High achiever!

Always a high achiever?

A comparison of student achievements on mathematical tests
with different aims and goals

Elisabet Mellroth

Faculty of Health, Science and Technology

Mathematics

LICENTIATE THESIS | Karlstad University Studies | 2014:67

High achiever!
Always a high achiever?
A comparison of student achievements on mathematical tests
with different aims and goals

Elisabet Mellroth

LICENTIATE THESIS | Karlstad University Studies | 2014:67

High achiever! Always a high achiever? - A comparison of student achievements on
mathematical tests with different aims and goals

Elisabet Mellroth

LICENTIATE THESIS

Karlstad University Studies | 2014:67

urn:nbn:se:kau:diva-34516

This thesis is also part of the series Studies in Science and Technology Education
ISSN 1652-5051 at Linköping University

ISSN 1403-8099

ISBN 978-91-7063-607-3
©
The author

Distribution:
Karlstad University
Faculty of Health, Science and Technology
Department of Mathematics and Computer Science
SE-651 88 Karlstad, Sweden
+46 54 700 10 00

Print: Universitetstryckeriet, Karlstad 2014

WWW.KAU.SE
To all my students whose strength I have not
discovered, acknowledged or encouraged
Abstract

This study explored changes in relative achievement over time. It also

investigated differences in how two groups of students activate mathematical
competencies. The aim of the study was to investigate students’ relative
achievement in mathematics over time, and how mathematical competencies
can be used to explore differences between groups of students on a non-
curriculum based test in mathematics. The study was divided in two parts.
Study 1 compared students’ (n=568) relative achievement in two national tests
in mathematics (years 3 and 6). Study 1 explored changes in relative
achievement between the two national tests as well as differences in relative
achievement between the national test in year 6 and the mathematical kangaroo
in year 7 (age 13). The study identified, from a sample (n=264) of study 1, two
groups of students with high achievements in only one of the tests, the national
test in year 6 or the mathematical kangaroo. Study 2 explored how differences
between those students relative achievement on the mathematical kangaroo
could be explained through activation of mathematical competencies. The
results in study 1 show that students undergo large changes, both increases and
decreases, in relative achievement between the national tests in years 3 and 6.
Study 2 shows how the two identified groups activate the mathematical
competencies differently on the mathematical kangaroo. 9% of the students
achieve highly in the mathematical kangaroo although they do not in the
national test. The study implicates the importance of using non-curriculum
bounded tests to identify strength in mathematical competencies among
students that not are able to show them through the national test.

Keywords: Achievement, alternative assessment, curriculum based assessment

mathematical competency, mathematical kangaroo, mathematics tests, national
tests.

iii
Acknowledgements

There are so many people that have been very important to me in different
ways during this work.

Arne Engström, my main supervisor, for your support and I would especially
like to thank you for your quick response and constructive comments in the
final phase of my work. Jari Appelgren, my second supervisor, I could not have
done this work without your explanations of the statistical methods, followed
by a millions of question that you have always answered with incredible
patience.

Per Nilsson for encouraging me, thanks to you I started to do research.

Inger Wistedt and Marianne Nolte, both for inspiring and encouraging me.

Jorryt van Bommel and Yvonne Liljeqvist for all your support, advice,
encouragement and jokes.

Annika Pettersson my own personal “hopms”, you helped me find laughter and
you shared the joy in the small as well as in the big things during this process. I
love to cry and laugh with you.

Torodd Lunde, Sergio Carlos, Matthias Müller and Ella Cosmovici Idsøe for
every interesting discussion, your critical comments and for sharing your
knowledge.

Thank you to everyone in SMEER at Karlstad university, especially to Shu-Nu

Chang-Rundgren for your support and your encouragement, also an extra thank
you to Daniel Olsson, Teresa Berglund and Jennifer Flint who all joined me on
this journey.

Ann Dyrman for your hawk eyes, thank you.

I want to give a huge thanks to all the teachers who helped me collect my data,
and also a special thanks to the teachers who helped me with the analysis. Of
course also many thanks to all the students who chose to participate in my
study.

iv
Thank you Karlstad municipality and Älvkullegymnasiet that allowed me to
participate in the research school. Especially I want to thank my bosses Inger
Thysell and Yvonne Toth, you both believed in me and made me feel good.

Thanks to The Swedish National Graduate School Science, Mathematics and

Technology Education Research (FontD) and to all friends and colleagues I
made through FontD in this process.

Jamie Pattle and Lindsay Fallis thanks for your friendship and your English
support.

I most likely forgot to mention someone, so a deep thank you to everyone that
I have spoken to about my work during this process.

At last I want to thank those people who I am afraid have suffered because of
my work, although always supported me. Emma and Carl, thank you for being
my wonderful children, for bringing me back to earth – I love you! And Jonas
you are incredible, always supporting, never complaining, taking care of
everything, and lifting me up when I am down – you are the very best for me!

v
vi
Contents

1 Introduction .......................................................................................................... 1
1.1 Main concepts used in the thesis ................................................................ 2
1.1.1 Relative achievement ................................................................................ 2
1.1.2 Mathematical competence and a mathematical competency .............. 3
1.2 Aim ................................................................................................................. 4
1.3 Research questions ....................................................................................... 5
1.4 Structure of the thesis .................................................................................. 5
2 Literature review ................................................................................................... 7
2.1 Curriculum, mathematical competencies and national test ..................... 7
2.2 Assessment .................................................................................................... 9
2.3 Achievement ................................................................................................ 11
2.4 High achievement, measured relatively.................................................... 13
2.5 The Swedish national test in mathematics .............................................. 17
2.6 The mathematical kangaroo ...................................................................... 18
2.7 Different aims in the two tests.................................................................. 19
3 Theoretical framework ...................................................................................... 21
3.1 Mathematical competencies as physical and mental activities .............. 21
3.2 Mathematical competencies ...................................................................... 22
3.2.1 Mathematical competencies – MCRF, KOM, NCTM ...................... 23
3.2.2 Mathematical competencies – Krutetskii and MCRF ........................ 24
3.3 Conceptualising the mathematical competencies ................................... 29
3.3.1 Applying procedures competency ........................................................ 29
3.3.2 Reasoning competency ........................................................................... 30
3.3.3 Communication competency ................................................................ 31
3.3.4 Representation competency .................................................................. 33
3.3.5 Connection competency ........................................................................ 34
3.3.6 Problem solving competency ................................................................ 35
3.4 Competency Related Activities, CRA ...................................................... 36
3.5 Situations for studying mathematical competencies .............................. 37
3.6 Master a competency .................................................................................. 38
3.7 Achievement ................................................................................................ 39

vii
 3.7.1 The national test in year 3 vs the national test in year 6 .................... 39
3.7.2 The national test in year 6 vs the mathematical kangaroo ................ 41
4 Method................................................................................................................. 42
4.1 Design of the study .................................................................................... 42
4.1.1 The national test in year 3 (2009) ......................................................... 44
4.1.2 The national test in year 6 (2012) ......................................................... 45
4.1.3 The mathematical kangaroo (2013) ...................................................... 46
4.2 Sample .......................................................................................................... 47
4.2.1 Representativeness of the sample ......................................................... 48
4.3 Procedure of the study ............................................................................... 58
4.4 Empirical data collection ........................................................................... 60
4.5 Data analysis ................................................................................................ 61
4.6 Guide to analysing mathematical competencies in tasks ....................... 62
4.7 Competency profile .................................................................................... 65
4.8 Validity and reliability ................................................................................. 69
4.8.1 Sample ...................................................................................................... 69
4.8.2 Non-participants ..................................................................................... 69
4.8.3 Error in identification ............................................................................. 69
4.8.4 Coding ...................................................................................................... 70
4.8.5 Required and reasonable solution ......................................................... 70
4.8.6 Task analysis ............................................................................................ 71
4.8.7 Problem solving competency ................................................................ 71
4.9 Ethical considerations ................................................................................ 72
5 Results .................................................................................................................. 74
5.1 Comparison of relative achievement on the national test in years 3 and
6, categories I and II .............................................................................................. 74
5.1.1 Interpretations of the results in categories I and II ............................ 75
5.2 Comparison of relative achievement on the national test in years 3 and
6, categories II and IV ........................................................................................... 77
5.2.1 Interpretations of the results in categories III and IV ....................... 83
5.3 Comparison of the national test in year 6 with the mathematical
kangaroo in year 7 .................................................................................................. 83
5.3.1 Interpretations of the comparison between the national test in year 6
and the mathematical kangaroo in year 7 ....................................................... 88
5.4 Summary of results – Study 1 ................................................................... 89

viii
 5.5 Competency profiles in the tests .............................................................. 89
5.5.1 The mathematical kangaroo .................................................................. 90
5.5.2 The national test in year 6 ...................................................................... 93
5.5.3 Comments on the competency profiles of the tests .......................... 93
5.6 Mathematical competencies activated by Group 1 and Group 2......... 93
5.6.1 Favoured competency ............................................................................ 94
5.7 Comparison of Group 1 and Group 2 on task level............................ 104
5.7.1 Interpretations ....................................................................................... 105
5.8 Summary of results – Study 2 ................................................................. 106
6 Conclusions and discussion ............................................................................ 108
6.1 Conclusions ............................................................................................... 108
6.2 Discussion.................................................................................................. 109
6.2.1 Ceiling effect and tests ......................................................................... 109
6.2.2 Movements in relative achievement ................................................... 110
6.2.3 Assessment and challenges .................................................................. 111
6.2.4 The problem solving competency ...................................................... 113
6.3 Method discussion .................................................................................... 114
6.3.1 Mathematical competencies................................................................. 115
6.3.2 Relative comparison ............................................................................. 116
6.3.3 Empirical data........................................................................................ 116
6.3.4 Participants ............................................................................................ 116
6.3.5 Implementation of the mathematical kangaroo ................................ 117
6.3.6 Representativeness of the sample ....................................................... 117
7 Implications ....................................................................................................... 118
7.1 Practice ....................................................................................................... 118
7.2 Research ..................................................................................................... 119
7.2.1 Further research .................................................................................... 120
8 Words ending the thesis or “What if?” ......................................................... 122
References ................................................................................................................. 123
Appendices................................................................................................................ 128

ix
Figures

Figure 1. Mathematical Competence and Competency. Picture inspired by

(Niss & Höjgaard, 2011, p. 51)............................................................... 4
Figure 2. Components of a mathematical competency and the parts activated
when working with a mathematical challenge. ................................... 22
Figure 3. The abilities defined by Krutetskii connected to the competencies in
MCRF. The underlined parts are those connected to physical acts.
The text of the abilities is taken from Krutetskii (1976, p. 87-88)... 26
Figure 4. The iterative process between conceptual and procedural
competency. ............................................................................................ 30
Figure 5. Relation between representations and connections. The picture is an
adaptation from Lithner et al. (2010). .................................................. 34
Figure 6. Test 1a and 2a show a movement of achievement that follows the
pattern of regression towards the mean. Test 1b and 2b show a
movement of achievement that breaks the pattern. .......................... 40
Figure 7. National test year 3; population no 1, sample no 2. .......................... 50
Figure 8. National test year 6; population no 1, sample no 2. .......................... 50
Figure 9. Left graph illustrates the ceiling effect and the right graph illustrates
the floor effect. ....................................................................................... 51
Figure 10. Frequency graph of results in the national test in year 3; the
population solid line, the sample dotted line. ..................................... 51
Figure 11. Frequency graph of results in the national test in year 6; the
population solid line, the sample dotted line. ..................................... 52
Figure 12. Results in the national test in year 6 versus results in the national test
in year 3.................................................................................................... 53
Figure 13. The process of the study, not related to time. .................................... 58
Figure 14. Reduced matrix showing the competency classification for task no. 1
.................................................................................................................. 65
Figure 15. Results on the task level in the mathematical kangaroo for student
S1. ............................................................................................................. 65
Figure 16. Student matrix of activated competencies for student S1. ............... 66
Figure 17. Competency distribution in the task in an example test. .................. 67
Figure 18. Example of a competency profile for one test and one student. ..... 68
Figure 19. Ranking position on the national test in year 6 versus the ranking
position on the national test in year 3. ................................................ 75
Figure 20. Frequency distribution in the national test in year 3, top 10%
shaded. Percentage results on the x-axis ............................................. 76

x
Figure 21. Frequency distribution in the national test in year 6, top 10%
shaded. Percentage results on the x-axis ............................................. 77
Figure 22. Illustration of possible movements of ranking position for students
in category III. ........................................................................................ 78
Figure 23. Group III, students with a large increase in ranking position from
year 3 to year 6. ....................................................................................... 79
Figure 24. Group IV, students with a large decrease in ranking position from
year 3 to year 6. ....................................................................................... 81
Figure 25. Distribution of students identified to be high achievers in one test
but not in the other. ............................................................................... 87
Figure 26. Competency profile for the mathematical kangaroo and the national
test in year 6. ........................................................................................... 92
Figure 27. Relative activation of the competencies App, Rea and Com in
comparison with the competency profile in the mathematical
kangaroo. The x-axis represents students. .......................................... 97
Figure 28. Relative activation of the competencies Rep, Con and Pro in
comparison with the competency profile in the mathematical
kangaroo. The x-axis represents students. .......................................... 98
Figure 29. Example of when it can be justified to change the limit of
comparison for the Fisher exact test. ................................................ 100

xi
Tables

Table 1 Showing how points (actual achievement) are connected to ranking (relative
achievement). ................................................................................................. 2
Table 2 Comparison of the frameworks on the basis of the competencies defined in the
MCRF. ..................................................................................................... 28
Table 3 The national test in year 3 (2009); mathematical content and maximum points
for each part. ............................................................................................... 45
Table 4 The national test in year 6 (2012); mathematical content and maximum points
for each part. ............................................................................................... 46
Table 5 The mathematical kangaroo (Benjamin) in year 7 (2013); mathematical content
and maximum points for each part, described by NCM. ............................... 47
Table 6 Comparison of statistical parameters between the population and the sample. . 49
Table 7 Distribution of points in the national test in year 6 for students who do not have
results in the national test in year 3. ............................................................. 54
Table 8 Wilcoxon test – the national test in year 3 and the national test in year 6, data
from students who have results in the national test in year 3 (2009) and in year
6 (2012). ................................................................................................... 57
Table 9 Wilcoxon test – the national test in year 6 and the Kangaroo test; data from
students who had results in the national test in year 6 (2012) and the
mathematical kangaroo in year 7 (2013). .................................................... 57
Table 10 Competency analysis guide used for each task. ............................................... 64
Table 11 Example of data used to produce a competency profile for one test and one
specific student. ............................................................................................ 67
Table 12 Statistical measures based on the actual total sum on the national test in year 3
and/or year 6 for the students in category III, Group R and Group S. ......... 80
Table 13 Statistical measures based on the actual total sum on the national test in year 3
and/or year 6 for the students in category IV, Group X and Group Y. ....... 82
Table 14 Students ranked top 5% (white), top 10% (light shaded) and top 20% (dark
shaded) in the mathematical kangaroo and the bottom 80% in the national test
in year 6, 13 boys and 12 girls. ................................................................... 85
Table 15 Students ranked top 5% (white), top 10% (light shaded) and top 20% (dark
shaded) in the national test in year 6 and bottom 80% in the mathematical
kangaroo, 7 boys and 16 girls. .................................................................... 86
Table 16 Statistical measures of ranking position for the two groups compared. ............ 88
Table 17 Results of the task analysis in the mathematical kangaroo. ........................... 91
Table 18 Result of competency analysis in the case of full points. .................................. 95
Table 19 Result of competency analysis of student S71. ............................................... 95

xii
Table 20 Summary of Fisher exact test. .................................................................... 101
Table 21 Summary of favoured competency on the mathematical kangaroo. ................ 103
Table 22 Comparison of response rate in each task. .................................................. 104
Table 23 Competencies activated by task 7,9, 15, 16, 17 and 21 ............................. 105

xiii
1 Introduction

In Sweden, one way to test the level of mathematical competence students

achieve through school mathematics is by national tests in mathematics.
Working as a teacher for more than 15 years, I have observed that some of my
students were very good in another mathematical test, called the mathematical
kangaroo, but achieved very low scores on the national test in mathematics. My
observation was confirmed in discussions with other teachers of mathematics
on different levels in Swedish schools. Mattsson (2013) writes about Swedish
teachers who have made the same observation; those teachers identified these
students as gifted. Students who achieve low in school mathematics get low
grades in mathematics and will most certainly not continue university studies in
mathematics, science or technology. Although my belief as a teacher is that, to
achieve highly on a mathematical test, no matter what the purpose of the test is,
cannot be done without possessing some mathematical competencies.

So how can this happen? Was it just sporadic observations made by me and
other teachers of mathematics?

In Sweden, it is a legal right for each school student to be supported to develop

their knowledge as far as possible (SFS 2010:800). There is a risk that there is a
group of students with high mathematical competencies that not are made
visible through the traditional assessment system. For example, teachers have
noted that there are students who achieve highly on non-curriculum bounded
tests. Despite the fact that they do not succeed on curriculum bounded tests
some teachers suspect that those students are gifted in mathematics (Mattsson,
2013). With suitable support, those students might be able to succeed in school.
Without suitable support, they might drop out of school for example for
reasons such as boredom (Stamm, 2008). There is a possibility that those
students are gifted in mathematics and not supporting them to develop their
competencies as far as possible is a waste to society. It is therefore important
for the individual and for the society to find ways to identify those students and
to give them suitable support. It is also important to help teachers with tools
that might provide a way to find some of those students.

It has been difficult to find research that compares students’ achievement on

curriculum bounded versus non curriculum bounded tests such as
competitions. I have not found any in languages available to the author

1
(Swedish, English and German). Research investigating how mathematical
competencies differ on different sorts of curriculum bounded tests have been
done in Sweden, for example by Boesen (2006), who compares tasks in teacher-
made tests with tasks in national tests according to mathematical competencies.

1.1 Main concepts used in the thesis

1.1.1 Relative achievement

In this study test results in three different mathematical tests are used as
empirical data. The three tests are of different characters and have different
aims. Here to use test results to compare achievement in the tests, despite the
differences in the tests, relative achievement instead of actual achievement is
used. An example is given to illustrate what is meant by relative achievement.

Example:
On a test it is possible to get a maximum of 50 points. 10 students participate in the test and
their points on the test are shown in Table 1. The student with the lowest points will be
ranked as 1, and the student with the highest point will be ranked as 10. The ranking number
indicate the student’s achievement in relation to the other students. This is how relative
achievement is used in this study.

Table 1
Showing how points (actual achievement) are connected to ranking (relative achievement).

Student Points Ranking/

Relative
achievement
S1 38 6
S2 48 10
S3 31 3
S4 13 1
S5 25 2
S6 44 8.5
S7 43 7
S8 36 5
S9 34 4
S10 44 8.5

2
1.1.2 Mathematical competence and a mathematical competency

The definition of the words competence and competency described in the

Danish KOM project (Niss & Höjgaard, 2011) will be used in this thesis:

A person possessing competence within a field is someone able to master the

essential aspects of that field effectively, incisively, and with an overview and
certainty of judgement (Niss & Höjgaard, 2011, p. 49).

In mathematics this means that a person possessing mathematical competence

has knowledge of, understands, can do and use, and has an opinion about
mathematics (Niss & Höjgaard, 2011). A mathematically competent person can
act in mathematical activities in different contexts where mathematics plays or
can play an important role. It implies factual and procedural knowledge as well
as concrete skills within the mathematical field (Niss & Höjgaard, 2011). On an
individual level, I interpret this to mean that there may be people who have
some well-developed mathematical competencies but who still do not possess
mathematical competence, for example, if he or she is unable to use the
competencies in different contexts, or he or she lacks in other competencies.
The difference between mathematical competence and mathematical
competency is described by Niss:

What then is a mathematical competency? It is an independent, relatively

distinct major constituent in mathematical competence as described above.
One could also say that a mathematical competency is a wellinformed readiness to act
appropriately in situations involving a certain type of mathematical challenge (Niss &
Höjgaard, 2011, p. 49).

A mathematically competent person possesses mathematical competencies that

are distinct but intertwined. In the Danish KOM project (Niss & Höjgaard,
2011) eight mathematical competencies are distinguished as the content of
mathematical competence, Figure 1.

3
 Mathematical Thinking
Representing Competency
Competency
To ask and answer in, with and

Mathematical Competence

To deal with mathematical

language and tools
about mathematics

Problem Tackling Symbol and Formalism

Competency Competency

Modelling Communicating
Competency Competency

Reasoning
Aids and Tool Competency
Competency

Figure 1. Mathematical Competence and Competency. Picture inspired by (Niss & Höjgaard, 2011, p.
51).

1.2 Aim

The overall aim was to investigate if there are students who possess good
mathematical competencies although they fail in school mathematics. In more
detail the aim is two folded.

One aim was to describe groups of students with similar movements in relative
achievement by means of results on three different mathematical tests over a
four-year period.

A further aim was to investigate a method used to explain differences in

achievement on curriculum bounded and non-curriculum bounded tests.

4
1.3 Research questions

1. How does relative achievement in the national test change between year
3 and year 6?

2. How do students who are ranked highly through the mathematical

kangaroo achieve in the national test (year 6)?

3. How do students who are ranked highly through the national test (year
6) achieve in the mathematical kangaroo?

Results from research questions 2 and 3 will be further investigated through

research question 4.

4. How can differences in achievement on the mathematical kangaroo be

explained by mathematical competencies?

1.4 Structure of the thesis

The thesis consists of eight chapters. The second chapter is a literature review
that gives a background for the aims and research questions. The chapter has to
do with assessment and achievement, with an extra discussion of high
achievers. The tests included in the empirical data are also described.

In the third chapter, theories used in the process of working with the empirical
material are discussed. The theory of mathematical competencies is given
especially large room. Each mathematical competency that is used in the
analysis is discussed in depth, both through the theoretical framework chosen
for the study and also connected to other research involving mathematical
competencies.

The fourth chapter describes the methods used in the thesis. The tests involved
are described in more depth. The sample is compared with the population and
the representativeness of the sample is described through statistical
measurements and tests. Validity, reliability and ethical considerations are
discussed in this chapter.

5
Analysis and results are described in chapter five. Chapter five starts with
describing analysis and results concerning the descriptive part about
movements in relative achievement. The last part of the chapter presents
analysis and results of mathematical competencies in the tests and within
groups of identified students.

Interpretations from both parts of the study are presented in chapter six, and a
discussion of the interpretations connected to the aims of the thesis is included.
Chapter seven gives a discussion of the contribution of knowledge together
with suggestions for further research. The last chapter, chapter eight, is words
ending the thesis.

6
2 Literature review

In this study curriculum bounded and non-curriculum bounded tests and

students’ results in the tests are used as empirical data. This chapter serves to
give an overview of the literature and earlier research that have inspired and
shaped this study. Several aspects must be discussed, such as curriculum,
assessment and achievement, high achievement and the tests involved in the
study.

2.1 Curriculum, mathematical competencies and national test

This section describes different levels of a curriculum and discusses how a

standardised international test such as PISA, which uses concepts similar to
mathematical competencies, influences national curricula like the one in
Sweden. In the section the connection between mathematical competencies, the
Swedish curriculum and the Swedish national tests are discussed and how the
construction of different tests gives different opportunities for students is also
mentioned.

A curriculum can be divided into three levels (Mesa, Gómez, & Cheah Hock,
2013), the intended, the implemented and the attained. As interpreted in the
Swedish system:

 The intended is the curriculum that is the national curriculum,

 The implemented is the same as mentioned by Mesa et al. (2013) which
is what happens in the classroom. What happens in the classroom is
partly dependent on how the teachers interpret the curriculum,
 The attained is what the students have learned (Mesa et al., 2013).

How the attained curriculum is manifested is mainly measured through

students’ achievement on class assessment and external mandated tests, such as
national tests or international standardised tests, for example PISA (Mesa et al.,
2013, p. 866). The international test PISA has influence on national levels; the
mathematical framework for PISA is based on capabilities (earlier named
competencies) and not on content knowledge (OECD, 2013). For example the
development of the Swedish curriculum (Skolverket, 2011a) is influenced by

7
international measurements such as PISA (Skolverket, 2011b). In European
countries, curricula in mathematics are today often related to mathematical
competencies (Mesa et al., 2013), which in turn means that assessment must
also relate to mathematical competencies.

From the year 2000 the Swedish curriculum has changed from earlier having a
stronger focus on the content of mathematics to having a focus on what kinds
of mathematical competencies are needed to work with the subject (Boesen,
2006). This shift also influenced the assessment system, and especially the
national tests, since they are meant to guide Swedish teachers in assessments
and grading (Skolverket, 2014). Because of the shift in the curriculum, the
national tests now also aim to assess conceptual understanding instead of only
factual knowledge.

The curriculum for compulsory school in Sweden today is goal oriented;

students are graded in mathematics according to how well their mathematical
abilities develop (Skolverket, 2011a). In education, the teacher is responsible for
giving students opportunities to develop their abilities (Skolverket, 2011a) and
also for judging how well those abilities are developed, finally giving a subject
grade for each semester, starting in year 6 (age 12) (SFS 2010:800). It is
common to use tests as a part of the judgement of students’ grades and teachers
both construct own tests and use national tests. These tests are not necessarily
equal in terms of content or as to which competencies they require the student
to succeed.

Boesen (2006) started filling the gap of research concerning the relation
between national tests and teacher-made tests in the Swedish context. In one
part of his research, he compared what kind of reasoning the students need to
be able to solve the tasks in teacher-made tests and in national tests. He
compares imitative reasoning versus mathematical creative reasoning. In short,
imitative reasoning is a kind of reasoning the student has met before and has
been trained in; the student does not need to invent anything new. Central in
creative reasoning is “…the reasoning that goes beyond just following strict
algorithmic paths or recalling ideas provided by others.” (Boesen, 2006, p. 18).
The national tests give tasks that cannot be solved by imitative reasoning and
therefore give tasks that differ from textbook tasks (Boesen, 2006). These
national tests, since they have a guiding position, should give students tasks that
demand more than imitative reasoning. This was confirmed in Boesens (2006)
study. The national tests give the possibility to use mathematically creative

8
reasoning to a much higher degree than teacher-made tests, and, in the teacher-
made tests, many of the tasks could be solved using only imitative reasoning
(Boesen, 2006).

2.2 Assessment

This section indicates that assessment according to the Swedish curriculum is

supposed to relate to mathematical abilities, similar to mathematical
competencies. It also discusses how assessment is implemented and how it is
used.

Assessment is supposed to be used for learning (A. Pettersson, 2004), meaning

that assessment should be used in a formative way. There is an increased
interest in external assessment (Mesa et al., 2013); external means that those
tests are constructed outside the schools and can for example be national tests
and/or international achievement tests like PISA or TIMSS. The externally
constructed tests give each student a total mark that gives the students’
achievement on that specific test. The Swedish national tests are given at the
end of the school year; students either get a grade in that class (years 6 and 9) or
a teacher opinion (year 3) of the students’ knowledge (Skolverket, 2014). Since
there is little time left of the school year it is difficult to work in a formative way
with the students after the national tests. Achievement in those tests is
therefore important, and it is partly used to decide students’ subject grade.

When teachers are asked how they assess their students, they relate this to tests,
portfolios etc., but if they are asked how they know that their students have
learned something, they relate for example to classroom questions and group
activities (Dorr-Bremme & Herman, 1986). One possible interpretation is that
teachers do not completely think that assessments measure what students have
learned. However, when using assessments, both formal and informal, the
purpose of assessment is to determine the existing status of a student’s
knowledge (Wiliam, 2007). The results of the assessment can be used both
summatively and formatively depending on the purpose of the assessment.

One aim of the Swedish national tests is to support the teacher in the
assessment process (A. Pettersson, 2007). Sometimes both teachers and

9
students focus on succeeding in the test instead of focusing on the learning (A.
Pettersson, 2007), that is, to achieve highly on the tests. To get both teachers
and students to focus on students’ learning was one of the reasons for
developing a new curriculum (Skolverket, 2011a) and new national tests. The
national tests are supposed to be an assessment instrument for learning instead
of of learning (A. Pettersson, 2007). Pettersson (2007) asks what it means to
have knowledge in a subject; she divides the subject of knowledge into two
parts – one personal and one official. Personal knowledge is how the individual
looks at knowledge, for example in mathematics, while official knowledge is
dictated by the curriculum. The national tests serve as the assessment of official
knowledge, according to the intended curiculum. In the national test today, it is
important to be able to apply knowledge to tasks that demand conceptual
understanding, argumentation, communication and logical competency (A.
Pettersson, 2007).

Assessment can be perceived positively or negatively to the student. Through

the use of conventional tests in mathematics, some students achieve at the top
of the class, getting good grades and teacher praise, while others achieve
bottom results. Most students are aware of their place in this created hierarchy
(Boaler, 2006). Assessment can be used to lift each student’s positive sides and
to help the student to develop those parts that can improve. For a teacher to
observe all students’ strength and weaknesses, it is important to observe and
document students’ knowledge in many ways (Jönsson & Svingby, 2008).

Put simply, when there are many ways to be successful, many more students
are successful. Students are aware of the different practices that are valued
and they feel successful because they are able to excel at some of them
(Boaler, 2006, p. 42).

Assessing each student individually is one of a teacher’s difficult missions. In a

mathematics classroom there are students without motivation, those with weak
knowledge, those who are highly motivated and those with advanced
knowledge (Boaler, 2006). Boaler (2006) showed that through a collaborative
problem-solving approach, students achieved better in mathematics and also
chose more advanced mathematics courses than students in comparison
schools. When students asked for help, the teacher tried not to give the answers
but tried to lead the groups into finding the solutions together. They also used
more open-ended problems than usual (Boaler, 2006).

Using different practices, grouping, using open-ended problems can influence

quality learning, and quality learning leads to higher achievement, also when

10
externally mandated tests are used (Wiliam, 2007). Internationally, the use of
summative tests constructed externally (i.e. by someone or some organisation
outside the school) has increased. As a criticism, some say that that type of tests
discriminates against certain groups of students owing to psychological issues.
For example there are some qualities that can not be measured through tests,
such as the metacognitive process (Gipps, 1999).

It is clear in the curriculum that it is the students’ mathematical abilities that are
to be assessed in mathematics (Skolverket, 2011a). For the curriculum between
1994 and 2011, Jönsson (2008) writes that many of the goals in the compulsory
school are complex and difficult to assess and that there is a lack of models for
how to assess those goals in the classroom. The present curriculum (Skolverket,
2011a) does not describe what is meant by all mathematical abilities. It is
therefore plausible that teachers still think that the curriculum goals are
complex and difficult to understand.

In school, students’ performance is assessed, either comparing students with

one another or according to goals in the curriculum. In the current Swedish
curriculum for compulsory school (Skolverket, 2011a), students are assessed
according to goals in the curriculum. What the purpose of assessment is does
not matter; as long as students are being assessed, it is possible to study their
relative achievement, rank their achievement and identify a top percentage
population, for example the top 10%. The achievement can change as learning
progresses; it may decrease as well as increase (Gagné, 2005). However,
according to Gagné (2005), most talented (top 10%) students maintain their
top-position through their formal schooling.

2.3 Achievement

Two studies that follow students’ achievement in different ways in the

Scandinavian context are in this section discussed. Both studies investigates
movements in achievement over time and are therefore of interest for this
study.

In a large longitudinal (7517 students in 29 municipalities, over three years)

study (A. Pettersson, 1990), students’ achievement behaviour and their

11
achievement development were investigated. Achievement behaviour (A.
Pettersson, 2007) meant ways of solving tasks. She investigated whether there
were differences in achievement behaviour between students who developed
differently in achievement. Achievement was measured in a test with 15 tasks in
year 3, with the addition of four new tasks in year 6. The tasks were
dichotomously scored (either credit or no credit) and the students were
grouped into five groups according to their achievement development (A.
Pettersson, 1990).

A. Students who achieved highly (13 points or more), both in year 3 and in
year 6,
B. Students who achieved low (less than 4 points), both in year 3 and in
year 6,
C. Students who achieved better in year 6 compared to in year 3 (at least 9
points more),
D. Students who achieved lower in year 6 compared to in year 3 (a decrease
of at least 2 points),
E. Students who achieved at an intermediate level both in year 3 and in year
6 (8 points in year 3, 11 points in year 6).

131 students belonged to group A, approximately 2% of the sample, and are

called “good at computation”. Group B called “weak at computation”,
consisted of 222 students, which is approximately 3%. Group C is an
improvement group, containing 226 students, approximately 3%. Group D is a
decreasing group and contains 180 students, approximately 2.5%. Group E is a
group containing students that have average results both in years 3 and 6; this
group consists of 118 students, which is approximately 1.5%.

Essential aspects for students’ achievement are teaching, learning and the
students’ individual prerequisites. The students in the study were followed up in
year 9. Those who achieved highly in both years 3 and 6 also achieved highly in
year 9, and those who achieved low in both year 3 and 6 also achieved low in
year 9 (A. Pettersson & Boistrup, 2010). Those students who had poor results
in both school years (3 and 6) had difficulty understanding explanations given
by the teachers and wanted more help than was given. In the study of
Pettersson & Boistrup (2010), the students who achieved highly in all grades are
not discussed.

By following mathematical achievement in the number sense, calculation skills

and text tasks among children from the age of 6 to the age of 15 Häggblom

12
(2000) show that there are movements in relative achievement. Low achievers
at age 6 can become high achievers at age 15 and high achievers at age 6 can
become low achievers at age 15. When looking at number sense, the results
show that movements among the high achievers are more common than
among low achievers. However, as a summary, less than 20% of the children
belong to the same achievement group throughout their time in school.
Häggblom (2000) therefore concludes that mathematical achievement at the age
of 6 says very little about how a child will achieve at the end of compulsory
school.

2.4 High achievement, measured relatively

Most humans have different competence in different subjects. Some have

competence to become a well-paid soccer player, some have competence to
become an opera singer and some have competence to become a well-known
(maybe not well-paid) mathematician. It is natural to believe that this is also the
case for students in school; they are all different from each other and have
different competencies in different subjects.

The aim of this section is to describe how some other studies have used relative
achievement in the perspective of studying high achieving students. Those
studies have guided percent limits that are used in the analysis in the present
study. Using relative achievement means that there will always be students at
the top and at the bottom. This study especially uses the perspective of those
who achieve highly one way or another.

What does it mean to achieve highly or low on a test in school mathematics, or

any other subject? It is possible to define a high achiever in a test as someone
who scores above a certain number of points, and define that specific number
of points for each test. An alternative is to say that someone who scores among
the top percent of the participating students is a high-achiever, and to define
that percent limit. In this study, a high achiever is someone who scores among
the top percent of the participating population. The opposite is a low achiever,
that is, someone who achieves among the bottom percent of the participating
population, according to this study.

13
High achieving students are seen by some as gifted, although it is important to
distinguish between giftedness and high achievement. Gifted students are not
necessarily high achievers; and, vice versa, there are many high achievers who
are not necessarily gifted (Bar-On & Maree, 2009). This study is not about
giftedness; it uses relative achievement in three different mathematical tests as
empirical data to describe and identify groups of students. Of special interest in
this study are those students who achieve highly in one test but not on the test
that is used in the comparison. It is therefore interesting to explore how high
achievers and high achievement are discussed in earlier research and literature.

In mathematics, how do schools judge who is competent in mathematics and

easily reach the goals of the curriculum? A note should be included here that, in
teacher education in Sweden, there is very little, if any, information about how
to identify and support students with a capacity to develop further than is stated
in the curriculum (Mattsson, 2013).

Despite that high achievement is not equivalent to giftedness, the concept of

high achiever is sometimes connected to talent, for example in Gagnés
Differentiated Model for Giftedness and Talent, DMGT (Gagné, 2005). Very
briefly the DMGT is a model that combines giftedness with talent. A gifted or
talented individual, according to this model, possesses and uses outstanding
natural abilities, aptitudes (Gagné, 2005). A talented person masters
systematically developed knowledge and skills to an outstanding degree in a
field of human activity (Gagné, 2005), for example mathematics or soccer. This
outstanding degree is defined in the DMGT as the top 10% among age peers
that are or have been active in that field (Gagné, 2005). DMGT mentions that
some of the top 10% students will develop a talent with help of different
internal and external support (Gagné, 2004). It is how those top 10% are
measured that connects talent with high achievement.

In school, achievement is measured in assessments of different tests and/or

teacher observation; the summary of the assessments finally gives a subject
grade. Vialle (2007) made a longitudinal study to investigate relationships
among personal factors, social support, emotional well-being and academic
achievement. To identify the relevant students, she used the model of Gagné
and selected those who scored among the top 10% in two standardised tests
used in Australia: ELLA, which tests language and literacy, and SNAP, which
measures numeracy skills in problem solving, number, measurements, data and
space (Vialle et al., 2007).

14
In Ireland, the Irish Centre for Talented Youth (CTYI) identifies students for
participating in their program that aims to challenge and encourage talented
youth. To participate, the student must show, either through testing or in some
other way that he or she is among the top 5% of the school population (Mönks
& Pflüger, 2005).

In terms of giftedness in mathematics, up to 20% of students are in need of

special education because of that specific giftedness (Advisory Committee on
Mathematics Education, 2012; E. Pettersson, 2011). “Well above average
ability” is a term used by Renzulli (2005) to describe those who perform or
possess the potential for performance that belongs to the top 15 to 20 % in any
given area.

As soon as percentages are used to form a group in a population, individuals

are compared relative to each other, in one way or another. In the field of
giftedness relative comparison is often made by testing, mostly by combining
tests for example cognitive- and domain specific tests (Nolte, 2012b; Pitta-
Pantazi, Christou, Kontoyianni, & Kattou, 2011; Vanderbilt University, 2014;
Vialle et al., 2007). Individuals with the best results are chosen for further
development or investigation.

In research, relative achievement is sometimes used to choose participants. One

example is a large longitudinal study in the USA, the “Study of Mathematically
Precocious Youth (SMPY)” (Vanderbilt University, 2014). This study uses high
achievement as a criterion in choosing participants. SMPY is an ongoing study;
it is planned as a 50-year study and started in the early 1970s (Gross, 2009). The
participants enter the project in their adolescence on the basis of their scores on
the math or verbal scale of the Scholastic Aptitude Test (SAT) that place them
among the top 1% of the population (Gross, 2009). This means that the
participants were chosen through their relative achievement on tests. In this
project, it has been found that the participants seem to engage and manipulate
math or language in ways that are more characteristic of students that are many
years older. They also take great individual responsibility for their academic
success or difficulties. They blame themselves and not external factors for
difficulties. They accept that they have a math or verbal talent and this together
with motivation and endeavour, contributes to their academic success (Gross,
2009). In an academic perspective, the project is successful; in their early 30s
90% had Bachelor’s degrees and 25% held doctoral degrees (Benbow, Lubinski,

15
Shea, & Eftekhari-Sanjani, 2000). The highest achievers among the participants
tend to maintain their pre-eminence in adult life.

However, Talent Searches have also identified that achievement and success
are by no means built in for gifted students. Where schools have not
provided structured opportunities for talent development, these students
perform, in school, and in later life, at levels significantly below their true
capacity. Even remarkably high ability is not by itself sufficient; exceptional
ability does not develop into exceptional achievement unless the educational
system accepts its responsibility to actively facilitate this process (Gross, 2009
p. 347).

When high achievers are mentioned in research in the Swedish context, it is

common to measure achievement in traditional tests in school and/or grades
for example (Hallesson, 2011; Szabo, 2013). In Sweden, as of 2009, there has
been an opportunity for students with a special interest in mathematics to enter
so called “spetsprogram”, a gifted program in English, in upper secondary
school. To get a place in one of those programs, the student has to do a special
test for that specific program; together with the test result and with their
average grade from compulsory school, they can be placed in one of those
programs (Mattsson, 2013). In addition, in the next educational step, i.e.
university, being accepted to a university is to the greatest extent based on
students’ average grades from upper secondary school. In order to be accepted
in the most popular university programs through grades, a student must have
the highest grade in each subject from upper secondary school, and, to get
highest grade in any subject the student must achieve highly on almost every
assessment during the time in upper secondary school1.

Mattsson (2013) found that students with well-educated parents are over-
represented in the mathematical tracks of the gifted programs and that females
are under-represented. This is supported in an international perspective by
Sivelman & Miller (2009), who stress that, when gifted programs are reserved
for high achievers, they serve a primarily higher socioeconomic group.
However, being gifted does not have to do with social class (Silverman &
Miller, 2009). Mathematics is an academic subject, and academic achievement is
correlated to social economic status (SES), ethnicity and language status; the
relation is neither perfect nor deterministic (McCoach et al., 2010). Factors such
as school, teacher and parent have also been associated with student
achievement (McCoach et al., 2010).

1
The author has been working as an upper secondary teacher for 15 years, on four different schools
in two cities.

16
As described in this section, relative achievement connected to high
achievement is used in research. In the Swedish school context, to gain a place
in the programs in upper secondary school, mentioned gifted programs by
Mattsson (2013), students must be high achievers in compulsory school. For
example, it is very likely that they achieved highly on the national tests.

2.5 The Swedish national test in mathematics

Results in the national test in year 3 (2009) and year 6 (2012) are used as
empirical data. One main goal of the national tests is to support equality and
fairness in assessments and grading (Skolverket, 2014). The tests provide a basis
for analysis of the extent to which the demands on knowledge are fulfilled on
different levels - the school-, organisation-, and national levels. The national
tests are not examination tests; they are meant to be one of teachers’ collective
information about each student’s knowledge. The tests are summative; they
shall function as a checkpoint at the end of a school year or in a subject course.
The test should show what qualities the student has in his or her knowledge of
the subject (Skolverket, 2014).

For compulsory school, the “PRIM-gruppen2” (Stockholm University, 2013)

has the responsibility for developing the national test. Each test consists of
several parts and aims to give opportunities to show as many aspects of
knowledge as possible (A. Pettersson & Boistrup, 2010). The tests offer a
variety of contexts and a range of ways to respond. The tasks shall be
constructed to give students the opportunity to demonstrate different areas of
knowledge and different levels of quality in their knowledge (A. Pettersson &
Boistrup, 2010).

In Sweden students take national tests in mathematics in years 3, 6 and 9 in

compulsory school. Students’ knowledge in mathematics is assessed according
to mathematical abilities and the level of these that each student shows
(Skolverket, 2011a). The national tests in mathematics in Sweden have an
influence on the student’s grade in a subject. Korp (2006, p. 79) writes that, in

2PRIM-gruppen is a research group at Stockholm University; their main focus is on assessment of

knowledge and competence. They develop different instruments for assessment and evaluation, for
example national tests for compulsory school.

17
upper secondary school, it is most common for a student to either get a higher
grade in mathematics or the same grade as was achieved in the national test.
There are exceptions, however, when students are given a lower grade than the
national test shows. It is reasonable to assume that the pattern between subject
grade and results on the national test is the same in compulsory school.

2.6 The mathematical kangaroo

The non-curriculum bounded test used to collect empirical data is the

mathematical kangaroo. The mathematical kangaroo is an international
competition in which more than 50 countries and more than 6 million children
participate (Wettbewerbsbedingungen, 2013). Each country constructs its own
test from a selection of problems that an international group of researchers and
well acknowledged teachers together construct3. In Sweden (2013), the tests
were given on five school year levels, year 0―2, year 3―4, year 5―7, year 8―10
and year 10―12. The mathematical kangaroo is not connected to any
curriculum; the aim of the mathematical kangaroo is to stimulate interest in
mathematics and to arouse curiosity and a desire to learn mathematics
(Nationellt centrum för matematikutbildning, 2013). Another aim is to offer
interesting challenges (Wettbewerbsbedingungen, 2013).

Using kangaroo problems in mathematics education can be a part of a

successful learning path; kangaroo problems are mentioned in discussions of
problem solving (E. Pettersson, 2011), and problem solving is also mentioned
in the context of challenging students (Krutetskii, 1976; Nolte, 2012a). The
mathematical kangaroo has inspired a test used to measure mathematical
competencies in the process of identifying mathematically gifted students (Pitta-
Pantazi et al., 2011). The mathematical kangaroo, in comparison with other
mathematical competitions given in Sweden for students at age 13, does not
need training in mathematics, which is important for the choice of test. It is not
natural to think that a student that achieves low in school mathematics would
voluntarily participate in a training program aimed for competition in
mathematics. Students who compete in competitions in advanced mathematics

3
The researcher has asked people involved in the process of constructing the Mathematical
Kangaroo from Romania and Sweden.

18
most likely achieve highly in school mathematics4; those students are not the
focus of this study.

In the Swedish context it is therefore interesting to explore whether students

who achieve highly on the mathematical kangaroo are given the opportunity to
activate mathematical competencies that differ from the mathematical
competencies they are given the opportunity to activate in the national test in
mathematics.

2.7 Different aims in the two tests

The tests included in this study have different aims and are not constructed to
measure the same things. However, they are mathematics tests, and in this
study they will be analysed according to mathematical competencies and the
distribution of those in the tests.

A student who achieves highly on a national test can be seen to have both a
broad and deep understanding of the mathematical abilities in the Swedish
curriculum, since the national test is constructed to broadly test goals and
criteria given by the curriculum (Stockholm University, 2013). The
mathematical content in the tests, for example what kind of geometry, what
kind of equations, is chosen according to what the curriculum tells about the
mathematical content in that special age group. It is therefore possible to say
that the national tests measure mathematical abilities through the mathematical
content determined by the current curriculum.

It is very important to offer students mathematical challenges, partly because of

the Swedish Education Act (SFS 2010:800), which states that:

 Education in school shall promote all children’s and students’

development and learning,
 Education shall take into account that children and students have
different needs,

4
No research about which students who participate in mathematics competition has been found in
Sweden. By asking teachers who are known in Sweden to participate in those competitions, the
claim that those students are also high achievers in school mathematics is supported.

19
  Children and students shall be supported and stimulated in a way such
that they develop as far as possible,
 There should be an ambition to compensate for differences in children’s
and students’ abilities for being benefited by the education.

Internationally, supporting students who need more challenge than average is

discussed as an important issue. This is shown in the following:

Moreover, we should not forget the legal or ethical perspective. As clearly

mentioned in article 26 of the UNESCO definition of the rights of children,
all children – including intellectually gifted ones – have a right to an education
that will foster the development of their abilities and personality to their
fullest (Stoeger, 2004, p. 169).

Mathematically challenging tasks also give support and stimulation to students

who are gifted in mathematics (Nolte, 2012a).

This literature review ends by saying that the Swedish curriculum in

mathematics is connected to mathematical competencies, and that the national
tests have a guiding position for the curriculum and should assess students’
mathematical abilities. Further, different mathematical tests are used to assess
mathematical competence, and how these tests are constructed gives students
different opportunities to show different mathematical abilities, for example
imitative or creative reasoning. A teacher-made test and a national test are both
connected to the curriculum. This study compares a curriculum bounded test
and a non-curriculum bounded test using mathematical competencies in its
analysis.

20
3 Theoretical framework

The purpose of this chapter is to describe how mathematical competencies are

to be interpreted in this study. A further purpose is to describe how percent
limits are chosen in grouping students according to relative achievement.

Words such as competency, capability, proficiency, processes and ability are

used in different texts for similar situations. It is probably possible to write a
separate thesis in linguistics about these words, how they are connected and
how they have been used through history. The words are not congruent
although they are indeed related. Translations between different languages from
one original language to a second can also complicate the meaning of these
words. They can be used as synonyms in some situations but not in others. I
choose not to go deeper into a discussion of their meaning.

Mathematical competence is sometimes divided into parts called mathematical

competencies. To be mathematical competent, one must master not only one
competency but all (Krutetskii, 1976; Niss & Höjgaard, 2011), although a lack
of one competency may be offset by another (Krutetskii, 1976). In the second
part of this study, the parts, the mathematical competencies, are considered,
that is, not mathematical competence.

3.1 Mathematical competencies as physical and mental activities

Competence descriptions in mathematics education are used to describe the

“aim” for teaching in mathematics (Niss & Höjgaard, 2011).

…mathematical competence comprises having knowledge of, understanding,

doing, using and having an opinion about mathematics and mathematical
activity in a variety of contexts where mathematics plays or can play a role
(Niss & Höjgaard, 2011, p. 49).

Mathematical competence includes procedural and factual knowledge, but

mathematical competencies have more to do with mental and physical activities
in how to treat mathematical challenges. The competencies are behavioural in
nature; the focus is on the ability (as in being able to) to carry out relevant
activities (Niss & Höjgaard, 2011).

21
All competencies are dual in nature; they have an “investigative” and a
“productive” side. Both sides are behavioural because they are about people’s
competence in being able to carry out activities, mental or physical (Niss &
Höjgaard, 2011). Using a mathematical competency in a mathematical activity
therefore demands both mental and physical activity, Figure 2.

Mathematical competency
Mental

Procedural Mathematical challenge

Factual
Physical

Understand
Carry out
Reflect Investigative Productive
Use
Assess

Figure 2. Components of a mathematical competency and the parts activated when working with a
mathematical challenge.

3.2 Mathematical competencies

Mathematical competencies: A research framework, MCRF (Lithner et al.,

2010) is used here as a framework in analysing tasks in tests. The framework
has been developed by a group of researchers at Umeå University, Sweden, and
constructed to analyse empirical data with a focus on students’ opportunities to
activate mathematical competencies (Lithner et al., 2010). It has been developed
for a project called “National tests in mathematics as a catalyst for
implementing educational reforms” (Lithner, 2011). Swedish national tests in
mathematics are analysed with this framework in one part of the project. The
framework is not used to analyse what students actually learn but to analyse
opportunities to learn (Lithner et al., 2010). As the study involves an analysis of

22
what is needed to come to a solution of a mathematical task, it is possible to use
this framework.

It is not only in the framework chosen here that mathematical competence is

divided into smaller parts that together build competence. The MCRF was
developed in Sweden and has been used for analyses of mathematical
competencies in the Swedish context (Boesen et al., 2014; Säfström, 2013) . To
show that the framework is not restricted to the Swedish context this chapter
will discuss two other frameworks that use parts to describe mathematical skills,
knowledge or competence as a wholeness. The comparative frameworks are
chosen because they have inspired or are related to the MCRF. Those are
“Principal and standards for school mathematics” (NCTM, 2000) and the
Danish KOM-project (Niss & Höjgaard, 2011). The frameworks chosen for
comparison are well known and are used in research in Western culture.

Another framework (Krutetskii, 1976) is chosen for comparison because it

deals with mathematical giftedness and is a well-known framework in this field
of research. The frameworks will be used in the discussion of the study, and it
is therefore important to discuss similarities and differences in the framework
used in the analysis. This study deals with high achieving students in different
ways. Teachers’ suspect that students in one of the groups studied might be
gifted (Mattsson, 2013). The framework of Krutetskii will not be used in the
analysis, partly because giftedness is a complex phenomenon that, above
domain specific competencies, also involves cognitive and affective factors, for
example (Nolte, 2012a; Pitta-Pantazi et al., 2011), factors that are not involved
in this study. It is important to note that there is not an equivalence between
high achievement and giftedness (Bar-On & Maree, 2009).

3.2.1 Mathematical competencies – MCRF, KOM, NCTM

NCTM (2000) is intended to be a resource for all involved in mathematics

education in the US. The standard defines 10 curriculum standards, 5 related to
content goals and 5 related to processes. The 5 processes are: problem solving,
reasoning and proof, communication, connections and representation. Content
goals and processes are thoroughly described for education from pre-

23
kindergarten to grade 12. Those processes together with content knowledge,
build mathematical competence (NCTM, 2000).

The aim of the KOM project (Niss & Höjgaard, 2011) is to contribute to a
coherence and progression of mathematics education in the Danish school
system, lengthwise and crosswise. Eight overlapping mathematical
competencies divided into two groups together build up mathematical
competence in the KOM project, see Figure 1. The competencies are, in one
group “To ask and answer in, with, about mathematics”, reasoning-, modelling-
, problem tackling and mathematical thinking competency. The other group
consists of, “To deal with mathematical language and tools”, representing-,
symbol and formalism-, communicating- and aids and tools competency.

The MCRF is mainly inspired by the NCTM and the KOM-project, although it
was developed for research and not for education. The main difference is that
the competencies used are made more distinct and are differentiated from each
other. The framework defines 6 mathematical competencies:

 Applying procedures,
 Reasoning,
 Communication,
 Representation,
 Connection,
 Problem solving.

The competencies will each be discussed in section 3.3.

3.2.2 Mathematical competencies – Krutetskii and MCRF

Krutetskii (1976) uses the word ability which is used here when linking to his
work. Mathematical ability and the possibility to make progress in mathematical
activities are seen as a complex set of mathematical abilities (Krutetskii, 1976).
The combination of mathematical abilities in a mathematical activity is a
condition for high achievement. However, weakness in one ability can be
compensated for by another ability so that successful or high achievement is
still possible (Krutetskii, 1976).

24
Krutetskii (1976, p. 87-88) lists nine component mathematical abilities.

1. An ability to formalize mathematical material, to isolate form from content,

to abstract oneself from concrete numerical relationships and spatial forms,
and to operate with formal structure – with structures of relationships and
connections.

2. An ability to generalize mathematical material, to detect what is of chief

importance, abstracting oneself from the irrelevant, and to see what is
common in what is externally different.

3. An ability to operate with numerals and other symbols.

4. An ability for “sequential, properly segmented logical reasoning”

(Kolmogrov, 180, p. 10), which is related to the need for proof,
substantiation, and deductions.

5. An ability to shorten the reasoning process, to think in curtailed structures.

6. An ability to reverse a mental process (to transfer from a direct to a reverse

train of thought).

7. Flexibility of thought – an ability to switch from one mental operation to

another; freedom from the binding influence of the commonplace and the
hackneyed. This characteristic of thinking is important for the creative work
of a mathematician.

8. A mathematical memory. It can be assumed that its characteristics also

arise from the specific features of the mathematical sciences, that this is a
memory for generalizations, formalized structures, and logical schemes.

9. An ability for spatial concepts, which is directly related to the presence of a

branch of mathematics such as geometry (especially the geometry of space).

Krutetskii (1976) addresses the abilities as mental activities. However, I believe

that when someone operates with something it is also a physical act, meaning
that, to be able to observe those abilities, the student needs to do a physical act
– verbally through thinking aloud and sometimes also in combination with a
writing process. Figure 3 shows a comparison of the abilities defined by
Krutetskii (1976) with the competencies defined in MCRF (Lithner et al., 2010).

25
 88).
An ability to formalize mathematical material, to isolate form from content,
to abstract oneself from concrete numerical relationships and spatial forms,
and to operate with formal structure – with structures of relationships and
connections.

(all are connected

to this competence)
Problem solving
An ability to generalize mathematical material, to detect what is of chief
importance, abstracting oneself from the irrelevant, and to see what is
common in what is externally different.

Reasoning
An ability to operate with numerals and other symbols.

An ability for “sequential, properly segmented logical reasoning”

(Kolmogorov, 180, p. 10), which is related to the need for proof,
substantiation, and deductions.

Applying
procedures

26
An ability to shorten the reasoning process, to think in curtailed structures.

An ability to reverse a mental process (to transfer from a direct to a reverse

train of thought)

Representation
Flexibility of thoughts – an ability to switch from one mental operation to
another; freedom from the binding influence of the commonplace and the
hackneyed.

Connection
A mathematical memory. It can be assumed that its characteristics also arise
from the specific features of the mathematical sciences, that this is a memory
for generalizations, formalized structures, and logical schemes.

An ability for spatial concepts, which is directly related to the presence of a

branch of mathematicians such as geometry (especially the geometry of
Communication

space).

Figure 3. The abilities defined by Krutetskii connected to the competencies in MCRF. The underlined
parts are those connected to physical acts. The text of the abilities is taken from Krutetskii (1976, p. 87-

both frameworks divide the mathematical competence or ability into parts.

work, when letting the students think aloud during the problem solving process
Figure 3 shows that there are some abilities described by Krutetskii that partly

Some of these parts are connected, for example the reasoning competency in

comparison, the communication competency in the MCRF is only connected to

one of the abilities in Krutetskii’s work. The problem solving competency is
comparison of MCRF (Lithner et al., 2010) with Krutetskii (1976) shows that
or fully can be connected to physical activity. The method used in Krutetskii’s

the MCRF can be interpreted in four of the abilities defined by Krutetskii. As a

shows that students’ communication competence is also of importance. The
special, since Krutetskii uses problem solving as an activity and analyses all
other abilities in that activity. The problem solving competency is therefore
marked in another way. The reasoning competency and the problem solving
competency in the MCRF are those that recur most in Krutetskii’s work.

The four frameworks discussed in this chapter are compared with the MCRF as
a base in Table 2. The comparison shows that all competencies defined in the
MCRF (Lithner et al., 2010) can also be found in the framework of Krutetskii,
although not evenly distributed as Figure 3 shows. When the NCTM (NCTM,
2000) is combined with the Danish KOM project (Niss & Höjgaard, 2011),
they together cover all competencies defined in the MCRF.

27
 Table 2
Comparison of the frameworks on the basis of the competencies defined in the MCRF.

Problem Reasoning Procedure Representation Connection Communi-

Solving cation

MCRF Solution Justify Accepted Concrete Connecting Exchange of

method is choices and (math) replacements of or linking mathematical
not conclusions. actions used abstract between for information
known in to solve the mathematical example between a
advanced. task. entities. relationships sender and a
and/or receiver.
representa-
tions.

Krutet- An activity Generalise, Operate with To abstract Operate with Operate with
skii to detect formal from the structures of mathematical
what is of structures. concrete. connections. symbols.
chief Operate with Spatial Flexibility in Clear, short,
importance. numerals. concepts. thoughts rational
Logical solutions
reasoning.
Shorten
reasoning.
Reverse a
mental
process.

NCTM An Proof is the External Make Communicate

activity, ultimate observable and connections and
the solver form. Use internal in between understand
is unaware mathematical people’s mind. different mathematical
of the conjectures topics. knowledge.
solution in all areas.
method.

KOM Detect, Follow and Carry out Understand and Symbol: to

formulate, assess informal and use different decode
delineate mathematical formal kinds of symbol and
and reasoning. mathematics. mathematical formal
specify Devise representations. language.
different informal and Communica-
kinds of formal tion: to study
math mathematical and interpret
problems. reasoning. others and to
express
oneself.

28
3.3 Conceptualising the mathematical competencies

In this study mathematical competencies are used to investigate a method to

explain differences in achievement on mathematical tests. The following section
aims to explain how the competencies are interpreted in the analysis of the tests
in this study. Each competency is presented and discussed – this starts with
presenting how the competencies are described in the MCRF. The reason for
including this comprehensive section is to make it possible to fully understand
how each competency is interpreted in the analysis.

When the word ability is used, it is in the sense of being able to, the same way
as it is used in the Danish KOM project (Niss & Höjgaard, 2011).

In the characterisations of the individual competencies below, the word

“ability” is sometimes used. It must be pointed out that this is merely a
linguistic substantivation of “being able to”, and by no means a psychological
term aimed at referring to a person’s mental personality traits general mental
faculties (Niss & Höjgaard, 2011, p. 50).

In the theory of mathematical competencies used, extra attention is given to

whether it is necessary to activate a competency or not. In the MCRF, it says
that “…a competency is the ability to handle something…”(Lithner et al., 2010, p.
161). This something is defined for each competency in the MCRF. The
purpose is to analyse the opportunities a task gives to activate a competency; to
handle something is defined in this study as an action (mental or physical) that is
necessary and cannot be avoided if the task is going to be solved.

3.3.1 Applying procedures competency

In the MCRF (Lithner et al., 2010), applying procedures competency is defined

as “a sequence of mathematical actions that is an accepted way of solving a
task.” In the analysis conceptual and procedures knowledge are both
interpreted as parts of the applying procedures competency.

Procedures are seldom studied in research in mathematics education (Säfström,

2013). In the Swedish curriculum one aim in mathematics education is to give
students opportunities to develop the ability to use and analyse mathematical
concepts (Skolverket, 2011a). Another ability the students should be given an

29
opportunity to develop is the ability to choose and use mathematical methods
in order to calculate and to solve routine tasks (Skolverket, 2011a). I interpret
those two abilities as conceptual and procedural competencies. Procedural and
conceptual competencies can be seen as intertwined where deep understanding
is thought to be reached through connecting those two competencies (Baroody,
Feil, & Johnson, 2007). The two types are developed together, each type
interacts with and influences and is influenced by the other (Voutsina, 2012),
Figure 4.

Figure 4. The iterative process between conceptual and procedural competency.

Voutsina (2012) means that the application of procedures strengthens the

conceptual knowledge and that it is important to justify and explain procedures
both to get more effective procedures and for the conceptual understanding.

3.3.2 Reasoning competency

Reasoning is “the explicit act of justifying choices and conclusions by

mathematical arguments.” (Lithner et al., 2010, p. 161). Reasoning can be
connected to mathematical proof (NCTM, 2000; Niss & Höjgaard, 2011).
Reasoning competency can be seen as the ability to follow and understand
mathematical proof and the difference between proofs and intuition and/or
special cases. The competency is also about the ability to form valid proofs
when people “devise and carry out informal and formal arguments (on the basis
of intuition)” (Niss & Höjgaard, 2011, p. 60).

Lithner defines reasoning as “the line of thought adopted to produce assertions

and reach conclusion in task solving” (Lithner, 2008, p. 257). The ultimate form

30
of mathematical reasoning is mathematical proofs (NCTM, 2000). To reason
mathematically is essential to understand mathematics and it includes
developing ideas, exploring phenomena and using mathematical conjectures in
all areas. To reason mathematically “is a habit of mind, and like all habits, it
must be developed through consistent use in many contexts.” (NCTM, 2000, p.
56). According to Lithner (2008) the line of thought does not need to be based
on formal logic and does not need to be restricted to proofs; it can even include
incorrect arguments, as long as the one who is doing the reasoning has reasons
to back it up.

Mathematical arguments are seen to be mathematical if they motivate why

conclusions are true or plausible based on mathematics, or if they are anchored
in mathematical properties (Lithner, 2008).

Reasoning is tightly connected to problem solving and modelling; Lithner

(2010) calls it the “juridical counterpart”. In the situation of working with tasks,
I interpret this as when activating the reasoning competency, it validates the
chosen strategy for the solver. Since a mathematical problem in this framework
is seen as a problem with an unknown mathematical solution, it is natural to
believe that reasoning is a part of the solution strategy. It is likely that reasoning
is an implication of problem solving, but it is not necessarily that problem
solving is an implication of reasoning.

3.3.3 Communication competency

To communicate is “to engage in a process where information is exchanged

between individuals through a common system of symbols, signs, or
behaviour” (Lithner et al., 2010, p. 165). The definition involves knowledge of
the mathematical language (symbols, signs and behaviour) and the ability to use
the language. The definition demands that individuals are involved, but it is not
necessarily a direct communication, the communication can for example be
from an author - through a book - to the reader. Another example is, from a
student - through a solution or an answer - to the teacher.

What does it mean to communicate in mathematics? It could mean that you use
a language, I would like to call it Mathematish, which means that you use words

31
and connect them grammatically correctly. Mathematicians, mathematics
educators, teachers in mathematics, all need to have something in common in
their view of mathematics; otherwise it would not be possible to talk about
mathematics together (Maier & Schweiger, 1999). Also, students being taught
mathematics need to be given an adequate picture of mathematics to enable a
discussion of the importance of mathematics for culture and society (Maier &
Schweiger, 1999). Mathematical texts are special; they endeavour
unambiguously; all objects, actions and relations should be clarified without any
doubt for misunderstanding (Maier & Schweiger, 1999). The mathematical
language also involves symbols, for example:
𝑖
𝑏
1 𝑛
∫ 𝑓(𝑥)𝑑𝑥 lim (1 + ) ∑ 𝑔(𝑥)
𝑎 𝑛→∞ 𝑛
0

Those symbols are impossible to understand without being taught in some way.
The special terminology is used in the mathematical language, which means
three things that are clearly distinguished from each other (Maier & Schweiger,
1999):

 Words that do not occur in everyday language, such as Prime numbers,

Divisor, Logarithms, Orthogonal …
 Words used in everyday language that have the same or a similar
meaning in measurements, such as Even, Triangle, Quadrate …
 Words used in everyday language in the first glimpse with a different
meaning; at a closer analysis they often have a common origin. An
example is the word Product, meaning in math the result of a
multiplication. The meaning in everyday life is the result of a production
process.

The goal of education in mathematics is to reach an understanding of

mathematics. Communication can either be seen as part of the treatment to
reach the goal – understanding mathematics or learning to communicate in
mathematics can be seen as the goal where the instructional intervention is the
treatment (Lampert & Cobb, 2003). Those two views of communication cannot
be separated. To participate in the activities of mathematics a student both
needs to be able to understand, for example, the mathematical symbols, which
he/she ought to have learned - and learn to communicate as a goal. The
mathematical language is also fundamental to come further and to engage in

32
certain areas of mathematics, for example calculus, and is used as part of the
treatment to gain deeper understanding.

Communication competency requires both a sender and a receiver therefore

“communication is about reaching shared meaning” (Säfström, 2013). In
education the sender is usually a teacher, student or a textbook and the receiver
is often a student or a teacher (Lithner et al., 2010). Communication also
requires a medium within which the sender and the receiver can understand the
communicated information (Lithner et al., 2010). The medium is often physical,
such as writing and, gesturing or auditory, such as listening and talking.

In the Swedish curriculum (Skolverket, 2011a) both the learning of

communication and the use of mathematical communication are goals that the
student should be given possibilities to develop. In the Danish KOM-project
(Niss & Höjgaard, 2011) and in NCTM (2000) communication competency is
the ability to express one’s own mathematical knowledge and to understand
others’ mathematical communication.

3.3.4 Representation competency

In the MCRF (Lithner et al., 2010, p. 163) representation competency is “the

concrete replacements (substitutes), mental or real, of abstract mathematical
entities”.

Dörfler (2006) puts the mathematical signs at the centre of mathematical

activities and includes both writing and reading in this activity. The need to use
external representations for mathematical objects derives from the abstraction
in them. “They are not accessible to the senses, not palpable, not perceivable,
they cannot be shown directly…” (Dörfler, 2006, p. 98). The representations
are either used to learn or to investigate the abstract object; they serve
predominantly a mediating role between the learner (or researcher) and the
abstract object. Dörfler means that learning through representations could help
students who believe they are incapable of learning abstract objects.

The basic idea is that the learner, by the use of external representations,
constructs or develops in his/her mind a mental representation (cognitive
structure, schema, or the like) which then permits him/her to think with and
about the respective mathematical concept (object) (Dörfler, 2006, p. 100).

33
According to Dörfler (2006) representations are important for the process of
learning mathematics; therefore the ability to interpret, use and judge
mathematical representations are important in school mathematics.
Representations can be external, such as symbols, graphs, diagrams, tables and
concrete material, or internal, such as mental pictures (Dörfler, 2006; NCTM,
2000; Niss & Höjgaard, 2011). The representation competency includes the use
of mathematical forms in expressions (Skolverket, 2011a). It also includes the
ability to understand, use and compare different mathematical representations,
to choose the best suited representation for specific situations (Niss &
Höjgaard, 2011).

3.3.5 Connection competency

Connection competency is “the process to use something that connects or makes a

link between two things, e.g. a relationship in fact or a causal or logical relation or
sequence” (Lithner et al., 2010, p. 163). Connection ability is about being able
to find meaningful relationships/connections between mathematical entities
and their representations, Figure 5.

Connection
Entity Entity

Connection Connection

Representation Representation Representation

Figure 5. Relation between representations and connections. The picture is an adaptation from Lithner
et al. (2010).

One way to speak about connections is with the word flexibility. Flexibility can
refer to the ability to adopt a known procedure to meet new demands (Baroody
et al., 2007) or perhaps to transfer knowledge from one mathematical entity to
another. To do this, it is necessary to make connections between the entities.

34
According to Kilpatrick et al (2001), flexibility is a major cognitive requirement
for solving non-routine problems. Krutetskii (1976) also mentioned flexibility,
as I interpret it in the meaning of the ability to change solution strategies for the
same problem.

The connection competency can also be seen as the ability to go back and forth
between different mathematical representations such as: symbols, formal
language and natural language (Niss & Höjgaard, 2011). The competency can
also include the ability to make connections between different contexts to get
deeper and long lasting knowledge (NCTM, 2000).

3.3.6 Problem solving competency

Lithner et al. (2010, p. 161) defines the problem solving competency as

“engaging in a task for which the solution method is not known in advance”.

What is a mathematical problem? Today, at least in Sweden a common way to

define a mathematical problem is a mathematical task where the solution
method is not known in advance by the solver (Skolverket, 2011b). This means
that a mathematical problem is not a mathematical problem for every student,
since all students do not have the same mathematical knowledge (Skolverket,
2011b). A similar definition is given in the NCTM (2000, p. 52) for the activity
of problem solving: “engaging in a task for which the solution method is not
known in advance.”

What defines a mathematical problem solving competency? In the Swedish

curriculum (Skolverket, 2011a) it involves both mental and physical activities, it
involves the ability to formulate and solve the problem and to assess the
selected strategies and methods. This is also supported by the educational
frameworks of the Danish KOM project (Niss & Höjgaard, 2011) and the
NCTM (2000): problem solving competency is about the ability to interpret,
solve and assess mathematical problems. Niss (2011, p. 55) also adds “if
necessary or desirable in different ways”.

In some educational frameworks such as the Danish KOM project (Niss &
Höjgaard, 2011), modelling competency is set as a separate competency.
Modelling competency is closely related to problem solving competency, which

35
can be seen when comparing the definition of a mathematical model with the
definition of a mathematical problem.

Mathematical problem: A mathematical problem is a problem for which the

solution method is not known in advanced by the solver.

Mathematical model: A mathematical model is a connection between the real

world and the mathematical world. Using a mathematical model either means to
de-mathematise mathematical models and interpret them into the real world, or
to mathematise real-life situations (beyond mathematics) and make
mathematical models that explain them (Niss & Höjgaard, 2011). These
situations are most often situations for students where the solution methods are
not known in advance, which makes mathematical models mathematical
problems. In NCTM (2000) it is written about contextualising mathematics,
which can be interpreted as using mathematics in mathematical models.

Problem solving can be used as a general activity in which the object of study is
engaged. This was how Krutetskii (1976) used problem solving. He assumed
that a student has to be in a situation containing mathematical problem solving
to be able to observe the student’s mathematical abilities. He also meant that
the abilities are primarily individual psychological characteristics of mental
activity (Krutetskii, 1976).

3.4 Competency Related Activities, CRA

Each competency has an investigative and a productive side (Lithner et al.,

2010; Niss & Höjgaard, 2011). This duality can be divided into three
competency related activities (CRA:s) (Lithner et al., 2010) that can be used to
describe the aspects of mastering each competency (Säfström, 2013). These
aspects are: Interpret, Do and use, and Judge (Lithner et al., 2010). Some parts
of the MCRF are questioned by Säfström (2013), for example the competency
related activity - interpret. Säfström feels that this activity can not be separated
from the other two, do and use and judge (Säfström, 2013). She believes that the
definition of interpret, that is taking in information, relates to building
knowledge, understanding, identification and recognition. The CRAs, do and use
and judge rely on interpretations and cannot be implemented without them. Also

36
the other way around, through the activities do and use, and judge, interpretations
are manifested. Säfström therefore decided to remove the first CRA interpret and
to use do and use as the productive aspect, and judge as the analytical aspect and
merge the interpret activity into both.

I chose to exclude all the three CRAs. No students’ solutions are available in
this study, and no other interaction with students has been attempted. The lack
of information about how students actually act with the tasks in the empirical
data makes it difficult to separate the CRAs from each other.

3.5 Situations for studying mathematical competencies

Mathematical competencies can be studied through observing and/or

interacting with students. Opportunities for activating the competencies can be
studied for example through investigating material (textbook tasks, tests et.)
that students meet. Examples of both are given in this section.

To be able to study the actual competencies a student possesses, Krutetskii

(1976) felt that the student must be in a mathematical activity. Some qualitative
studies involving mathematical competencies in Sweden, for example (E.
Pettersson, 2011; Szabo, 2013), have chosen problem solving as a mathematical
activity, as Krutetskii did.

Studying whether tasks give students the opportunity for imitative and/or
creative reasoning has been done in Sweden both through textbooks and task
analysis and through more students’ interactive studies (Boesen, Lithner, &
Palm, 2010). In one part of the project “National tests in mathematics as a
catalyst for implementing educational reforms” (Lithner, 2011), tasks in
national tests were analysed with the MCRF (Lithner et al., 2010) to investigate
which mathematical competencies the tasks gave the students the opportunity
to use.

37
3.6 Master a competency

Analysis of abilities is about the qualities or traits of the person who is

performing the activity; in an analysis of skills or habits it has to do with
analysing the features of the activity a person is carrying out (Krutetskii, 1976).
Ability has to do with the psychological traits of a person, skills or habits are
about something that proceeds from the concept of an operation. Krutetskii
stresses that, regardless of whether the analysis deals with skills and habits or
abilities, it is about analysing an activity. “Therefore the investigation of a
pupil´s mathematical ability is also an investigation of his mathematical activity, but
from a certain standpoint” (Krutetskii, 1976p. 72).

The competencies can be measured in three dimensions: degree of coverage, radius

of action and technical level (Niss & Höjgaard, 2011). By observing those three
dimensions, the progression of the competencies can be measured and a
person’s mastery of a mathematical competency can be assessed.

The Swedish grading system in each subject ranges from F to A; F means fail, E
is the passing level and A is the highest grade. Each mathematical competency
has knowledge requirements for the grades of E, C and A. To get the subject
grade E, C or A, all knowledge requirements for each competency must be
fulfilled for the knowledge requirement. To get the subject grade of D or B, all
knowledge requirements for the lower levels (E and C) must be fulfilled as well
as the majority of the knowledge requirements for the higher levels (C and A)
(Skolverket, 2011a). Connected to Niss (2011), it means that a student must
master each competency to a certain extent in a technical level, in the radius of
action and in the degree of coverage. Tests are a necessary tool for a teacher to
be able to assess each student in a classroom with many other students. Results
of tests are part of the assessment; a student who succeeds and achieves well in
tests will have better possibilities to get a high subject grade than a student who
achieves low on tests. In the spring semester in years 3, 6 and 9 of compulsory
school, the students also take a national test in mathematics. The national test
has a large influence on the subject grade, although it is only supposed to be
one test among others.

38
3.7 Achievement

3.7.1 The national test in year 3 vs the national test in year 6

In this study students are categorised according to relative achievement. In the

first part of the study, comparing results in national tests from the same
individuals in year 3 and in year 6, achievement categories are inspired by the
work of Pettersson (1990) and of Gagné’s model, DMGT (Gagné, 2004),
although changes are made to make categorise more suitable for the purpose of
this study.

Pettersson (1990) defines an increase in points that are more than four times
larger than the decrease (an increase of 9 points compared to a decrease of 2
points). However she works with actual scores and I work with relative
achievement.

I define a large increase or a decrease in achievement with the same size of

measure; here a large movement is defined in the following way. A large
increase or decrease has occurred if the ranking position changes up or down
more than 40% of the total number of ranking positions. There are 568 ranking
positions; 568 students took the national test both in year 3 and in year 6; 40%
of 568 is 227.2. A movement of more than 227.2 ranking positions up or down
is defined as a large movement. Only students that in year 3 ranked at the
bottom 60% or the top 60% have the possibility to make an increase or
decrease defined in this way.

Due to the statistical phenomenon “regression towards the mean”, measuring

changes in achievement will result in an increase of achievement for a group of
students with bottom results from the beginning. This means that, if the
bottom group had a lower average result than the average of the population on
the first test (year 3), the average of the group on the second test (year 6) will
approach the average of the population on the second test (year 6), see Test 1a
and Test 2a in Figure 6. An increase of the average for the group has occurred.
The opposite will happen for the group that were top achievers on the first test.

If an increase or a decrease is to be studied that cannot be explained by “the

regression towards the mean”, the increase or decrease has to be chosen such

39
that the identified group breaks the pattern of regression towards the mean,
resulting in that the group who were bottom achievers in the first test (year 3),
achieved in the second test (year 6) so that their average will be higher than the
average of the population, see Test 1b and Test 2b in Figure 6. The definitions
of a large increase and a large decrease are chosen because they break the
pattern of regression towards the mean.

Frequency Test 1a Frequency Test 2a

Average
of bottom
group on
Test 1a

Marks Marks

Average Average
of Average of Average
bottom of bottom of
group population group population

Frequency Test 1b Frequency Test 2b

Average
of bottom
group on
Test 1b

Marks

Average Average
of Average Average of
bottom of of bottom
group population population group

Figure 6. Test 1a and 2a show a movement of achievement that follows the pattern of regression
towards the mean. Test 1b and 2b show a movement of achievement that breaks the pattern.

The second difference is that Pettersson (1990) has a category for those who
were intermediate achievers. This category is removed in this study as there is
no focus on intermediate students who do not change their relative
achievement.

40
To decide who is a high achiever and who a low achiever, Pettersson (1990)
uses points. Since I use relative achievement, I use percentages limits; those
limits are chosen from the work of Gagné (2004).

The combination of the work of Pettersson (1990) and Gagné (2004) gives the
following categories:

I. Students highly ranked, top 10 %, in both years 3 and 6,

II. Students low ranked, bottom 10%, in both years 3 and 6,
III. Students who showed a large increase in ranking position from year 3 to
year 6,
IV. Students who showed a large decrease in ranking position from year 3 to
year 6.

3.7.2 The national test in year 6 vs the mathematical kangaroo

The aim of comparing relative achievement in the national test in year 6 with
the relative achievement in the mathematical kangaroo is to explore how those
who achieve highly on one test achieve on the other. To define who is a high
achiever in this study, percentage limits are used. Those percentage limits are
inspired by other research that in some way mentions a part of a population
defined by percentage limits (Advisory Committee on Mathematics Education,
2012; Gagné, 2004; Mönks & Pflüger, 2005; E. Pettersson, 2011; Renzulli, 2005;
Vialle et al., 2007). The aim of those percentage limits is different; for more
detail see chapter 2.4. Some mention a limit indicating that a certain percentage
of students need more challenges (Advisory Committee on Mathematics
Education, 2012). Others mention a percentage limit to indicate how many
students are gifted according to different models (Gagné, 2004; Mönks &
Pflüger, 2005; E. Pettersson, 2011; Renzulli, 2005), while some use top
percentage limits to choose participants for further investigation (Vialle et al.,
2007).

Inspired by those references, three percentage limits are used in this study to
define high achievers, top 5%, top 10%, and top 20%. Those who are not seen
as high achievers therefore make up the bottom 80%.

41
4 Method

This study starts with comparing results in national tests in mathematics over a
three-year period in a whole population, and comparing results in the national
test and the mathematical kangaroo in a large part of the same population. The
quantitative approach gives answers to the correctness of the observation by a
teacher with 15 years of experience. The statistical analysis of the quantitative
material identifies students that were further investigated. The second part of
the study investigates tasks and what mathematical competencies they give the
opportunity to activate. Further, a method will be investigated that explores
how mathematical competencies can explain differences in achievement
between two identified groups.

Empirical data in this study are results in three mathematical tests all of them
with different purposes.

 The national tests in mathematics given in year 3 (2009) tests the passing
level according to the curriculum (Skolverket, 2010),

 The national test in mathematics given in year 6 (2012) tests

mathematical knowledge according to the curriculum, focusing on the
passing level (Skolverket, 2012),

 The mathematical kangaroo given in year 7 (2013) is not connected to

the syllabus. The aim of the mathematical Kangaroo is to stimulate
interest in mathematics and to arouse curiosity and lust to learn
mathematics (Nationellt centrum för matematikutbildning, 2013).

4.1 Design of the study

The full study contains two parts, study 1 and study 2.

Study 1 is descriptive, where movements in relative achievement are the object

of study. It is relative achievement on the individual level that is compared
between the tests, although they are analysed as groups. Movements over a
four-year period (2009―2013) through the results in three tests are analysed,

42
two curriculum bounded and one non-curriculum bounded. The same
individuals are followed through the years; all students in public schools are
involved in the national test since results are official data. To employ students’
results in the mathematical kangaroo, students and guardians needed to sign an
informed consent form, which reduced the number of participants to 264 from
611. In study 1, two comparisons are made:

 Relative achievement on the national test in year 3 (2009) is compared

with relative achievement on the national test in year 6 (2012), n=568,

 Relative achievement on the national test in year 6 (2012) is compared

with relative achievement on the mathematical kangaroo in year 7
(2013), n=264.

Study 2 aims to investigate a method that explores how mathematical

competencies can be used to explain why some students are high achievers in
one test but not in another. The two tests compared are the national test in year
6 and the mathematical kangaroo in year 7. Test competency profiles are
constructed through analysis of the tests on the task level, exploring which
mathematical competencies a student is considered to be in need of activating
to solve the task. Individual results in the tests, on the task or part level,
produces individual competency profiles, in turn leading to the group of
students identified in study 1 being able to be analysed. Study 2 involves five
steps:

 Competency analysis of the two tests involved on the task level,

 Producing competency profiles for the tests and individual competency

profiles based on results in each task in the mathematical kangaroo,

 Comparison of competency profiles in the mathematical kangaroo for

the identified groups,

 Investigating differences between the two identified groups in activated

competencies by means of the mathematical kangaroo,

 Investigating tasks in the mathematical kangaroo that are of special

interest.

43
4.1.1 The national test in year 3 (2009)

The national test in year 3 (2009) is connected to the curriculum that existed
before 2011. It was the first national test given in year 3 in Sweden (Skolverket,
2008). The aim of the test was to assess students’ knowledge and to support
teachers in the process of assessing students’ fulfilment of goals in the
curriculum. The test contains nine parts, A―J; three of them (B, C and J) are
meant to be solved in pairs or in groups, and one part (J) is a game (Skolverket,
2010). One part of the test (part A) is about self-assessment; the students are to
tell how secure they feel in different mathematical situations. What can be said
about the content and maximum points on each part is given in Table 3.
Getting full points on each part results in a test result of 93 points. Public data
on the test from the municipality office gives each student’s points on each part
except part A of the test.

44
Table 3
The national test in year 3 (2009); mathematical content and maximum points for each part.

Part Content Maximum points

A (individual) Self-assessment No points

B (group) Follow instruction to build a 3D object Together with C, 8 points

C (pair) 2D and 3D geometrical object. Positions Together with B, 8 points

measurements

D (individual) Properties of geometrical objects, 2D and 3D 7 points

E (indicidual) Mental arithmetic, addition and subtraction, Together with F, 39

without context points

F (individual) Mental arithmetic, addition and subtraction, with Together with E, 39

context points

G (individual) Written calculation, addition and subtraction 14 points

H (individual) The meaning of the four basic operations 9 points

I (individual) Length, measurements and comparisons 8 points

J (group) A game about number sense 8 points

4.1.2 The national test in year 6 (2012)

The national test in year 6 (2012) was obligatory for all students and, was the
first national test for year 6 connected to the curriculum Lgr 11 (Skolverket,
2011a). The test focuses on the lowest passing level according to the curriculum
(Skolverket, 2012). There was one oral part on the test (part A) in which the

45
content is geometry. In the three written parts (B―D), the content is: number
sense, algebra, geometry, problem solving, statistics and relationships and
changes. For parts B and C, some tasks demand written solutions in the test
booklet and some “only” an answer. Part D demands written solutions on a
separate paper. At the time of this writing the test is still classified, and thus the
description of this test must be limited. What can be said about the content and
maximum points on each part is given in Table 4. Getting full points on each
part results in a test result of 106 points. Public data on the test from the
municipality office indicate each student’s points on each part of the test.

Table 4
The national test in year 6 (2012); mathematical content and maximum points for each part.

Part Content Maximum points

A (oral in groups of 3-4 students) Geometry Credit (1) or

no credit (0)

B (individual, no calculator) Number sense, Algebra, 38, of which 9 points are of

Geometry and Problem Solving higher level

C (individual, no calculator) Number sense, Algebra, 37, of which 3 points are of

Geometry higher level

D (individual, with calculator) Number sense, Probability and 30, of which 9 points are of
statistics, Relationships and higher level
changes, Problem solving

4.1.3 The mathematical kangaroo (2013)

The mathematical kangaroo in Sweden was given on five levels in 2013, Milou
for preschool and years 1―2, Ecolier for years 3―4, Benjamin for years 5―7,
Cadet for years 8―9 and one track, Junior, for upper secondary school. Students
in year 7 participated in this study, and therefore Benjamin is the choice here.

The mathematical kangaroo is a multiple choice test. For Benjamin there are five
choices in each task. All tasks are dichotomously scored, either credit or no
credit, although there are three levels in the scoring. The first seven tasks give 3
points, task numbers 8―14 give 4 points and task numbers 15―21 give 5
points.

46
The multiple choice tasks give an opportunity to test whether one of the five
options is a possible solution, but the tasks can also be solved mentally or in
writing. No solutions are required of the students. Getting full points on each
part results in a test result of 84 points.

The national centre of mathematics (NCM) in Sweden distributes the

mathematical kangaroo to the schools that want to participate in the
competition. After the implementation of the competition, it is possible to
download all material, the test, mark scheme and suggestions for how to
continue the work with the students (Nationellt centrum för
matematikutbildning, 2013). In this material, NCM has also divided the task
into four topics, numbers, geometry, time, problem solving and logical
reasoning, although it is not described how this division of the content is made.
The division of the contents according to NCM is shown in Table 5.

Table 5
The mathematical kangaroo (Benjamin) in year 7 (2013); mathematical content and maximum points for each part,
described by NCM.

Part Content Maximum points

3 mark part Numbers – 3 tasks 9 points

Geometry – 2 tasks 6 points
Time – 2 tasks 6 points

4 mark part Number – 1 task 4 points

Geometry – 3 tasks 12 points
Problem/Logic – 1 task 4 points

5 mark part Number – 2 tasks 10 points

Geometry 1 task 5 points
Problem/Logic – 4 tasks 20 points

Task numbers 8 and 13 These two tasks are not placed 8 points
in any of the four contents

4.2 Sample

The sample consists of the students who participated in the study; those are a
part of the whole population of students in year 7 in the spring of 2013 in a
municipality in Sweden.

47
For all students in public schools in a municipality, the results of the national
tests in mathematics, given in year 3 (2009, n=654) and in year 6 (2012, n=611),
were collected. The results are official data. Most students (n=568) had results
on both tests. The gap in the data is explained by some students moving to or
from the municipality and some students having attended private schools
before year 6 or choosing a private school after year 3. Results of national tests
from students attending private schools are not official.

All students in public schools (8 schools) were asked to do the mathematical

kangaroo in year 7 (spring 2013). The students and their guardians were asked
to allow the results on the mathematical kangaroo be used in this study; 264
students and guardians agreed to do so, and 247 of those had results in all three
tests involved in the study. The sample used in the study is thereby
approximately 43 % of the population.

4.2.1 Representativeness of the sample

When using statistics and talking about the whole population, one must specify
what is meant by the whole population. Is it all the humans in the world, or is it
all 7-year old children in a specific school? If a whole population participates in
a study, it must nevertheless be seen as a sample; for example, the study is done
in a specific time range or at a specific geographic place. This means that the
results can not be generalised for something that will happen ten years later or
happened ten years earlier or at another geographic place (Lisper & Lisper,
2005).

In this study, results of different mathematical tests from a sample of a whole

population are compared. It is important to analyse how well the sample
represents the population.

The statistical program R (R Core Team, 2013) was used for statistical analysis
of the representativeness of the sample, and a significance level of 5% was used
in the statistical tests. The whole population is counted as those students with
results in the national test in year 6 who went in year 7 to public schools in the
spring of 2013. The whole population was invited to participate in the study,
although naturally not everyone did. The sample group is those who chose to

48
participate in the mathematical kangaroo and also agreed to let their results be
used in this study.

When a sample is chosen from a population, it is important to investigate how

well the sample pictures the population. This can be done by comparing the
statistical parameters m (arithmetic mean value for the selection) with µ
(arithmetic mean value for the population), and s (standard deviation for the
selection) with σ (standard deviation for the population) (Lisper & Lisper,
2005).

4.2.1.1 Statistical parameters

To statistically investigate whether the sample is a representative sample, mean,

maximum, minimum, lower and upper quartile, median values and standard
deviation of the results in the national tests in year 3 and year 6 were compared
between the whole population and the sample group, Table 6. Boxplots were
drawn to visualise the statistical parameters, Figure 7 and Figure 8.

Table 6
Comparison of statistical parameters between the population and the sample.

National test National test National test National test

grade 3 grade 3 grade 6 grade 6
population sample population sample

Minimum 19 56 9 23

Lower quartile 75 78 61 68.75

Median 81 83 74 79

Upper quartile 86 87 85 88.25

Maximum 93 93 102 101

Mean 79.28 81.80 71.57 76.92

Standard 9.25 7.00 18.20 15.33

deviation

49
Figure 7. National test year 3; population no 1, sample no 2.

Figure 8. National test year 6; population no 1, sample no 2.

In Table 6, Figure 7, and Figure 8 it is shown that the range of variation is

larger in the population than in the sample. The reason for this is that low
achieving students are less represented in the sample as compared to in the
population; this also has the effect that the median for both tests is higher in
the sample than in the population.

4.2.1.2 Frequency distribution graph

If a test is constructed so that most students pass it, the test has a ceiling effect.
A test that is very difficult, resulting in most students failing it, has a floor
effect, Figure 9 (Statistik för samhällsvetare, Lisper & Lisper2005).

50
 Frequency Frequency

Points in the test Points in the test

in percentage in percentage
100 % 100 %

Figure 9. Left graph illustrates the ceiling effect and the right graph illustrates the floor effect.

One way to judge whether empirical data are normal or skewed in any direction
is to draw a frequency density graph of the data. The frequency distribution
graph of the sample also has to be compared with the frequency distribution
graph of the population, with the aim to show whether the sample describes the
population (Lisper & Lisper, 2005). Differences and similarities as well as
s on the test", ylab="Density", xlim=c(0,100), ylim=c(0,0.06))

arks on the test", ylab="Density", xlim=c(0,100), ylim=c(0,0.06), lty=3)

ceilings effects are visualised through frequency distribution graphs. To
tics
compare the distribution of the sample and the population, frequency density
graphs were drawn for the results in the population and the sample for each
test, Figure 10 and Figure 11.

Frequency graph, national test in year 3

Frequency

Achieved points in the test

Figure 10. Frequency graph of results in the national test in year 3; the population solid line, the sample
dotted line.

51
xlab="Marks on the test", ylab="Density", xlim=c(0,110), ylim=c(0,0.03))

m=c(0,0.03), lty=3)

atistics

Frequency graph, national test in year 6

Frequency

Achieved points in the test

Figure 11. Frequency graph of results in the national test in year 6; the population solid line, the sample
dotted line.

In the frequency graphs, Figure 10 and Figure 11, it can be seen that the sample
represents those students who achieved highly on both national tests more as
compared to those that had low achievement. It is also visualised that both
national tests have a ceiling effect, which means that data (actual results on
tests) must be transformed, for example to ranked data, to be able to do
statistics (Polit, 2008).

4.2.1.3 Scatter plot

Plotting the actual results in the national test in grade 6 against the actual results
in the national test in grade 3, and marking (in black) those students who
belong to the sample, gives a picture of the distribution of the sample over the
population, Figure 12.

52
ead.csv2("130808filt4.csv", header=TRUE)
ad.csv2("130808filt5.csv", header=TRUE)
lt4)
m3,Sum6, main=" ", xlab="Marks on national test in Year3 ", ylab="Marks on national test in Year6 ", xlim=c(0,100), ylim=c(0,110), col="gray47")
=TRUE)
ilt4)
lt5)
um3,Sum6, main=" ", xlab=" ", ylab=" ", xlim=c(0,100), ylim=c(0,110),pch=21,bg="gray")
Texten inklistrad via textrutor över gammal R-text

Achieved points in the national test in year 6

Achieved points in the national test in year 3

Figure 12. Results in the national test in year 6 versus results in the national test in year 3.

There is a lack in the sample of students who had low achievement in both
national tests. However, those students are not very many (12―14 students),
which can be seen in Figure 12.

Some students (n=43) have results in the national test in year 6 but not in the
national test in year 3. Of those, some (n=17) are included in the sample, i.e.
they completed the national test in year 6 and agreed to participate in the study
with their results in the mathematical kangaroo. The distribution of their points
in the national test in year 6 is shown in Table 7.

53
Table 7
Distribution of points in the national test in year 6 for students who do not have results in the national test in year 3.

Statistical measures Results from year 6 but not The sample - with results from
from year 3 (n=47) year 6 but not from year 3
(n=17)

Minimum 12 21

Lower quartile (25%) 50 32

Median (50%) 66 39

Upper quartile (75%) 86 49

Maximum 100 71

4.2.1.4 Godness of fit test

To find out, statistically, how well the sample represents the population, a
“goodness of fit” test, the 2 test, was used (Wackerly, Mendenhall, &
Scheaffer, 2002). With the null hypothesis, H0:

There is no significant difference between the distribution of the sample and

the population.

The 2 test is suitable to use since it is possible to compare two groups even if
data not are normally distributed. The test compares the sample with the
population through their results on the national test in year 3 and their results
on the national test in year 6, data are given in appendix 1 and appendix 2.

Comparing the sample with the population through the national test in year 3,
shown in appendix 1, gives a calculated 2-value of 28.68615. The degree of
freedom, 23, with the critical 2 value (α = 0.05) 35.172, says that the null
hypothesis is retained.

The calculated 2 value for year 6 is 36.99813. The degree of freedom is 24 and
has the critical 2 value (α = 0.05) of 36.415. Since the calculated value is
greater than the critical one, there is a significant difference between the sample
and the population. Looking at the table in appendix 2, it is very clear that the

54
reason is mainly the three groups with low points on the test, 0―32 points,
33―40 points and 56―58 points, meaning that the reason for why the sample is
not representative of the population is primarily because there is a lack of
students who participated who were low achievers in the national test in year 6.

The sample is not a perfect representation of the population, which the 2 test
shows. The discrepancy can be explained by there being a lack of students who
were low achievers in the national test in year 6. In the study, students will be
identified by different criteria; if they are identified by their low achievement in
the national test, there is a chance that those students are underrepresented in
the population.

4.2.1.5 Change in ranking position

Statistical parameters describe the population’s location and shape in the

frequency distribution graph. For some measurements, knowledge of the
location on the scale of measurements is arbitrary (Wackerly et al., 2002). It is
more important to compare the location in comparison to other data. In this
case, data can be ranked and analysed with nonparametric statistical methods.
For example, if a group of people is defined by a specific percentage of a
population, it is only interesting to compare that group of people relatively to
each other, which ends up in ranking them. For example, in Gagné’s
differentiated model of giftedness and talents (Gagné, 2000), the gifted and
talented are seen as the top 10 % of a population, and this top is defined by
means of individual performance relative to the population.

If the distribution of data is not symmetric, for example there is extreme

skewness or the presence of a ceiling or floor effect, data must be transformed
to meet the requirements of certain statistical tests (Polit, 2008). One
transformation can be to rank the data; ranked data leads to the choice of
nonparametric statistical methods. Nonparametric tests shall be used if the
distribution of the population is unknown and if no assumptions are made
about the distribution. For a sample with large amount of data, nonparametric
tests should also be used if the data are ranked (Lisper & Lisper, 2005). An
example is used to describe the difference between empirical data best suited
for nonparametric methods and empirical data suited for parametric methods.

55
Example

In a sprint race among 8-year old children, there is a winner, a second place and a third place
and so on, in the same way as in a sprint race among elite athletes. Those numbers do not say
anything about the performance or about the differences in performance. The distributions of
the positions of winner to tenth place look the same for the children as for the elites. This
kind of data is more suited for nonparametric methods. In the same sprint race, individual
times for the races can be measured, which gives individuals’ actual performance. Parametric
methods are more suitable for analysing this kind of data.

For ranked paired data, where pair represents “pre-treatment” and “post-
treatment” observations, shifts in location can be analysed with a Wilcoxon
signed rank test (Hollander, Wolfe, & Chicken, 2014). This also leads to making
it possible to use a Wilcoxon signed rank test to compare individual ranking
positions for two measurement points, for example test A and test B. The null
hypothesis will in that case be: the ranking position is the same for each
individual on both tests. If the null hypothesis is retained, the (pseudo)median
is equal or close to zero, meaning that there has not been a shift in location. A
large p-value, p ≥ 0.05, says that there has not been a shift in location. The
changes measured by the Wilcoxon test is in the individuals and how those
changes are placed (in ranking). It is a measurement of “individual movement”
from test A to test B.

The tests used in this study are said to be not comparative because of the
different aim in each test. The purpose of the national test in year 3 is to
support formative assessment and it only tests whether the student has reached
the curriculum goal for year 3, the passing level (A. Pettersson & Boistrup,
2010). The national test in year 6 tests students’ achievement according to the
curriculum with some possibility to show deeper mathematical knowledge
(Skolverket, 2012). The mathematical kangaroo is not connected to any
curriculum; the aim of the mathematical kangaroo is to stimulate interest in
mathematics and to arouse curiosity and lust to learn mathematics (Nationellt
centrum för matematikutbildning, 2013). The Wilcoxon signed rank test is used
to statistically investigate whether students’ ranking position changes between
the tests. To be able to make a Wilcoxon signed rank test, values from the two
groups that will be compared must be in the same interval. The results of the
tests were therefore recalculated to percentages of the maximum value.

If the Wilcoxon test says that there has been a change, one must consider what
the cause of the change is. For top achieving students, Gagné (2005) stresses
that most of them maintain their ranking position.

56
To investigate whether movements in ranking positions occurred, the Wilcoxon
signed rank test was carried out first comparing the national test in year 3 and
the national test in year 6, Table 8, and second comparing the national test in
year 6 and the mathematical kangaroo in year 7, Table 9.

Table 8
Wilcoxon test – the national test in year 3 and the national test in year 6, data from students who have results in the
national test in year 3 (2009) and in year 6 (2012).

Wilcoxon values Percentage values

V 158038
p-value <2.2e-16
95 percent confidence interval [15.14599,∞[
(pseudo)median 16.06755

Table 9
Wilcoxon test – the national test in year 6 and the Kangaroo test; data from students who had results in the national test
in year 6 (2012) and the mathematical kangaroo in year 7 (2013).

Wilcoxon values Percentage values

V 34595.5
p-value <2.2e-16
95 percent confidence interval [23.2143,∞[
(pseudo)median 24.76195

A small p-value indicates that there has been a shift in the location of ranking
positions among the students. The value of the (pseudo)median is 16.1 when
the two national tests are compared and 24.8 when the national test in year 6 is
compared with the mathematical kangaroo. Since the percentage results of all
tests were used in the Wilcoxon signed rank test, the values of the
(pseudo)median indicate that there have been large movements in the ranking
position between the tests among the students.

There may be many reasons for a change in ranking position: trauma, change of
teacher, being ill on the test day, different kind of test etc. People have told me
“of course you get different results when using different test” but since it
surprises not only me but also other teachers (Mattsson, 2013) that there are
students who are low achievers in the national test and high achievers in the
mathematical kangaroo, it is interesting to analyse differences. Some teachers
suspect that those students are gifted in mathematics (Mattsson, 2013). In this
study, reasons for changes in ranking position between the national test in year
6 and the mathematical kangaroo are investigated by means of the

57
opportunities offered by the tasks to activate mathematical competencies; the
MCRF framework (Lithner et al., 2010) is used in the analysis.

4.3 Procedure of the study

The design of the study is visualised in Figure 13. This chapter discusses the
practical parts of the study.

Contact
Contact Schools
Municipality office

Data collection Data collection

kangaroo national tests

Processing data

Anonymise data

Representativeness

Analyse
data

MC analysis
Relative MC analysis of
on individual
achievement the tests
level

Interpreting
results

Figure 13. The process of the study, not related to time.

58
The practical process started with planning the study and having it inspected by
the ethics committee at the university.

All secondary public schools were contacted to ask whether the schools wanted
to participate. Contact with a teacher responsible for mathematics education at
each school was then established.

Informed consent asking permission to connect results of the national tests

with results on the mathematical kangaroo was delivered, through the teachers,
to all students and their guardians, see appendix 3 and appendix 4. Here, it was
necessary for the students to agree without the influence of their guardians, and
thus the guardians and the students received informed consent forms.

All material for the implementation of the mathematical kangaroo was

distributed to the schools by the author. After the implementation, which was
done by the teachers at the schools, the author collected all the material. All
tests were corrected by the author; after collecting the data needed for the
study, the tests were returned to the schools and the teachers returned them to
the students. Data needed for the study were results in each task among the
participating students.

The municipality office was contacted to get the results of the national test
among the population of students who were asked to do the mathematical
kangaroo. The municipality office was informed about the ethical
considerations and that the study had been reviewed by the ethics committee at
the university. National test results in public schools are official data in Sweden,
however.

The national test from 2012 among students in year 6 was classified at the time
of the study; a special agreement was made with the national agency of
education to gain access to the test.

When all empirical data were collected, all the individuals were anonymised
before the analysis started.

The representativeness of the sample was explored through statistical analysis.

An analysis of data aiming to answer the research questions was then made; this
analysis was done in three separate steps: Step 1, analysis of relative achievement
between the national test in year 3 and in year 6; Step 2, analysis of relative
achievement in the national test in year 6 versus the mathematical kangaroo;

59
Step 3, analysis of the mathematical competencies the tests gave opportunity to
activate and analysis of the distribution of activated mathematical competencies
in the national test in year 6 and in the mathematical kangaroo for students of
special interest.

The last step in the study is to interpret and discuss the results.

4.4 Empirical data collection

The national test is obligatory and individual results of the test are recorded at
the municipality office. Results of the national test from public schools are
official data; therefore those were collected from the municipality office at a
visit. The data of individuals with a secret identity are not public, and their data
are thus not included in the study.

It is optional to participate in the mathematical kangaroo. In this study, the

researcher contacted the headmasters at each school, who first gave permission
to carry out the study at their school and second gave the researcher the name
of a teacher for discussions. In order not to increase the work load for teachers
too much, most of the work with the implementation of the mathematical
kangaroo was done by the author. All material needed for the implementation
was distributed to each school by the author. The students’ ordinary
mathematics teachers received instructions about the implementation of the
mathematical kangaroo, both written, appendix 5, and verbal. The author
corrected all the tests and returned them to each school after collecting all the
necessary data, that is: results in each task connected to each student who
participated in the study. The mathematical kangaroo did not require more than
one hour of the teachers and the students. Since it is a multiple choice test, it
was also possible to gather a large amount of data and be handled in a one-
woman project.

60
4.5 Data analysis

All data were made anonymous: both the official results in the national tests
and the results in the mathematical kangaroo. A computer program (Wireflow
AB) produced a SHA-256 hash string of each identification number combined
with cryptographic salt (comparable to stretching passwords) to make it
impossible for an unauthorised to go back and identify someone.

The students were ranked according to their results to get their relative
achievement in each test, and an analysis was carried out in three steps:

Step 1: Comparing movements in relative achievement in two national tests

(year 3, 2009 and year 6, 2012) over a three-year period. The movements are
measured by categorising students into four groups, I-IV, defined in chapter
3.7.1.

Step 2: Comparing relative achievement in two different tests, one non

curriculum-bounded – the mathematical kangaroo – and one curriculum-
bounded – the national test. The mathematical kangaroo was given to the
students in year 7 (2013), while the national test had already been done in year 6
(2012). Analysing relative achievement in those two tests identifies groups of
high achieving students using percentage limits 5%, 10% and 20%, see chapter
3.7.2.

Students’ relative achievements in the national test in year 6 and in the

mathematical kangaroo in year 7 are compared. The focus is on students among
the top 20% on one test but not on the other. The overlapping group with
students among the top 20% on both tests is not included in the analysis.

It is investigated how high achievers on one test achieved on the second. The
students identified in this part of the study were explored by means of the
mathematical competencies they are given the opportunity to activate in the
tasks and the tests in the third step of the study.

Step 3: Analysis of what mathematical competencies the tasks give the

opportunity to activate in the national test in year 6 and in the mathematical
kangaroo. The analysis results in competency profiles for each test and for the
students identified in step 2; comparing those profiles investigates whether
differences in achievement can be explained by mathematical competencies.

61
4.6 Guide to analysing mathematical competencies in tasks

Analyse of the tasks is based on an unpublished analysis document used in

Boesen et al. (2014)5. The analysis of the tasks uses the perspective of what is
required and reasonable (RR) to expect of a student in year 7. The key idea in this
analysis is to show what is necessary, working through the so called exclusion
principle. In analysing the tasks for each competency, asking – in a required and
reasonable solution – is it necessary to activate the competency.

The guide in Table 10 is used to analyse tasks in the mathematical kangaroo and
in the national test in year 6. The results of the analysis indicate the opportunity
the test gives to activate the six competencies in the MCRF. To simplify, the
analysis will judge what is required of and is reasonable to believe that a student
in year 7 uses in the solution process of the task (Boesen et al., 2014). Through
experience, Boesen et al. (2014) have noticed that students do what is required
to solve a task but not much more; this is verified in the 15 years of teacher
experience of the author. In the analysis of the task, there is an underlying
question for each competency.

In a required and reasonable solution that solves a task correctly, is it necessary

to activate the competency? By necessary it is meant: expected to be necessary
for a student in year 7.

The analysis procedure was:

First, to find a reasonable solution or solutions. Reasonableness is determined

by teacher experience of how students normally solve tasks. The solutions
made for the analysis are possible solutions and were produced by the author.
A task that does not demand a written solution but only an answer is still seen
to have been solved when a correct answer is given. The student could have
solved the task mentally or physically. The empirical data do not include any
student solutions.

Working through each of the six competencies in the MCRF, try to show that it
is necessary to activate the competency in a reasonable solution (that fulfils the
task conditions).

5
This study had access to written analysis documents that are used but not published in the
referenced article.

62
 - If it is not necessary to activate, for this task, the competency is given
the classification ‘0’.
- If it is necessary to activate the competency, for this task, the
competency is given the classification ‘1’.

There is one exception from this analyse procedure in this study, although, the
problem solving competency was not analysed in this way. To get an answer as
to whether the tasks in the mathematical kangaroo are seen as problems, eight
experienced teachers used to teaching mathematics in year 7 were asked to
mark the tasks they interpreted as problems according to the following
definition.

A mathematical problem is a problem where the solution method is not

known in advance for the solver (NCTM, 2000, p. 52).

Those tasks that the majority (five or more) of the teachers agreed to be
problems were seen in this study as giving an opportunity to activate the
problem solving competency and were given the classification ‘1’.

The national test, which was classified material at the time of this study, could
not be discussed with teachers in the same way. The tasks that the constructors
(Stockholm University, 2013) defined as problem solving tasks are classified as
giving an opportunity to activate the problem solving competency and are given
the classification ‘1’.

Table 10 is used as a guide for analysis of the competencies in a task, for each
task, and the reasonable solution. The guide is used and finally offers a
classification of ‘1’ or ‘0’ for each competency and task.

63
Table 10
Competency analysis guide used for each task.

Competency Question to ask in the analysis of the task. Classifi- Classifi-

For all competencies, if the answer is: cation ‘0’ cation ‘1’
Yes – the classification is ‘1’
No – the classification is ‘0’

Applying Is it necessary to activate the applying procedures

procedures competency in an RR solution? That is, is it necessary to
(App) involve a sequence of mathematical actions in an RR
solution?

Reasoning Is it necessary to activate the reasoning competency in an

(Rea) RR solution? That is, is it necessary to involve an explicit
act of using a mathematical argument to justify choices
and conclusions in an RR solution?

Communication Is it necessary to activate the communication competency

(Com) in an RR solution? That is, is it necessary to involve a
process where mathematical symbols, signs and/or
language are exchanged between the sender (the test) to
the receiver (the student) in an RR solution?

Representation Is it necessary to activate the representation competency

(Rep) in an RR solution? That is, is it necessary to use
representations in an RR solution?

Connection Is it necessary to activate the connection competency in

(Con) an RR solution? That is, is it necessary to draw a
connection between different entities and/or different
representations in an RR solution?

Problem solving No question

(Pro) For the mathematical kangaroo - statement of
experienced teachers decides.
For the national test – constructors’ choice decides.

64
The analysis results in a reduced competency matrix for each task, Figure 14.

Task 1 class
App 1
Rea 0
Com 1
Rep 0
Con 1
Pro 0

Figure 14. Reduced matrix showing the competency classification for task no. 1

Examples of the analysis of tasks from mathematical kangaroo are shown in

appendix 6.

4.7 Competency profile

The mathematical kangaroo has 21 tasks, 7 tasks give 3 points, 7 tasks give 4
points and 7 tasks give 5 points. Each student identified will be given a matrix
showing their results, Figure 15. Row one represents three-point tasks, row two
four-point tasks, and row three five-point tasks. Number 1 in a cell indicates
credit; 0 indicates no credit.

KT-S1
3m 1 0 1 1 0 1 1
4m 1 1 0 1 1 0 0
5m 0 1 1 1 0 1 1

Figure 15. Results on the task level in the mathematical kangaroo for student S1.

The analysis will result in one summary competency matrix for each identified
student that tells how often that student used each competency and how often
each competency is used in relation to the other that is used, Figure 16. ∑ class
is the sum of all times the students received a point in a task that gave the
opportunity to activate the competency. Rel class is the relative distribution of
the competency.

65
 Sum Rel class
∑ l
KT-S1

Pro 2 0.095
Rea 6 0.286
App 1 0.048
Rep 3 0.143
Con 4 0.190
Com 5 0.238
∑ 21 1

Figure 16. Student matrix of activated competencies for student S1.

A competency profile for the tests and for each student identified in study 1 is
made. The competency profile indicates how much each competency is
activated in relation to the total amount of activated competencies. An example
is given to explain how a competency profile is created.

Example
A test has 10 tasks and the competencies are distributed among the tasks in the example test
shown in Figure 17. To solve all 10 tasks, it is necessary to activate the competencies as
summarised in column 2 in Table 11. Relative values for each competency are shown in
column 3. In this example, there is a specific student that has failed in tasks no. 2, 7 and 8.
The student is therefore seen to have activated the reasoning competency two times less, the
communication competency one time less, the representation competency two times less and
the problem solving competency two times less than possible, summarised in columns 4 and 5
in Table 11. Data from column 3 result in the competency profile for the test; data from
column 5 result in the competency profile for the specific student, Figure 18.

66
 Task 1 Class Task 2 Class Task 3 Class Task 4 Class Task 5 Class
App 1 App 0 App 1 App 0 App 1
Rea 0 Rea 0 Rea 1 Rea 0 Rea 0
Com 1 Com 0 Com 1 Com 0 Com 1
Rep 0 Rep 1 Rep 0 Rep 0 Rep 1
Con 0 Con 0 Con 0 Con 0 Con 0
Pro 0 Pro 0 Pro 0 Pro 1 Pro 0

Task 6 Class Task 7 Class Task 8 Class Task 9 Class Task 10 Class
App 1 App 0 App 0 App 1 App 1
Rea 0 Rea 1 Rea 1 Rea 0 Rea 0
Com 0 Com 1 Com 0 Com 1 Com 0
Rep 0 Rep 0 Rep 1 Rep 0 Rep 0
Con 1 Con 0 Con 0 Con 1 Con 0
Pro 0 Pro 1 Pro 1 Pro 0 Pro 0

Figure 17. Competency distribution in the task in an example test.

Competency profile Competency profile
for a test for a student
0,40
0,40 0,40
Table 11
0,35 0,35
0,27 0,27
Example of data used
0,30
0,23to produce a competency profile
0,30 for one test and one specific student.
0,25 0,25
0,20
Competency 0,14 0,14 0,14 0,20of times
No. Relative 0,13 No. of times Relative
0,15 0,09 0,15
activated 0,07 values activated values
0,10 0,10 0,07 0,07
Maximum Maximum By the student By the student
0,05 0,05
0,00 0,00

Applying procedures (App) 6 0.27 6 0.40

Reasoning (Rea) 3 0.14 1 0.07

Communication (Com) 5 0.23 4 0.27

Representation (Rep) 3 0.14 1 0.07

Connection (Con) 2 0.09 2 0.13

Problem solving (Pro) 3 0.14 1 0.07

Total amount of competencies 22 1.01 15 1.01

67
 Competency profile

Pro;
Pro; 0,14 0,07

Con; 0,13
App; 0,27
Con; 0,09 App; 0,4
Rep; 0,07
Rep; 0,14
Rea; 0,14
Com; 0,27
Rea;
Com; 0,23 0,07

Test profile Student profile

Figure 18. Example of a competency profile for one test and one student.

A competency profile for a test shows the possibilities that have existed to
activate the six competencies in relation to each other. The numbers in Figure
18 show how much a competency is given opportunity to be activated by the
tasks in the test in relation to the total amount of possible competencies on the
test.

A competency profile for a student shows how the competencies are activated
in relation to each other in the tasks solved by the student. The numbers in
Figure 18 show how much a competency is activated by the student on the test
in relation to the total amount of activated competencies by the student on the
test.

The test profiles give possibilities to make comparisons between opportunities

the test gives to activate the competencies and how students activated the
competencies, both on an individual level and group level. The profiles are not
correlated to achievement; a student profile can look exactly or almost exactly
like the test profile without there being any demand that the student has full
points in the test.

68
4.8 Validity and reliability

4.8.1 Sample

The whole population of students from a school year in a municipality was

invited to participate in the part of the study that includes the mathematical
kangaroo; those who participated make up the sample. Statistical measurements
were made to validate whether the sample was representative for the
population; these are presented in chapter 4.2.1. The results of the statistical
measurements (chi square test) are that the sample is representative, with p =
0.025. The p-value is very close to 0.05; it is obvious in the data (appendix 2)
that the reason for not achieving a better representativeness is a lack of students
that were low achievers in the two national tests.

4.8.2 Non-participants

The students that did not participate in the study are primarily those who
achieved low on the national test in both years 3 and 6, this is visualised in
Figure 12. This affects the results in that there could be more students that
achieve highly in the mathematical kangaroo and low in the national test in year
6 than are identified in the sample.

4.8.3 Error in identification

Empirical data in this study are students’ results in three mathematical tests.
There is a risk that there are individuals that did not achieve the way they
should have under normal conditions. Factors such as stress, illness, social
circumstances etc. can influence individual achievement. It is therefore possible
that some students should have been identified that were not owing to factors
like those mentioned above. In turn, this leads to there being a possibility that

69
some of the students identified should not have been identified, since ranking
and percentages are used as identification criteria. When using identification
criteria, one must always chose a limit, which gives a risk of some identification
errors. Although students are treated as a group in this study, which reduces the
identification factors, the probability that all students had a bad day is
considered low.

4.8.4 Coding

The coding of personal numbers to hash codes was done by a coding expert
(Wireflow AB). The expert constructed a program that coded original data. The
program added cryptographic salt to the personal numbers before the SHA-256
hash was calculated. This makes it impossible for any unauthorised person to
go back to and identify individuals. Use of an expert for coding and security in
the program makes the coding reliable and safe.

4.8.5 Required and reasonable solution

The best way to find solutions for the task analysis would have been to let a
large group of students in year 7 produce solutions that could be used for the
analysis. This would have been time consuming, however. To make it possible
to manage the study, solutions were created by the author. The solutions were
compared with authentic students’ solutions in the classified material from the
national test in year 6. The solutions produced by the researcher and the
authentic student solutions were judged to be of the same complexity and on
the same mathematical level. The solutions used in the task analysis are
therefore seen to be comparable to solutions that students in year 7 would have
produced.

70
4.8.6 Task analysis

The analysis procedure of the tasks was discussed in a seminar with researchers
and research students in mathematics education. Comments given at the
seminar resulted in the analysis guide described in chapter 4.6. The analyser is
supposed to consider what a student in year 7 should need in order to solve a
task. The author has worked as a teacher in mathematics for more than 15
years, although in the upper secondary school. The author also has five years of
experience in teaching situations in mathematics with students at younger ages,
years 2―9. To test the analysis guide further, the author had a one-hour seminar
with six teachers from a lower secondary school (year 7―9). The author first
briefly introduced the analysis guide, and the teachers analysed 6 tasks in the
mathematical kangaroo working in pairs. After the analysis of the task, the
analysis procedure was summarised in a discussion between the teachers and
the author. After the analysis done by the teachers working in pairs, it was
found that the results of the analyses differed between the teachers and the
author. After the discussion, however, the author and the teachers, except for
one, agreed upon how the analyses of the competencies should be interpreted.

The discussion of the analysis tool with research colleagues, the teachers’
experience of the author, the seminar with the lower secondary school teachers,
the control of the solutions with the authentic solutions given in the material
for the national test in year 6 make the analysis process reliable and valid
enough to test the method, which aims to explore differences between the
groups that were identified. Nevertheless, it could have been even better if the
whole analysis process had been done in collaboration with a larger group of
lower secondary school teachers in mathematics, and/or more in depth with
research colleagues in mathematics education.

4.8.7 Problem solving competency

The definition of a mathematical problem that is used says that a mathematical

problem is an individual phenomenon. One student might be aware of a
solution method while another student is not. The task is therefore a problem
for one of them but not for the other. A good approximation is made by asking

71
teachers who are experienced in teaching students of this age, but this is still a
source of failure. The only way to find out whether the task is a problem for a
student is to interact with the student in some way, but this makes it difficult to
handle large groups of students.

There is also a complication that problem solving competency is not analysed in

the same way in the both tests. Problem solving competency on the national
test is judged by the constructors (Stockholm University, 2013). However, the
constructors use experienced teachers in the process of constructing the tests,
and this process involves categorising tasks according to which competency
they test.

The National Centre of Mathematics (NCM) in Sweden has categorised tasks

on the mathematical kangaroo that they judge to measure problem solving
competency. I choose not to use their classification because it is not described
whether they use experienced teachers in this process. The ways that are chosen
have one thing in common, i.e. it is the participation of teachers experienced in
teaching students for whom the test is aimed.

In the analysis, the identified groups are compared in one test at a time. It is
therefore not a dilemma that the competency is not defined in exactly the same
way.

The analysis of the problem solving competency is a weak point in this study,
but the conclusions are valid because the groups are compared in one test at a
time. There is no comparison that mixes problem solving competency between
the two tests.

4.9 Ethical considerations

The ethical perspective of the study has been discussed in a seminar with senior
researchers and research colleagues. The study has also been examined by the
ethics committee at the University without requiring additional review. There
has been no interaction between the researcher and the participants of the
study. Data connected to the national tests are public data. To be able to
include results from the mathematical kangaroo, it was necessary to obtain a
signed informed consent form from the students and their guardians.

72
All data were connected to the individuals’ personal number; the data were
anonymised and coded through a program that produced a SHA-256 hash
string. An expert in coding (Wireflow AB) implemented the coding procedure.
The security in the coding process is high.

Considering the seminar with colleagues and the advice of the ethics
committee, it is reasonable to draw the conclusion that no individual was
injured psychologically or physically by participation in this study.

73
5 Results

Chapters 5.1 to 5.4 give the results of study 1, the descriptive part comparing
relative achievement. Chapters 5.5 to 5.8 give the results of study 2, involving
the analysis with mathematical competencies.

5.1 Comparison of relative achievement on the national test in years 3

and 6, categories I and II

When comparing the change in ranking position between the national tests in
year 3 and the national test in year 6, it was required that the students had taken
both tests. There were 568 students with results on both tests. Since students
can have the same ranking position it means that the groups of the top 10% on
each test and the groups of the bottom 10% on each test do not need to
contain the same number of students.

Category I - Students highly ranked, top 10%, in both years 3 and 6

49 students were ranked among the top 10% in year 3, and 55 students were
among the top 10% in year 6; of those, 15 students, 10 girls and 5 boys, were
among the top 10% on both tests. Ranking positions for the top 10% students
in year 3 are placed to the right of the vertical solid line in Figure 19, and
ranking positions for the top 10% students in year 6 are placed above the
horizontal solid line in Figure 19. The circles inside the box in the upper right
corner in Figure 19 therefore represent students who are ranked among the top
10% on both tests.

Category II - Students ranked low, bottom 10%, in both years 3 and 6

56 students were among the bottom 10% in year 3, and 57 students were
among the bottom 10% in year 6; of those, 26 students, 10 girls and 16 boys,
were among the bottom 10% on both tests. Ranking positions for the bottom
10% students in year 3 are placed to the left of the vertical dotted line in Figure
19, and ranking positions for the bottom 10% students in year 6 are placed
below the horizontal dotted line in Figure 19. The circles inside the box in the

74
 bottom left corner in Figure 19 therefore represent students who are ranked
among the bottom 10% on both tests.
nk(Sum6), xlim=c(0,568),ylim=c(0,568), main=" ", xlab="Ranking place in Year 3", ylab="Ranking place in Year 6")

1)
1)
)
=2)

I textrutor via PP
orginal r-text
Ranking position in year 6

Ranking position in year 3

Figure 19. Ranking position on the national test in year 6 versus the ranking position on the national test
in year 3.

5.1.1 Interpretations of the results in categories I and II

The results in this study show that there are more students who stay among the
bottom 10% than those who stay among the top 10%. One explanation for the
results can be the existing ceiling effects in both tests. Considering the ceiling
effect in the two national tests, Figure 20 and Figure 21, there are more
students who achieve high (actual results) in the national test than who achieve

75
 low (actual results). The students who achieved among the top 10% in one of
the two national tests are shaded in Figure 20 and Figure 21. The ceiling effect
means that a change by only a few points can change the ranking position
remarkably among the high achievers but does not have the same affect among
the low achievers. With the results it can be concluded that it is more likely that
a student stays among the poorer achievers than it is that he or she stays among
the best achievers. In this study relative achievement is used in the comparison,
the frequency graphs based on actual results on the two tests, Figure 20 and
Figure 21, show that students among the bottom 10% have a wide distribution
of their actual results, while students belonging to the top 10% in any of the
=c(0,100))
x2,x2)]), y= c(0, y[x1:x2], 0), col="gray"),tests
main=" ",have all achieved highly in actual points on the tests.
xlab=" ",xlim=c(0,100))

a och placerad över orginal r-text

Frequency

Figure 20. Frequency distribution in the national test in year 3, top 10% shaded. Percentage results on
the x-axis

76
0))
]), y= c(0, y[x1:x2], 0), col="gray"), main=" ", xlab=" ",xlim=c(0,100))

PP som är placerad över orginal r-text

Frequency

Figure 21. Frequency distribution in the national test in year 6, top 10% shaded. Percentage results on
the x-axis

5.2 Comparison of relative achievement on the national test in years 3

and 6, categories II and IV

Category III – large increase in ranking position from year 3 to year 6

For a student to be categorised in group III the student must increase the
ranking position by at least 40 percentage points, i.e. by at least 227 ranking
positions, from year 3 to year 6, as illustrated in Figure 22. This means that
students that have a ranking position between 1 and 341 in year 3 have the
possibility to increase their ranking position by 40 percentage points and, after
the increase, they will have a ranking position between 228 and 568 in year 6.

77
 Ranking position 1 341 568
in year 3

An increase of
40 percentage points
in ranking position

Ranking position 1 228 568

in year 6

Figure 22. Illustration of possible movements of ranking position for students in category III.

The students that are ranked among the bottom 60% (below 341 in ranking
position) on the national test in year 3 are those who theoretically have the
possibility to increase their ranking position by at least 40 percentage points.
They are 333 in total: of those, 33 students, 17 girls and 16 boys, increased their
ranking position by at least 40 percentage points on the national test from year
3 to year 6, marked with filled circles in Figure 23.

78
ank(Sum6)-rank(Sum3)>227) %in% c(TRUE))
6), main=" ", xlab="Ranking place in Year 3", ylab="Ranking place in Year 6", xlim=c(0,568), ylim=c(0,568))
easeIndex],rank(Sum6)[IncreaseIndex], pch=21, bg="gray47")

ext via PP textrutor

Ranking position in year 6

Ranking position in year 3

Figure 23. Group III, students with a large increase in ranking position from year 3 to year 6.

To compare students that belong to category III, with those who had the
possibility to belong to the category, statistical measures of minimum, quartiles
and maximum values, for actual results on the two national tests were analysed.
The comparison of the statistical measures for different groups of students
show which students are most likely to perform a large increase in ranking
position. The groups of interest are:

 Group R is defined as the group of students (n=333) with ranking

positions 1 to 341 by their actual results on the national test in year 3,
that is, those who had the possibility to perform a large increase in
ranking position from year 3 to year 6, see figure 22,

 Group S is defined as the group of students (n=334) with ranking

positions 228 to 568 by their actual results on the national test in year 6,

79
 that is, the maximum range of ranking positions for students in Category
III, see figure 22,

 Category III are the students (n=33) who performed an increase of at

least 40 percentage points from year 3 to year 6, and are therefore
included in both Group R and Group S.

The statistical measures for the groups of interest based on their actual total
sum on the national test in year 3 and/or year 6 are shown in Table 12.

Table 12
Statistical measures based on the actual total sum on the national test in year 3 and/or year 6 for the students in category
III, Group R and Group S.

Statistical Group R Category III Group S Category III

measures Actual results on Actual results on Actual results on Actual results on
the national test the national test in the national test the national test in
in year 3 year 3 in year 6 year 6
Minimum 19 57 71 74
Lower quartile 71 70 77 81
Median 77 74 83 88
Upper quartile 80 79 90 91
Maximum 82 81 102 100

Comparing the distribution of actual results between category III and Group R
shows that students in category III are not among the lowest achievers on the
national test in year 3, due to the difference in minimum value in the two
groups, see Table 12. However, the distribution of actual results from lowest
quartile to the maximum are similar for category III and Group R. Comparing
category III and Group S shows that the distribution of actual results are
similar for both groups, see Table 12. These results indicate that among
students that perform a large increase from year 3 to year 6, however, low
achievers in year 3 are not represented.

Category IV - large decrease in ranking position from year 3 to year 6

For a student to be categorised in group IV, he or she must have decreased in

ranking position by at least 40 percentage points, i.e. by at least 227 ranking
positions, from year 3 to year 6. This means that students that have a ranking
position between 228 and 568 in year 3 have the possibility to decrease their
ranking position by 40 percentage points and, after the decrease, they can have
a ranking position between 1 and 341 in year 6.

80
 The students that are ranked among the top 60% on the national test in year 3
are those who theoretically have the possibility to decrease their ranking
position by at least 40 percentage points. They are 333 in total; 35 students, 15
k(Sum3)-rank(Sum6)>227) %in% c(TRUE))
main=" ", xlab="Ranking place in Year 3", ylab="Ranking place in Year 6", xlim=c(0,568), ylim=c(0,568))
seIndex],rank(Sum6)[DecreaseIndex],girls
pch=21,and 20 boys, decreased their ranking position by at least 40 percentage
bg="gray47")

trutor över orginal r-text

points in the national test from year 3 to year 6, marked with filled circles in
Figure 24.
Ranking position in year 6

Ranking position in year 3

Figure 24. Group IV, students with a large decrease in ranking position from year 3 to year 6.

To compare students that belong to category IV, with those who had the
possibility to belong to the category, statistical measures: minimum, quartiles
and maximum values, for actual results on the two national tests were analysed.
Comparing the statistical measures for different groups of students, shows
which students who are most likely to make a large decrease in ranking
position. The groups of interest are:

81
  Group X is defined as the group of students (n=333) with ranking
positions 228 to 568 by their actual results on the national test in year 3,
that is those who had the possibility to make a large decrease in ranking
position from year 3 to year 6,

 Group Y is defined as the group of students (n=333) with ranking

positions 1 to 341 by their actual results on the national test in year 6,
that is the maximum range of ranking positions for students in Category
IV,

 Category IV are the students (n=35) who made a decrease of at least 40

percentage points from year 3 to year 6, and are therefore included in
both Group X and Group Y.

The statistical measures for the groups of interest based on their actual total
sum on the national test in year 3 and/or year 6 are shown in Table 13.

Table 13
Statistical measures based on the actual total sum on the national test in year 3 and/or year 6 for the students in category
IV, Group X and Group Y.

Statistical Group X Category IV Group Y Category IV

measures Actual results on Actual results on Actual results on Actual results on
the national test the national test in the national test the national test in
in year 3 year 3 in year 6 year 6
Minimum 80 81 9 21
Lower quartile 82 85 54 54.5
Median 85 87 64 64
Upper quartile 88 89 72 72
Maximum 93 93 78 74

Comparing category IV and Group X shows that the distribution of actual

results are similar for both groups, see Table 13. Comparing the distribution of
actual results between category IV and Group Y shows that students in
category IV are not among the lowest achievers on the national test in year 6,
due to the difference in minimum value in the two groups, see Table 13.
However, the distribution of actual results from lowest quartile to the
maximum are similar for category IV and Group Y. These results indicate that
students that make a large decrease in ranking position from year 3 to year 6 are
not placed among the lowest achievers in year 6.

82
5.2.1 Interpretations of the results in categories III and IV

There are an equal number of students who decrease in relative achievement as

students who increase in relative achievement. The percentage of students who
did those large movements is 10% both for those that performed an increase in
achievement and for those who made a decrease in achievement.

The results of the analysis of category III and IV show that low achievers in any
of the tests did not belong to the groups that performed large movements in
relative achievement. Therefore it can be deduced that, as low achievers in the
national test in year 3 did not improve their results remarkable, and also
students belonging to category IV did not end up among the lowest achievers
in the national test in year 6, the results verify that low achievers continue to be
low achievers.

5.3 Comparison of the national test in year 6 with the mathematical

kangaroo in year 7

Students involved in the comparison of relative achievement on the national

test in year 6 and on the mathematical kangaroo in year 7 are those with results
in the national test in year 6 and in the mathematical kangaroo in year 7, 264
students, 109 boys and 155 girls. These 264 students are the sample discussed
in chapter 4.2. The results of the analysis are to answer research question 2 and
3.

The analysis aims to investigate the number of, and the distribution, of those
students who ranked among the top 20%, top 10% and top 5% in one of the
tests, but among the bottom 80% in the other test. The students who are
among the top 20% in both tests (n=23) are not investigated because one aim
of the study is to explore movements in relative achievement, and another aim
is to explain why students are high achievers in one test but not in another. To
fulfil those aims, comparable groups must be distinct and must not overlap. In
addition, to explore differences or movements in achievement, students in an
identified group must have made some sort of change in achievement between

83
the measuring points. Those who are among the top 20% in both tests can only
have made minor movements in achievement. It is not the aim in this study to
explore those students.

Two non-overlapping groups of students are identified for the comparison

between the national test in year 6 and the mathematical kangaroo in year 7.
Students are either identified to be among the top 20% achievers in the
mathematical kangaroo and among the bottom 80% achievers in the national
test in year 6, those belongs to what will be called Group 1. The alternative is
that students are identified to be among the top 20% achievers in the national
test in year 6 and among the bottom 80% achievers in the mathematical
kangaroo, those belongs to what will be called Group 2. Group 1 and 2 will be
further investigated in study 2.

Table 14 and Table 15 show detailed information of the identified students.

The tables show their actual results in the national test in year 6 (Sum6) and in
the mathematical kangaroo in year 7 (SumK). The tables also show each student
ranking position in each test (Rank6, RankK) and the change in ranking
position between the tests (Change in ranking). In the study ranking positions
are used, the tables make it possible to compare the ranking positions with
actual results for the identified students. In Table 14, students in Group 1 are
shown and in Table 15, students in Group 2 are shown.

84
Table 14
Students ranked top 5% (white), top 10% (light shaded) and top 20% (dark shaded) in the mathematical kangaroo and
the bottom 80% in the national test in year 6, 13 boys and 12 girls.

ID Gender Sum6 SumK Rank6 RankK Change in ranking

S51 female 84 79 169 264 95

S207 male 82 70 152.5 258 105.5
S146 male 90 64 207.5 250.5 43
S197 female 88 64 194 250.5 56.5
S87 female 78 62 122.5 248.5 126
S36 male 89 59 201 241 40
S56 female 81 58 143.5 238 94.5
S84 male 90 57 207.5 233.5 26
S245 male 73 57 85 233.5 148.5
S192 male 51 57 18.5 233.5 215
S230 female 76 56 108.5 229 120.5
S262 male 76 56 108.5 229 120.5
S72 female 84 55 169 225 56
S44 female 77 55 114.5 225 110.5
S231 male 68 55 64 225 161
S106 male 89 54 201 219.5 18.5
S108 female 89 54 201 219.5 18.5
S257 male 88 54 194 219.5 25.5
S2 male 85 54 175.5 219.5 44
S152 female 84 54 169 219.5 50.5
S143 female 88 53 194 213 19
S252 female 86 53 181.5 213 31.5
S255 male 86 53 181.5 213 31.5
S150 female 82 53 152.5 213 60.5
S19 male 81 53 143.5 213 69.5

85
Table 15
Students ranked top 5% (white), top 10% (light shaded) and top 20% (dark shaded) in the national test in year 6 and
bottom 80% in the mathematical kangaroo, 7 boys and 16 girls.

ID Gender Sum6 SumK Rank6 RankK Change in ranking

S110 female 100 51 261 197 -64

S172 female 99 38 258.5 118 -140.5
S205 male 99 51 258.5 197 -61.5
S201 female 98 41 256.5 140.5 -116
S94 female 96 42 246 149.5 -96.5
S135 female 96 41 246 140.5 -105.5
S157 female 96 31 246 59 -187
S98 female 95 52 241 205.5 -35.5
S145 male 94 39 235.5 126.5 -109
S162 male 94 51 235.5 197 -38.5
S194 female 94 52 235.5 205.5 -30
S198 female 94 46 235.5 174.5 -61
S99 female 93 44 229.5 161.5 -68
S112 female 93 45 229.5 167.5 -62
S82 female 92 47 223.5 180.5 -43
S121 female 92 52 223.5 205.5 -18
S147 male 92 37 223.5 108.5 -115
S222 female 92 52 223.5 205.5 -18
S7 female 91 29 216 50 -166
S216 male 91 48 216 185 -31
S227 female 91 42 216 149.5 -66.5
S236 male 91 37 216 108.5 -107.5
S240 male 91 42 216 149.5 -66.5

86
Figure 25 visualises the students identified as high achievers on one test but not
on the other and shows their distribution in the two tests used for comparison.
In Figure 25 a circle placed at the bottom of the diagram represents a student
that was ranked low on the mathematical kangaroo, and a circle placed at the
top of the diagram represents a student that is highly ranked on the
mathematical kangaroo. In the same way, a circle placed to the left in Figure 25
represents a student that was ranked low on the national test in year 6, and a
circle to the right represents a student that was ranked high on the national test
in year 6. In Figure 25, students belonging to Group 2 are marked with filled
circles on the right side of the diagram, and students belonging to Group 1 are
marked with filled circles at the top of the diagram. Students who are among
the top 20% on both the national test in year 6 and the mathematical kangaroo
in year 7 are represented by non-filled circles in the upper right corner: those
students are not the object of study here and are therefore the circles are not
filled.
Ranking position in year 7 (kangaroo)

Ranking position in year 6 (national test)

Figure 25. Distribution of students identified to be high achievers in one test but not in the other.

87
5.3.1 Interpretations of the comparison between the national test in year
6 and the mathematical kangaroo in year 7

Statistical measures used: minimum, lower quartile, median, upper quartile and
maximum, see Table 16, for the change in ranking positions in Group 1 and 2
show that students in Group 1 are more widely spread (have a larger variance)
in the change of their ranking position as compared to students in Group 2.

For students belonging to the upper half (median to maximum), the

distribution of students was broader for Group 1; this is visualised in Figure 25.
There is a lack of students in the sample that achieved low in the national test.
It is possible that some of those would also have achieved highly on the
mathematical kangaroo and thereby be included in the group of students that
change their ranking position more than 56.5 places (the median value) between
the two tests.

Table 16
Statistical measures of ranking position for the two groups compared.

Group 1, change in ranking Group 2, change in ranking

positions. positions.
Minimum 18.5 18
Lower quartile 31.5 40.75
Median 56.5 66.5
Upper quartile 110.5 108.25
Maximum 215 187

There is a difference in gender distribution between the groups summarised in

Table 14 and Table 15. 13 boys and 12 girls belonged to Group 1. 7 boys and
16 girls belonged to Group 2. This study does not deal with gender differences,
but it is still worth noting that in Group 1 the gender distribution is equal, but
in Group 2 the majority of the students are girls.

88
5.4 Summary of results – Study 1

The results of the descriptive analysis of relative achievement in three tests in

mathematics shows that the group of students that is among the bottom 10% in
both national tests is larger (n=26) than the group of students that is among the
top 10% on both national tests (n=15). The analysis of the national tests shows
that both tests have a ceiling effect, Figure 10 and Figure 11; this is a possible
explanation for why the top 10% group is smaller than the bottom 10% group.
There are many students that achieve highly (actual results) on both tests, which
leads to a difference of a few points having a large effect on the ranking
position. If both tests had a floor effect the opposite result would have been
expected. Both tests are mainly aimed to test the passing level, which they do,
because they discriminate at the bottom.

The groups of students who either increased or decreased more than 40

percentage points in ranking position were equal in number.

In comparing the national tests given in year 3 and year 6, it can be concluded
that the analysis indicates that there are more top students that change their
relative achievement than there are bottom students, the ceiling effect of the
tests are a plausible explanation. There are approximately the same numbers of
students that increase as decrease largely in relative achievement.

Students in Group 1 change in ranking positions between the national test in

year 6 to the mathematical kangaroo in year 7 in the range of 18.5 to 215, with a
median value of 56.5. The comparative group, Group 2 change in ranking
position in a range from 18 to 187, with a median value of 66.5. Students who
ranked high in one of the tests therefore show a wider range of achievement in
the other test, and there is not a large difference between the two groups.

5.5 Competency profiles in the tests

The results correlated to study 2 are to answer research question 4 and are
based on the comparison of two groups, identified in study 1, of how those
groups differ in the way they activate mathematical competencies through their
result on the mathematical kangaroo. A student is seen to have activated a

89
mathematical competency if she or he has been given points on a task that
according to the analyse guide, Table 10, requires the competency to solve the
task. The identified groups are;

Group 1 are those students identified to be among the top 20% achievers in the
mathematical kangaroo and among the bottom 80% achievers in the national
test in year 6.

Group 2 are those students identified to be among the top 20% achievers in the
national test in year 6 and among the bottom 80% achievers in the
mathematical kangaroo.

5.5.1 The mathematical kangaroo

The analysis of each task in the mathematical kangaroo resulted in the reduced
matrixes shown in Table 17. Abbreviations for the competencies are used as
follows: App = applying procedures competency, Rea = reasoning competency,
Com = communication competency, Rep = representation competency, Con =
connection competency, Pro = problem solving competency.

90
Table 17
Results of the task analysis in the mathematical kangaroo.

Task 1-10/
Competency 1 2 3 4 5 6 7 8 9 10
App 1 0 0 0 1 1 1 1 0 0
Rea 0 0 1 0 0 0 1 0 1 1
Com 1 0 1 1 1 1 1 0 0 0
Rep 0 1 1 0 1 0 0 1 1 1
Con 0 0 0 0 0 0 0 1 0 1
Pro 0 0 1 1 0 0 1 0 1 1

Task 11-21/
Competency 11 12 13 14 15 16 17 18 19 20 21
App 1 1 1 0 0 1 1 0 0 0 0
Rea 0 1 0 1 1 0 1 1 1 1 1
Com 0 0 1 1 0 1 1 1 1 1 1
Rep 1 1 1 1 1 0 0 0 0 0 0
Con 1 1 1 1 1 0 0 0 0 0 0
Pro 0 1 0 1 1 1 0 1 1 1 1

Pairwise comparison of the competencies shows that the reasoning competency

is always activated together with the problem solving competency except on
one occasion, task 17, which demands the reasoning competency but not the
problem solving competency. It is possible to show the problem solving
competency in two tasks, 4 and 16, without activating the reasoning
competency. The pairwise comparison also shows that it is not possible to
activate the connection competency without also activating the representation
competency. It is not a purpose of this study to explore the reason why those
competencies are related to each other. In the interpretation of the results, it is
important to remember the relation between the reasoning competency and the
problem solving competency, as well as the relation between the connection
competency and the representation competency.

Summary data from the task analysis results in a competency profile of the
whole test as shown in Figure 26. The numbers in Figure 26 show how much
each competency is given opportunity to be activated in relation to the total
amount of possible competencies on each test.

91
 Competency profiles

Pro;
App; 0,15 0,09
Con;
Pro; 0,19
0,07
App; 0,32

Con; 0,10 Rea; 0,18 Rep; 0,17

Rep; 0,16 Rea; 0,11

Com; 0,21 Com; 0,25

Mathematical kangaroo in year 7 National test in year 6

Figure 26. Competency profile for the mathematical kangaroo and the national test in year 6.

The competency profile for the mathematical kangaroo shows that the test
gives the greatest opportunity to activate communication competency (21%)
and the lowest opportunity to activate the connection competency (10%). The
relation between the connection competency and the representation
competency shows that taking away one or two tasks that activate the
connection competency also lowers the representation competency. The
relation between the reasoning competency and the problem solving
competency has a similar effect: taking away tasks that activate the reasoning
competency lowers the problem solving competency.

92
5.5.2 The national test in year 6

The national test in year 6 (2012) consists of four parts, A―D. Part A is
excluded from the analysis of the competencies. Part A is an oral test that gives
1 or 0 points out of 106 in total. As part A is an oral part and because of its low
effect on the results it is not addressed here. The analysis of parts B―D in the
national test was made in the same way as the analysis of the mathematical
kangaroo. Since the test is classified, results of individual tasks are not shown;
the competency profile of the complete test is shown in Figure 26.

5.5.3 Comments on the competency profiles of the tests

The two tests are similar in their competency profiles as regards the
communication, representation and connection competencies. The national test
gives greater opportunity to activate the applying procedures competency in
comparison to the mathematical kangaroo. The mathematical kangaroo gives
more opportunity for activating the reasoning and the problem solving
competency compared to the national test.

The different ways of analysing problem solving competency can be one

explanation for the difference in the problem solving competency. The
mathematical kangaroo was analysed by eight experienced year 7 teachers, and
the national test is constructed by a group at Stockholm University the “PRIM-
gruppen” (Stockholm University, 2013). However the “PRIM-gruppen” also
uses experienced teachers as co-constructers and discussion partners
(Stockholm University, 2013). There is therefore at least one common factor in
the analysis process.

5.6 Mathematical competencies activated by Group 1 and Group 2

Each competency is given a relative activation factor, calculated as the sum of

all times that the competency was activated divided by the total sum of

93
activated competencies. Thus the sum of all competencies relative activation
factors is 1.

Students who achieved among the top 20% in ranking position in one of the
tests, the national test in year 6 (2012) or the mathematical kangaroo in year 7
(2013), and at the same time are among the bottom 80% in the other test are
treated as two comparative groups in the following analysis.

In the analysis, one group is called Group 1 that is; “high achievers in the
mathematical kangaroo (bottom 80% in the national test)” and the other group
are called Group 2 that is, “high achievers in the national test (bottom 80% in
the mathematical kangaroo)”.

5.6.1 Favoured competency

To investigate whether one competency is in favour of being activated in

relation to the others, in the case of the mathematical kangaroo, a comparison
is made of how students activate that competency in relation to the possible
outcome of that competency. That a student activate a competency means that
the student have been given points on tasks that require that the specific
competency is activated according to the analyse guide used.

An example is given in Table 18 and Table 19 of how a competency analysis is

done in order to create competency profiles. Table 18 gives the competency
profile for the test, and Table 19 gives the competency profile for a specific
student. In Table 18, “Sum Comp” is the number of times a specific
competency is given the opportunity to be activated through the test; in Table
19, “Sum Comp” is the number of times the specific competency is activated
by the student, S71. In both tables, “Rel Comp” tells the activation of each
competency relative to the total number of competencies activated.

94
 Table 18
Result of competency analysis in the case of full points.

Competency Sum Comp Rel Comp

App 10 0,15
Rea 12 0,18
Com 14 0,21
Rep 11 0,16
Con 7 0,10
Pro 13 0,19
∑ 67 1

Table 19
Result of competency analysis of student S71.

S71 Sum Comp Rel Comp

Competency
App 10 0.17
Rea 9 0.16
Com 14 0.24
Rep 9 0.16
Con 6 0.10
Pro 10 0.17
∑ 63 1

To compare how groups of students activate the competencies in a test without

actually comparing their test results, a calculation of the relative deviation from
the competency profile of the students from the competency profile of the test
is introduced.

The relative deviation, denoted 𝑓𝑐𝑝 , is calculated for each student and each
competency using the formula:

𝑟𝑒𝑙. 𝑐𝑜𝑚𝑝𝑠 − 𝑟𝑒𝑙. 𝑐𝑜𝑚𝑝𝑡

𝑓𝑐𝑝 =
𝑟𝑒𝑙. 𝑐𝑜𝑚𝑝𝑡

Where:

95
𝑟𝑒𝑙. 𝑐𝑜𝑚𝑝𝑠 is the number of times a student activate a competency relative to
the total number of competencies activated by a student. In the example, it is
values from column 3 in Table 19.

𝑟𝑒𝑙. 𝑐𝑜𝑚𝑝𝑡 is the opportunities the test gives to activate a competency relative
to the total number of competencies the test gives opportunity to. In the
example, it is values from column 3 in Table 18.

The values of the relative deviation, 𝑓𝑐𝑝 , for students belonging to the same
identification group, Group 1 or Group 2, are collected and thereafter
differences in how groups of students activate each competency can be
investigated. In the comparison between the groups each 𝑓𝑐𝑝 value for each
student is represented by a marker in a diagram, one diagram for each group
and each competency, see Figure 27 and Figure 28. If a student activates a
competency to the same proportion as in the case of getting full points in the
mathematical kangaroo, the marker will be placed on the zero line. If the
competency proportionally is activated more for a student than in the case of
getting full points, the marker is placed above 0; another competency or other
competencies will then have to “pay” and will be placed below 0. A point on 0
does not mean that a student has activated the competency as much as possible
throughout the test, since it is a proportional measurement.

In the example given in Table 18 and Table 19, the applying procedures
competency and the communication competency are activated more relative to
the other competencies by the student than they are by getting full points in the
test. In the same way the reasoning competency and the problem solving
competency are less activated and the representation competency and
connection competency are activated to the same extent as in the test profile.

This analysis shows whether students in their competency profile activate a

competency more or less than the test profile of the mathematical kangaroo
shows. Therefore the analysis shows if groups of students favour or disfavour
any competencies in relation to each other. It gives a possibility to explore
individuals and patterns of how competencies are activated in the two groups.

Results of this analysis of the students identified are shown for Group 1 to the
left and for Group 2 to the right in Figure 27 and Figure 28.

96
 Group 1 Group 2

The 𝑓𝑐𝑝 value for each student for The 𝑓𝑐𝑝 value for each student for
the competencies: App, Rea, Com the competencies: App, Rea, Com

Applying procedures Applying procedures

0,5 0,5
0,4 0,4
0,3 0,3
0,2 0,2
0,1 0,1
0 0
-0,1 0 5 10 15 20 25 -0,1 0 10 20
-0,2 -0,2
-0,3 -0,3
-0,4 -0,4
-0,5 -0,5
-0,6 -0,6

Reasoning Reasoning
0,5 0,5
0,4 0,4
0,3 0,3
0,2 0,2
0,1 0,1
0 0
-0,1 0 5 10 15 20 25 -0,1 0 10 20
-0,2 -0,2
-0,3 -0,3
-0,4 -0,4
-0,5 -0,5
-0,6 -0,6

Communication Communication
0,5 0,5
0,4 0,4
0,3 0,3
0,2 0,2
0,1 0,1
0 0
-0,1 0 5 10 15 20 25 -0,1 0 10 20
-0,2 -0,2
-0,3 -0,3
-0,4 -0,4
-0,5 -0,5
-0,6 -0,6

Figure 27. Relative activation of the competencies App, Rea and Com in comparison with the
competency profile in the mathematical kangaroo. The x-axis represents students.

97
 Group 1 Group 2

The 𝑓𝑐𝑝 value for each student for The 𝑓𝑐𝑝 value for each student for
the competencies: Rep, Con, Pro the competencies: Rep, Con, Pro

Representation Representation
0,5 0,5
0,4 0,4
0,3 0,3
0,2 0,2
0,1 0,1
0 0
-0,1 0 5 10 15 20 25 -0,1 0 10 20
-0,2 -0,2
-0,3 -0,3
-0,4 -0,4
-0,5 -0,5
-0,6 -0,6

Connection Connection
0,5 0,5
0,4 0,4
0,3 0,3
0,2 0,2
0,1 0,1
0 0
-0,1 0 5 10 15 20 25 -0,1 0 10 20
-0,2 -0,2
-0,3 -0,3
-0,4 -0,4
-0,5 -0,5
-0,6 -0,6

Problem solving Problem solving

0,5 0,5
0,4 0,4
0,3 0,3
0,2 0,2
0,1 0,1
0 0
-0,1 0 5 10 15 20 25 -0,1 0 10 20
-0,2 -0,2
-0,3 -0,3
-0,4 -0,4
-0,5 -0,5
-0,6 -0,6

Figure 28. Relative activation of the competencies Rep, Con and Pro in comparison with the
competency profile in the mathematical kangaroo. The x-axis represents students.

Figure 27 and Figure 28 visualises the comparison between Group 1 and Group
2 in how they activate the competencies on the mathematical kangaroo.
Through the figures it can be suspected that Group 2 activates the applying
procedures competency, the representation competency and the connection
competency to a greater extent than Group 1, and that Group 1 activate the

98
reasoning competency and the problem solving competency to a greater extent
than Group 2. To verify or invalidate the suspicion a Fisher exact test is
performed.

5.6.1.1 Fisher exact test in favoured competency analysis

In the favoured competency analysis, competencies activated by individuals

shown in their individual kangaroo profiles are compared to the competencies
in the competency profile of the test. If a competency is shown more it
automatically means that another competency is shown less, because they stand
in relation to one another. Because the competency profiles are not related to
achievement, the two groups of students identified can be compared to each
other.

The two groups, Group 1 and Group 2, are compared to investigate whether
any competency is favoured or disfavoured in the mathematical kangaroo. The
Fisher exact test is used to investigate whether one group of students
significantly favours or disfavours a competency compared to the other group.
The Fisher exact test is a dichotomous test that calculates whether there is a
significant difference between two groups around a chosen limit, where data are
either above or below the chosen limit. The natural choice of limit is zero in
this case, since zero means that the competency is activated to the same
proportion to which the test gives an opportunity. However, if there is reason
to suspect differences in distribution between the groups although they do not
show a significant difference around the zero, a new limit can be chosen.

99
Example:

Group 1 Group 2

The 𝑓𝑐𝑝 value for each student for The 𝑓𝑐𝑝 value for each student for
the competency: Pro the competency: Pro

Problem solving Problem solving

0,5 0,5
0,4 0,4
0,3 0,3
0,2 0,2
0,1 0,1
0 0
-0,1 0 5 10 15 20 25 -0,1 0 10 20
-0,2 -0,2
-0,3 -0,3
-0,4 -0,4
-0,5 -0,5
-0,6 -0,6

Figure 29. Example of when it can be justified to change the limit of comparison for the Fisher exact
test.

It appears in Figure 29 that the 𝑓𝑐𝑝 values for Group 2 are distributed to a
greater extent on the negative side. A comparison of whether there is a
significant difference between the two groups around the zero shows that there
is not, the hypotheses that Group 1 activates the competency either less or
more than Group 2 is not supported (p=0.9678 alternatively p=0.1509).
However, when the limit of comparison is changed to -0.1, we see a significant
difference, the hypothesis that Group 1 activates the competency more than
Group 2 is supported (p=0.0005), indicating that more students in Group 1 are
placed above -0.1 compared to Group 2. This means that there are more
students in Group 2 that activate the problem solving competency to a lower
degree relative to the other competencies as compared to Group 1. If actual
achievement is taken into account it is expected that the students that succeed
among the best in the mathematical kangaroo also activate the problem solving
competency most, since the test offers many opportunities to activate the
problem solving competency. Here, the achievement factor is eliminated and
the two groups of students are compared to each other with regard to how they
activate the competencies relative to each other. It can be concluded that:
Group 1 favours the problem solving competency over Group 2 in the
mathematical kangaroo despite the achievement factor having been eliminated.

The results of the Fisher exact test are shown in Table 20. In column 2 the p-
value tells if the hypothesis that Group 1 activates the competency less than

100
Group 2 is significant. In column 3 the p-value tells if the hypothesis that
Group 1 activates the competency more than Group 2 is significant. In those
cases where a difference in distribution between the groups is suspected when
looking at Figure 27 and Figure 28 the Fisher exact test is done one more time
with a new limit to test around, 0.1 or -0.1. If the new Fisher exact test gives a
significant difference between the groups it can be concluded that one of the
groups activates that competency more or less than the other group.

Table 20
Summary of Fisher exact test.

Competency Group 1 activates Group 1 activates

the competency the competency
less than Group 2, more than Group 2,
p-value p-value

App 0.6971 0.5997

App(0.1 limit) 0.3611 0.8282

Rea 0.734 0.7757

Rea (-0.1 limit) 0.9736 0.1112

Com 0.7267 0.4905

Rep 0.114 0.9856

Rep (0.1 limit) 0.01342 0.9985

Con 0.0365 0.9935

Pro 0.9678 0.1509

Pro (-0.1 limit) 1 0.0004694

5.6.1.2 Comparison between Group 1 and Group 2

The connection competency is activated significantly less among Group 1. No

other competency shows any significant difference between the two groups
when comparing around the zero. Visually, in Figure 27 and Figure 28, it looks
as there are differences in the distribution between the groups in the applying
procedures, the reasoning, the representation and the problem solving

101
competency. By raising or lowering the limit in the Fisher exact test when
testing those competencies, it is shown that the representation competency is
activated less in Group 1 compared to Group 2. This result is expected since
the connection competency and the representation competency are related to
each other.

Lowering the limit in the Fisher exact test also indicates that the problem
solving competency is activated more by Group 1 compared to Group 2. The
Fisher exact test with a lower limit on the reasoning competency, which is
related to the problem solving competency, does not give a significant
difference.

5.6.1.3 Comparison of favoured competencies within Group 1 respective Group 2

In the mathematical kangaroo both groups activate the applying procedures-

and the representation competency to a higher degree than other competencies,
which means that both groups have to “pay” with other competencies.

Both groups “pay” with the reasoning competency and partly with the problem
solving competency, meaning that they activate those competencies to a lower
degree than other competencies. However the problem solving competency is
more evenly distributed and closer to zero for Group 1.

The communication competency is evenly distributed around the zero for both
groups, meaning that both groups activate that competency in the same
proportion that the test gives the opportunity to do.

The connection competency is evenly distributed around the zero for Group 1,
but activated to a higher degree for Group 2.

The results are summarised in Table 21, where a “+” indicates that the
competency is activated more in comparison to the other, a “–“ indicates that
the competency is less activated and a “0” indicates that the competency is
activated to the same degree as expected from the test competency profile.

102
Table 21
Summary of favoured competency on the mathematical kangaroo.

Competency Group 1 Group 2

App + +

Rea - -

Com 0 0

Rep + +

Con 0 +

Pro - -

5.6.1.4 Interpretations

The analysis of a favoured competency in the mathematical kangaroo shows

that Group 1 activate the applying procedures competency and the
representation competency more, relative to the other competencies and that
they pay with the reasoning competency and the problem solving competency.
As a group, they activate the communication competency and the connection
competency in the same amount that the test offers. Looking at Group 2, the
pattern is the same except for the connection competency, which is activated
more in this group, which in turn means that this group must pay to a larger
extent with the reasoning competency and the problem solving competency.

The comparison between the two groups shows that Group 1 activate the
problem solving competency more than Group 2. It verifies that Group 2 pay
more with the problem solving competency. The comparison between the two
groups also shows that Group 1 activate the connection competency and the
representation competency less than Group 2. This agrees with that Group 2
activating the connection competency more in relation to other competencies,
and that the connection competency and the representation competency are
closely related.

103
5.7 Comparison of Group 1 and Group 2 on task level

A comparison of the response rate in percentage for the two groups in the 21
tasks on the mathematical kangaroo is summarised in Table 22. There are 25
students in Group 1, and 23 students in Group 2. The students in Group 1
naturally achieve better on most tasks because of the identification process.
Nevertheless, it is interesting to explore whether there are some tasks that show
different behaviour than others when comparing Group 1 and 2. To identify
tasks that stand out, the difference in response rate between the groups is
therefore calculated and is shown in percentage units in Table 22.

Table 22
Comparison of response rate in each task.

Group1 Group 2 %-units

Task (n=25) (n=23) between
response rate (%) response rate (%) group 1 and group 2
1 100 96 4
2 96 83 13
3 88 83 5
4 88 83 5
5 96 100 -4
6 72 61 11
7 68 30 38
8 96 83 13
9 68 30 38
10 80 83 -3
11 88 91 -3
12 76 65 11
13 76 57 19
14 44 43 1
15 40 17 23
16 72 26 46
17 48 22 26
18 32 17 15
19 96 78 18
20 36 22 14
21 40 13 27

The comparison shows that there are some tasks for which the response rate
for group 2 is much lower than for Group 1. Those tasks are number 7, 9, 15,

104
16, 17 and 21, all with a larger difference in percentage units than 20. The tasks
and the competencies they give opportunity to activate are shown in Table 23.

Table 23
Competencies activated by task 7,9, 15, 16, 17 and 21

Task 7 9 15 16 17 21
/Competency
App 1 0 0 1 1 0
Rea 1 1 1 0 1 1
Com 1 0 0 1 1 1
Rep 0 1 1 0 0 0
Con 0 0 1 0 0 0
Pro 1 1 1 1 0 1

For those six identified tasks:

 All but one, (no. 17) gave opportunity to activate the problem solving
competency,

 All but one, (no. 16) gave opportunity to activate the reasoning
competency,

 No one but one, (no. 15) gave opportunity to activate the connection
competency.

5.7.1 Interpretations

The six tasks in Table 23 that differ most between the two compared groups, is
probably the reason to that Group 1 activate the problem solving competency
more than Group 2. A result in this study is that the six tasks are most likely
special in some way. One explanation can be that they give students the
opportunity to activate the problem solving competency and the reasoning
competency, combined with that they do not give opportunity to show the
connection competency. However, this could be explained by something other
than mathematical competencies offered by the task, although, that question
remains for another study to explore.

105
The problem solving competency and the reasoning competency are closely
related, as discussed in chapter 3.3.2. The close relationship explains the results
in the pairwise comparison of the competencies made in section 5.5.1, which in
turn explains that most of the tasks identified in Table 23 demand both the
problem solving competency and the reasoning competency. Reasoning is seen
to be the “juridical counterpart” to problem solving (Lithner et al., 2010), and
this should make it almost impossible to activate the problem solving
competency without activating the reasoning competency. Task number 16 is
one of the two tasks on the mathematical kangaroo in which this is possible
according to the analysis used in this study, although, following the analysis
guide, Table 10, no reasoning competency is considered to be needed in task
number 16.

5.8 Summary of results – Study 2

Two groups of students were identified in study 1 as being high achievers, top
20%, in one of the tests, the mathematical kangaroo, or the national test in year
6, but among the bottom 80% in the other test, Group 1 respective Group 2.

How those two groups activated mathematical competencies in the

mathematical kangaroo was investigated and compared to explore the
differences in achievement. The method used is based on an existing
competency framework, MCRF (Lithner et al., 2010), discussed and developed
both with in-service teachers and research colleagues. One of the aims of this
study was to investigate the method, which has been done relatively thoroughly.
When using the method, it can be said that the group of students in Group 1
activates the problem solving competency more in the mathematical kangaroo
than students in Group 2. Students in Group 2 activate the connection
competency and the representation competency more in the mathematical
kangaroo compared to students in Group 1.

The competency profiles of the mathematical kangaroo and of the national test
in year 6, Figure 26, show that the mathematical kangaroo gives greater
opportunity to activate the problem solving competency, although, it must be
remembered that the problem solving competency is not analysed in the same
way for the two tests. The reasoning competency that is closely related to the

106
problem solving competency is also given more opportunity for activation in
the mathematical kangaroo then in the national test. The connection
competency and the representation competency are given similar opportunity
for activation in the two tests.

There were six tasks in the mathematical kangaroo out of a total 21 where there
was a large discrepancy in the response rate between the two groups. Those
tasks have in common that, to come to a solution, the solver needs to activate
both the problem solving competency and the reasoning competency or one of
those competencies. Both are needed for four of the tasks and one of the
competencies is needed for two of the tasks.

The investigated method can compare how two different groups of students
activate mathematical competencies through a test. The method can also give a
competency profile of a test that says which competencies the test gives an
opportunity to activate, and how those competencies are given opportunity
relative to each other. The method can be used to analyse individual students
and indicate how each individual activates the competencies relative to other
students in a test, showing both the strength and the weakness for that
individual. Further, the method gives the opportunity to explore tasks of special
interest according to which mathematical competencies the solver needs to
activate.

107
6 Conclusions and discussion

This chapter summarises the conclusions of the study in point form. Thereafter
follow sections that discuss the study in relation to the ceiling effect of the
national tests, movements in achievement, assessment and the problem solving
competency. Then comes a section that discusses the method chosen and some
alternatives that could have been used. The chapter ends with a section of how
this study can influence both practice and research, and suggestions for further
research.

6.1 Conclusions

 The national tests given in year 3, 2009, and in year 6, 2012, both had a
ceiling effect, which means, when measuring over time, that it is more
likely for a student to remain among the bottom achievers than to
remain among the top achievers.

 The national tests given in year 3, 2009, and in year 6, 2012, did not
guide teachers to discover students that might have needed greater
challenge according to their high achievement, because the tests did not
discriminate at the top.

 There are students who do not achieve among the top in the national
test, some of whom are ranked very low, that, when given a non-
curriculum test, achieve among the top students. This group is not
negligible in size: in this study the group consists of 9% of the
population (25 of 264).

 The problem solving competency is of special interest in the explanation

of why some students succeed in a non-curriculum bounded test but not
in a curriculum bounded one.

 The method explored that compares tests and groups of students

according to activation of mathematical competencies can be used to:

108
 o Analyse mathematical tests to show what opportunity a test gives
students to activate mathematical competencies on a task level
and to give a summary of the test in a competency test profile.

o On the individual level, show how a student activates

mathematical competencies, which can be used to identify a
student’s weak competencies and strong competencies.

o Compare two groups of students to investigate how they differ in

activating the mathematical competencies relative to each other
in a test, without taking the achievement factor into account.

o Identify tasks that differ more than others in response rate

between the compared groups and to discover what
mathematical competencies these tasks give the opportunity to
activate.

6.2 Discussion

The main aim of this study was to investigate students that have good
mathematical competencies although they are not able to show them in
conventional school mathematics. A further aim was to describe the
phenomenon in a quantitative approach and to investigate a method that can
explain differences in achievement between groups according to mathematical
competencies.

6.2.1 Ceiling effect and tests

None of the national tests used in this study discriminates at the top which
makes it difficult or perhaps impossible to use those tests to identify students
who might need greater mathematical challenge in school mathematics to be
able to develop further. The aim of the two national tests involved here are
mainly to test the passing level (Skolverket, 2010; 2012). Although, the national
tests in mathematics should show the mathematical qualities a student has in

109
the subject (Skolverket, 2014) and, according to Swedish law, students have the
right to be challenged further when they meet the requirements of the curricula
(SFS 2010:800).

Teachers therefore need some sort of assessment system that also discriminates
at the top to be able to identify those students that might need greater
challenge. The national test influences the assessment of the students (Korp,
2006) and it is therefore important that it also discriminates at the top.
Assessment in education ought to be assessment for learning, to stimulate
students’ learning (A. Pettersson, 2007; Wiliam, 2007).

6.2.2 Movements in relative achievement

With the empirical data and definitions used in this study, it has been shown
that it is more certain for a student to stay among the bottom achievers as
compared to staying among the top achievers in the national tests. This
conclusion is a contradiction to both Gagné (2005) and Pettersson (2010), who
both in separate research fields concluded that most students who have once
been high achievers continue to be high achievers. Pettersson (2010) draws the
same conclusion for low achievers. On the other hand, Häggblom (2000) shows
that it is very difficult to predict achievement several years in advance according
to how a student achieves in the present on the basis of large movement in
achievement, both upwards and downwards. None of us make measurements
in the same way, which is most likely the main reason for the different
conclusions.

In the present study, there is a ceiling effect in the tests that can explain why
there are fewer students who continue to be top achievers compared to those
who continue to be bottom achievers. Most of the tests included in the work of
Häggblom (2000) do not discriminate at the top, which is a similar situation as
in the case that a test has a ceiling effect. Pettersson (2010) uses both specially
designed tests and standardised tests in her comparison, and she uses actual
achievement; I use relative achievement. Gagné drew his conclusions from six
other studies6 (Chen, Lee, & Stevenson, 1996; Dumay, Coe, & Anumendem,
2014; Marques, Pais-Ribeiro, & Lopez, 2011; Marsh, 1990; Muijs, 1997) that
6
Verified trough an email conversation with Professor Gagné in Spring 2014.

110
measured a wide range of factors including cognitive and environmental
factors.

My conclusion is that, when discussing achievement, it is very important to take

the context, the method and the analysis into consideration. What sort of test is
used, is it a normally distributed test or does it have a ceiling or a roof effect?
Does the test measure one thing in depth or is it broad? This is a current issue
in Swedish schools; some schools are discussing measuring students’ knowledge
from the age of six to be able to help each child to develop according to his or
her potential. If this mapping will be implemented, schools need a variety of
tests that are normally distributed to be able not only to find students who need
more help but also those who need greater challenges in school.

In Sweden, in terms of support to students in schools, the focus is mostly on

students with special needs. If a student is at risk of failing a subject, it is the
headmaster’s responsibility to produce an action plan for that student in order
to help him or her to pass the subject (SFS 2010:800). However, there is one
group of students that needs more attention and perhaps also needs an action
plan, these being the students who easily reach the goals in school subjects and
are most often not challenged at a level that benefits them most. Not being
challenged on a suitable level can make it harder for a student to concentrate
and to achieve.

6.2.3 Assessment and challenges

Teachers have to analyse and collect evidence concerning their students’

knowledge (Wiliam, 2007). However, it is important to use different kinds of
assessment tools or tests in a multiple context (Boaler, 2006; Jönsson &
Svingby, 2008; Moltzen, 2009). This study shows that there are students that
achieve among the top in one kind of test but not necessarily do in another.

The present study describes a group of students that achieves high in a non-
curriculum bounded test but not in a curriculum bounded test. Those students
might have developed a personal mathematical knowledge (A. Pettersson, 2007)
on their own, perhaps through different out-of-school activities. However, the
school serves to develop students’ official mathematical knowledge (A.

111
Pettersson, 2007), i.e. knowledge defined by the curriculum that curriculum
bounded tests such as the national test measure.

To challenge and motivate a student, the task or the problem has to be on the
right level – not too easy, not too difficult - which of course is individual.
Especially gifted students are less motivated when they work with tasks or
problems that are too easy for them (Nolte, 2012a). To achieve in mathematics,
it is therefore important that each student, gifted or not, is given tasks and
problems that are challenging for him or her. If the tasks and problems are too
easy, this can actually lead to student achieving lower than if they were given
more challenging tasks or problems. The student can naturally not be passive if
she/he should succeed and achieve highly; the student must be motivated and
have endurance and self-discipline (Lucas & Claxton, 2010; Nolte, 2012a). Time
limits in the present study did not give the opportunity to explore a student’s
motivation, endurance or self-discipline. The author is aware that the
achievement of a student also depends on the student, not only on the teacher
or other environmental aspects.

Challenging a student at the correct individual level is important but is not an

easy mission. The fact that in a heterogeneous classroom at year 9 the 15%
lowest achieving students possess mathematical knowledge corresponding to
expected knowledge in year 4 (Engström & Magne, 2006) does not make it
easier. This wide distribution of knowledge in a heterogeneous classroom,
together with the fact that the “challenge factor” is important, tells us that it
might be appropriate to let classmates work with different tasks and problems
on different levels of challenge, at least if all students should get the
opportunity to develop as far as possible, which they have the legal right to do
(SFS 2010:800).

Students are different: some need one sort of challenge, perhaps in problem
solving, while other students need another sort of challenge, perhaps in
communicating mathematics. Some students may have good mathematical
competencies but are not able to show them in conventional teacher made tests
or in the national test. By giving those students opportunities to show their
skills in other, different tests for example the mathematical kangaroo, is one
way to lift students that are traditionally not noticed by the teacher. It might
give the teacher the possibility to, through positive feedback, help the student
to gain better self confidence in mathematics and perhaps also in the end help

112
the student to succeed in school mathematics, also improving their official
knowledge.

It is important to see the group of students who need to be challenged as fluent

and not fixed (Moltzen, 2009). In the present study, the fluidity of students’
achievement is seen in the fact that the groups of students who made large
movements up or down in relative achievement are large: approximately 10%
of those who had the possibility to make large movements did so. It is
reasonable to conclude that, if students’ achievement is fluent, then the need of
challenges and support is also fluent.

6.2.4 The problem solving competency

This study compares two groups with each other, one being of special interest.
The group of the greatest interest is Group 1 because those students’
achievement is most surprising. The mathematical kangaroo did not affect the
students’ subject grade as it was done by the students on a normal school day.
Competing in the mathematical kangaroo is a very relaxed happening in
Sweden, although students are told to do their best. The reason that the
competition is relaxed and that it does not affect the students’ grade might give
the effect that students do not do their best. The situation is the same for all
students who participate. It can explain why students that are highly ranked in
the national test decrease their ranking position in the mathematical
kangaroo - why should they make any effort? Nevertheless, it does not explain
why some students who are ranked low and have a low achievement in the
national test increase their ranking position in the mathematical kangaroo to
become among the top 20% ― why should they make any effort?

The problem solving competency seems to be an important factor in the

explanation. The mathematical kangaroo offers more opportunity to activate
the problem solving competency than does the national test in year 6. Group 1
activate the problem solving competency more than Group 2, and the tasks that
differ most in response rate between the two groups are also connected to the
problem solving competency.

113
What is special in the problem solving competency? According to the chosen
framework MCRF (Lithner et al., 2010) a problem is a problem if the solution
strategy is not known in advance by the solver. Tasks in the mathematical
kangaroo are discussed in problem solving situations (E. Pettersson, 2011), and
mathematical problems are used in research on gifted students i.e. (Krutetskii,
1976; Nolte, 2012a). Using a problem solving approach can also result in
students achieving better and that they continue their mathematics education
path by choosing more advanced mathematics courses (Boaler, 2006).

The tasks in the mathematical kangaroo are often mentioned as mathematical

problems when talking to people in the field of mathematics, both teachers and
researchers. The present study confirms that many of the tasks in the
mathematical kangaroo give the opportunity to activate the problem solving
competency. One aim of the mathematical kangaroo is to offer interesting
challenges (Wettbewerbsbedingungen, 2013). It is important to challenge
students with tasks at the right level; the task cannot be too easy or too hard if
the student is to be motivated. Nolte (2012a) writes that especially gifted
students are at risk of losing motivation if the tasks are too easy. I believe it is
plausible that all students are at most motivated when working with tasks at the
right level for them. The fact that the mathematical kangaroo aims to offer
challenging tasks and that the challenge level is important for gifted students
makes it interesting to mention that the mathematical kangaroo has inspired
part of a model that aims to identify mathematically gifted students (Pitta-
Pantazi et al., 2011).

A conclusion could be that, since the tasks in the mathematical kangaroo are
interesting and challenging and to relatively high degree mathematical problems,
some students find those tasks challenging and become motivated and, for that
reason, achieve better than they normally do in the national test.

6.3 Method discussion

One aim of the study is to investigate a method that can be used to explain
differences in achievement and connect this to mathematical competencies. The
framework used to analyse mathematical competencies is crucial for the results
of this study. This chapter discusses some of the methods used in the study:

114
Were there other choices? In what way could those choices have affected the
study?

6.3.1 Mathematical competencies

The framework chosen to analyse mathematical competencies is the MCRF

(Lithner et al., 2010). This is chosen partly because it has been used in Sweden
in situations involving national tests (Lithner, 2011).

However analysing tasks can be done in many different ways; for example it is
possible to analyse tasks according to the work of Krutetskii (1976). Using
Krutetskii (1976) would change the focus from competencies mentioned in
school mathematics and curricula to mathematical abilities mentioned as being
important for students gifted in mathematics. This shift would have been
interesting, but the main aim of the study has to do with students who not are
visualised through school mathematics despite their probably possessing good
mathematical competencies.

With the aim to connect the study with school mathematics, there are still other
choices for analysing the tasks that could have been chosen. For example it
could have been possible to analyse what opportunities the tasks in the
different tests gives for imitative reasoning and creative reasoning (Lithner,
2008). The results would probably have been different and would have been
discussed from another perspective, with different kinds of reasoning in focus.
The reasoning framework (Lithner, 2008) is partly used in a study that
concludes that learning through creative mathematically-founded reasoning
(CMR) is more beneficial for students with low cognitive ability (Jonsson,
Norqvist, Liljekvist, & Lithner, 2014). The use of creative and imitative
reasoning could make it possible to analyse the differences between the groups
in a different light, or as a complement to the one used, to view the phenomena
from different perspectives.

115
6.3.2 Relative comparison

The analysis uses relative measurements, which sometimes makes the

comparison complex and difficult to follow. With regard to individuals, the
purpose is to, within an individual measure how the competencies are
distributed. This means that calculating how many times a competency is
activated in relation to the total amount of activated competencies for that
individual fulfils the purpose. Another option that at first sight seems most
natural is to calculate for each individual how much a competency is activated
according to how much the test gives the opportunity for that competency. If
that method is used, however, the results would be connected to the
individual’s achievement in the test. In the comparison of activation of
mathematical competencies between the two groups identified, the desire is to
eliminate the achievement factor. The groups are identified according to
achievement on tests, but the achievement factor is not important when
comparing how mathematical competencies are activated.

6.3.3 Empirical data

This study uses already existing data on students’ results in the national test. As
an alternative, the study could have collected data from national tests in “real
time”, this would have given a richer material with access to results on the task
level and students’ solutions. A richer material would have given possibility for
a deeper analysis.

6.3.4 Participants

Students participating in this study all come from the same municipality, this
choice was made to make it possible to carry out the study. Another way to
choose participants would be to distribute the participants over a larger
geographic area, either randomly or consciously trying to achieve, for example,
an even distribution of socio economics and other factors known to influence

116
achievement. It is difficult to speculate how it would have affected the study
since, in the method used here, all public schools in the municipality
participated, which means that the participants come from a widely spread
background. However, no investigation of their socio economic status or
background is involved in the data.

6.3.5 Implementation of the mathematical kangaroo

To be able to smoothly collect as much data as possible, the participants’

ordinary teachers were asked to implement the mathematical kangaroo. All
teachers were given the same instructions, both verbally and in writing.
However, the author was not present at the time of the implementation of the
mathematical kangaroo in the schools. Another way to implement the kangaroo
and secure the equality of the implementation would have been to either let one
person take care of the implementation in all schools or to gather the
participants in one single place during the implementation.

6.3.6 Representativeness of the sample

The thesis has a strong focus on showing the representativeness of the sample
which is important because the study starts from a whole population and
continues to deal with a sample: the validity of the results is therefore directly
connected to how well the sample represents the population. Several methods
are used see: the box plots in Figure 7 and Figure 8, the frequency graphs in
Figure 10 and Figure 11, and a chi-square test 4.2.1.4. The combination of
methods both reveals the ceiling effect in the national tests and where there is a
lack of individuals in the sample compared to the population. There are other
methods that come to mind for measuring the representativeness, such as the t-
test. However, the t-test is most often used when data follow a normal
distribution, which is not the case for the data in this study.

117
7 Implications

7.1 Practice

Looking only at high achievement can be wrong when we try to identify

students who need greater challenge; for example students from low socio
economic homes are disadvantaged (Mattsson, 2013; Silverman & Miller, 2009).
Support from schools, teachers and parents is important for nurturing
achievement (Nolte, 2012a). Schools need to learn how to reward, support and
stimulate those that do not necessarily achieve high in a school subject but still
have the potential to develop a talent. Some patterns are of extra importance in
helping those students such as supportive adults and opportunities to be
awarded honours and take advanced classes (Reis, Colbert, & Hébert, 2004).

The national test aims to mirror the curriculum. Students who succeed in the
national test show that they fulfil the curriculum. The national tests are aimed
to support teachers in the assessment of students’ mathematical knowledge
according to the curriculum. Therefore the tests should discriminate both in the
bottom and in the top. The national tests used in this study both have a ceiling
effect which makes them unsuitable to use as support in the assessment,
however, both test aimed to test the passing level.

This study implicate that it is important to take into account if a tests have a
ceiling effect if it is used to support assessment. In Sweden the curriculum is
goal oriented which means that in theory all students can reach the highest
grade. Despite the goal oriented curriculum, in reality there is always a
distribution of students’ knowledge. To be able to develop their knowledge
students need to be challenged in their education, using tests that discriminate
both in the bottom and in the top helps a teacher to find the correct level of
challenges for each student. According to Nolte (Nolte, 2012a) it is of extra
importance to challenge gifted students, why the discrimination at the top can
not be omitted.

As shown in this study, there are students who achieve low in the national test
but high in the mathematical kangaroo. It might be that they do not have the
mathematical knowledge that fulfils the curriculum, but it is still possible that

118
they have some competencies that are important according to the curriculum.
According to Krutetskii (1976), one well developed competency can help to
outweigh another, although the students probably need some teacher guidance
to help one competency compensate for another. Another possibility is that
they have learned mathematics outside the curriculum; they have their own
informal personal curriculum that does not follow the national official
curriculum. This personal curriculum might give them good mathematical
competence but still low grades in school mathematics – they do not follow the
rules.

The results of this study say that there are students that can achieve in
mathematics although they do not do so in traditional school mathematics. The
results also indicate that some of those students can be identified by means of
the mathematical kangaroo. The study indicates that those students are better in
the problem solving competence compared to some of the students who
achieve among the best in the national test. This study thus verifies what other
studies have held, i.e. (Boaler, 2006; Jönsson & Svingby, 2008): that variation is
important not only in the teaching situation but also in assessment situations.
This study does not say that using the mathematical kangaroo is the solution
and using it does not help teachers to discover all students. The study says that
using the mathematical kangaroo as a complement in assessments is one way to
find students that possess some mathematical competencies that not are
visualised in traditional tests such as the national test.

7.2 Research

Mathematical competencies are important in mathematics education today,

partly because of the great influence that international standardised
measurements such as PISA (OECD, 2014) have on the curriculum in an
international perspective. The mathematical framework of PISA uses the word
capability; competency was earlier used (OECD, 2013). It has also become
more and more common for countries to use external standardised tests, which
has brought equality in education (Jönsson & Svingby, 2008). Equality is good
in one perspective because it makes it easier to compare different countries and
contexts. However, it is important to remember that education should,
according to Swedish law, give opportunities to students to show their

119
competencies and to be challenged (SFS 2010:800). It is therefore not only
important to use variation in teaching but also in assessments.

This study shows that one way to “see” students that not are “seen” through
national tests is to use a non-curriculum bounded test – for example the
mathematical kangaroo. The study also investigates a method that aims to
explore differences in achievement according to mathematical competencies.
The results say that some students that achieve among the highest in the
mathematical kangaroo to a greater extent activate the problem solving
competency than do some of the highest achievers in the national test.

The results of the study tell the research society in mathematics education that
it is also important to look at non-curriculum bounded assessment activities and
how those can be used to lift up and support students that might possess good
mathematical competencies.

7.2.1 Further research

The method used in this study can be further developed and explored. In
further research, the method can be used in studies that compare different
groups of students and how they activate mathematical competencies.
Especially the part that analyses tasks according to what mathematical
competencies they give the opportunity to activate can be made more reliable,
for example by using a larger amount of active teachers to analyse and discuss
the tasks or discussing the analysis in greater depth with more research
colleagues in mathematics education.

The national tests in the study both had a ceiling effect. It would be interesting
to follow movements in achievement over time using tests where the response
rates are normally distributed, discriminating both at the bottom and at the top,
thereby offering challenges to almost all students by means of that specific test.

The work of investigating what factors there are in tasks that allow one group
of students to achieve while another group does not is also important in further
investigations. Factors such as mathematical competencies can be investigated
in greater depth, but it is also important to investigate other factors such as
creativity, challenge level and cognitive aspects such as for example motivation.

120
However, the most important thing is to develop strategies in teaching and
learning situations that aim to challenge each student in the classroom, the non-
gifted as well as the gifted. The importance of using tasks that are challenging
has already been stated; the next step is to place challenging tasks into the
heterogeneous classroom situation and develop strategies that help to create a
meaningful school day with opportunities to be a challenge for all students, not
least the gifted ones.

121
8 Words ending the thesis or “What if?”

In a seminar held by Professor Jeppe Skott I was inspired to always ask myself
when I have come to a conclusion or make a statement “What if?” to never be
satisfied with finding an answer, because there might be another explanation. In
the last words of this thesis I want to relate to the “what if?” question, because
I am not satisfied; this study has given me more questions than it has answers.
At the beginning of the treatment of a “what if?” question, most of the
thoughts that come to the mind during the focused work of trying to answer a
question exist inside some chosen frames.

The phenomenon that there are students who are high achievers in the
mathematical kangaroo but not in the national test is described quantitatively in
this study. The phenomenon has been noticed by me and other in-service
teachers (Mattsson, 2013). Some teachers have wondered whether students who
achieve highly in the mathematical kangaroo but not in the national test are
gifted (Mattsson, 2013). This study has not shown in any way that the students
identified are gifted or not.

But what if they are?

Or…

What if the tasks in the mathematical kangaroo are creative and make students
with less cognitive abilities succeed?

Or…

What if …?

The questions will never end…

122
References

Advisory Committee on Mathematics Education. (2012). Raising the bar:

Developing able young mathematicians. Retrieved 12/12, 2012, from
http://www.acme-uk.org/home
Bar-On, R., & Maree, J. G. (. (2009). In search of emotional-social giftedness.
In L. V. Shavinina (Ed.), International handbook on giftedness (pp. 559-570).
Dordrecht: Springer.
Baroody, A. J., Feil, Y., & Johnson, A. R. (2007). An alternative
reconceptualization of procedural and conceptual knowledge. Journal for
Research in Mathematics Education, 38(2), 115-131.
Benbow, C. P., Lubinski, D., Shea, D. L., & Eftekhari-Sanjani, H. (2000). Sex
differences in mathematical reasoning ability at age 13: Their status 20
years later. Psychological Science, 11(6), 474-480.
Boaler, J. (2006). How a detracked mathematics approach promoted respect,
responsibility, and high achievement. Theory into Practice, 45(1), 40-46.
Boesen, J. (2006). Assessing mathematical creativity: Comparing national and teacher-
made tests, explaining differences and examining impact. Doctoral thesis, Umeå:
Umeå University.
Boesen, J., Helenius, O., Bergqvist, E., Bergqvist, T., Lithner, J., Palm, T., et al.
(2014). Developing mathematical competence: From the intended to the
enacted curriculum. The Journal of Mathematical Behavior, 33, 72-87.
Boesen, J., Lithner, J., & Palm, T. (2010). The relation between types of
assessment tasks and the mathematical reasoning students use. Educational
Studies in Mathematics, 75(1), 89-105.
Chen, C., Lee, S., & Stevenson, H. W. (1996). Long-term prediction of
academic achievement of American, Chinese, and Japanese adolescents.
Journal of Educational Psychology, 88(4), 750-759.
Dörfler, W. (2006). Inscriptions as objects of mathematical activities. In J.
Maasz, & W. Schloeglmann (Eds.), New mathematics education research and
practice (pp. 95-96). Rotterdam: Sense Publishers.
Dorr-Bremme, D. W., & Herman, J. L. (1986). Assessing student achievement: A
profile of classroom practices. Retrieved 06/24, 2014, from
http://files.eric.ed.gov/fulltext/ED338691.pdf
Dumay, X., Coe, R., & Anumendem, D. N. (2014). Stability over time of
different methods of estimating school performance. School Effectiveness and
School Improvement: An International Journal of Research, Policy and Practice, 25(1),
64-82.
Engström, A., & Magne, O. (2006). Medelsta-matematik III: Eleverna räknar.
Örebro: Örebro University.
Gagné, F. (2005). From gifts to talents: The DMGT as a developmental model.
In R. J. Sternberg, & J. E. Davidsson (Eds.), Conceptions of giftedness (pp. 98-
119). New York: NY: Cambridge University Press.
Gagné, F. (2000). A differentiated model of giftedness and talent (DMGT). Retrieved
10/06, 2014, from

123
 http://www.curriculumsupport.education.nsw.gov.au/policies/gats/assets
/pdf/poldmgt2000rtcl.pdf
Gagné, F. (2004). Transforming gifts into talents: The DMGT as a
developmental theory. High Ability Studies, 15(2), 119-147.
Gipps, C. (1999). Socio-cultural aspects of assessment. Review of Research in
Education, 24(1), 355-392.
Gross, M. U. M. (2009). Highly gifted young people: Development from
childhood to adulthood. In L. V. Shavinina (Ed.), International handbook on
giftedness (pp. 337-351). Dordrecht: Springer.
Häggblom, L. (2000). Räknespår: Barns matematiska utveckling från 6 till 15 års ålder.
Doctoral thesis, Åbo: Åbo Akademis Förlag.
Hallesson, Y. (2011). Högpresterande gymnasieelevers läskompetenser. Licentiate thesis,
Stockholm: Stockholm University.
Hollander, M., Wolfe, D., A., & Chicken, E. (2014). Nonparametric statistical
methods (3rd ed.). New Jersey, NJ: John Wiley & Sons.
Jönsson, A., & Svingby, G. (2008). Underlag till ramverk för en provbank i
grundskolan. Educare, (2), 57-93.
Jonsson, B., Norqvist, M., Liljekvist, Y., & Lithner, J. (2014). Learning
mathematics through imitative and creative reasoning. Journal of
Mathematical Behavior, 36, 20-32.
Kilpatrick, J., Swafford, J., & Findell, B. (2001). Adding it up: Helping children learn
mathematics. Washington, DC: National Academies Press.
Korp, H. (2006). Lika chanser i gymnasiet? en studie om betyg, nationella prov och social
reproduktion. Doctoral thesis, Malmö: Malmö University.
Krutetskii, V. A. (1976). The psychology of mathematical abilities in schoolchildren.
Chicago, IL: The University of Chicago Press.
Lampert, M., & Cobb, P. (2003). Communication and language. In J. Kilpatrick,
W. G. Martin & D. Schifter (Eds.), A research companion to principles and
standards for school mathematics (pp. 237-249). Reston, VA: National Council
of Teachers of English.
Lisper, H., & Lisper, S. (2005). Statistik för samhällsvetare. Stockholm: Liber.
Lithner, J. (2008). A research framework for creative and imitative reasoning.
Educational Studies in Mathematics, 67(3), 255-276.
Lithner, J. (2011). National tests in mathematics a catalyst for implementing educational
reforms. Retrieved 03/02, 2014, from
http://www.umu.se/sok/english/research-database/view-research-
projects?code=483&currentView=base&doSearch=&scbCode=&searchStr
ing=&uid=bjnpag02&guise=anst1
Lithner, J., Bergqvist, E., Bergqvist, T., Boesen, J., Palm, T., & Palmberg, B.
(2010). Mathematical competencies: A research framework. In C. Bergsten,
E. Jablonka & T. Wedege (Eds.), Mathematics and mathematics education:
Cultural and social dimensions. The seventh mathematics education research seminar,
Stockholm, Janurary 26-27, 2010 (pp. 157-167). Stockholm: Swedish
Mathematics Education Research (SMDF).
Lucas, B., & Claxton, G. (2010). New kinds of smart: How the science of learnable
intelligence is changing education. Maidenhead: Open University Press.

124
Maier, H., & Schweiger, F. (1999). Mathematik und Sprache: Zum Verstehen und
Verwenden von Fachsprache im Mathematikunterricht. Wien: öbv und hbt.
Marques, S. C., Pais-Ribeiro, J., & Lopez, S. J. (2011). The role of positive
psychology constructs in predicting mental health and academic
achievement in children and adolescents: A two-year longitudinal study.
Journal of Happiness Studies, 12(6), 1049-1062.
Marsh, H. W. (1990). Causal ordering of academic self-concept and academic
achievement: A multiwave, longitudinal panel analysis. Journal of Educational
Psychology, 82(4), 646-656.
Mattsson, L. (2013). Tracking mathematical giftedness in an egalitarian context.
Doctoral thesis, Gothenburg: Chalmers University of Technology and
University of Gothenburg.
McCoach, D. B., Goldstein, J., Behuniak, P., Reis, S. M., Black, A. C., Sullivan,
E. E., et al. (2010). Examining the unexpected: Outlier analyses of factors
affecting student achievement. Journal of Advanced Academics, 21(3), 426-468.
Mesa, V., Gómez, P., & Cheah Hock, U. (2013). Influence of international
studies of student achievement on mathematics teaching and learning. In
M. A. Clements, A. J. Bishop, C. Keitel, J. Kilpatrick & F. K. Leung (Eds.),
Third international handbook of mathematics education (pp. 861-900). New York,
NY: Springer.
Moltzen, R. (2009). Talent development across the lifespan. In L. V. Shavinina
(Ed.), International handbook on giftedness (pp. 352-379). Dordrecht: Springer.
Mönks, F. J., & Pflüger, R. (2005). Gifted education in 21 european countries: Inventory
and perspective. Nijmegen: Radboud University Nijmegen.
Muijs, R. D. (1997). Predictors of academic achievement and academic self‐
concept: A longitudinal perspective. British Journal of Educational Psychology,
67(3), 263-277.
Nationellt centrum för matematikutbildning. (2013). Vad är kängurun -
matematikens hopp? Retrieved 02/12, 2013, from
http://www.ncm.gu.se/node/1525
NCTM. (2000). Principles and standards for school mathematics. Reston, VA: National
Council of Teachers of Mathematics.
Niss, M., & Höjgaard, T. (Eds.). (2011). Competencies and mathematical learning
(English ed.). Roskilde: Roskilde University.
Nolte, M. (2012a). Challenging math problems for mathematically gifted children. Paper
presented at the The 7th Mathematical Creativity and Giftedness
International Conference, Busan, Repulic of Korea, July 12-18 2012. pp.
27-45.
Nolte, M. (2012b). Fostering mathematically talented children (PriMa). Retrieved
08/14, 2012, from http://blogs.epb.uni-hamburg.de/nolte/?page_id=671
OECD. (2013). PISA 2012 mathematics framework. PISA 2012 assessment and
analytical framework. mathematics, reading, science, problem solving and financial
literacy (pp. 23-58) OECD Publishing.
OECD. (2014). Program for international student assessment. Retrieved 03/02, 2014,
from http://www.oecd.org/pisa/

125
Pettersson, A. (1990). Att utvecklas i matematik: En studie av elever med olika
prestationsutveckling. Doctoral thesis, Stockholm: Stockholm: Almquist &
Wiksell.
Pettersson, A. (2004). Assessing students´ knowledge. Retrieved 06/03, 2014, from
http://www.mai.liu.se/SMDF/AbstractAP.pdf
Pettersson, A. (2007). Pedagogisk bedömning–bedömning för lärande. In M.
Olofsson (Ed.), Symposium 2006: Bedömning, flerspråkighet och lärande (pp. 11-
20) Stockholm: HLS Förlag.
Pettersson, A., & Boistrup, B. L. (2010). National assessment in Swedish
compulsory school. In B. Sriraman (Ed.), The first sourcebook on Nordic
research in mathematics education (pp. 373-385). Charlotte, NC: Information
Age Publishing.
Pettersson, E. (2011). Studiesituationen för elever med särskilda matematiska förmågor.
Doctoral thesis, Växjö: Linnaeus University Press.
Pitta-Pantazi, D., Christou, C., Kontoyianni, K., & Kattou, M. (2011). A model
of mathematical giftedness: Integrating natural, creative, and mathematical
abilities. Canadian Journal of Science, Mathematics and Technology Education, 11(1),
39-54.
Polit, D. F. (2008). Nursing research: Generating and assessing evidence for nursing
practice (8th ed.). Philadelphia, PA: Lippincott Williams & Wilkins.
R Core Team. (2013). R: A language and environment for statistical computing.
Retrieved 03/02 2013, from http://www.R-project.org/
Reis, S. M., Colbert, R. D., & Hébert, T. P. (2004). Understanding resilience in
diverse, talented students in an urban high school. Roeper Review, 27(2), 110-
120.
Renzulli, J. S. (2005). The three-ring conception of giftedness. In R. J.
Sternberg, & J. E. Davidsson (Eds.), Conceptions of giftedness (pp. 246-279).
New York: NY: Cambridge University Press.
Säfström, A. I. (2013). Exercising mathematical competence: Practising representation
theory and representing mathematical practice. Doctoral thesis, Gothenburg:
Chalmers University of Technology and University of Gothenburg.
SFS 2010:800. Skollag. Stockholm: Utbildningsdepartementet.
Silverman, L. K., & Miller, N. B. (2009). A feminine perspective of giftedness.
In L. V. Shavinina (Ed.), International handbook on giftedness (pp. 99-128).
Dordrecht: Springer.
Skolverket. (2008). Informationsmaterial om nationellt prov i åskurs 3, matematik.
Stockholm: Skolverket.
Skolverket. (2010). Ämnesproven i grundskolans årskurs 3. en redovisning av
utprövningsomgången 2009. Stockholm: Skolverket.
Skolverket. (2011a). Curriculum for the compulsory school, preschool class and the leisure-
time centre 2011. Stockholm: Skolverket.
Skolverket. (2011b). Kommentarmaterial till kursplanen i matematik. Stockholm:
Skolverket.
Skolverket. (2012). Ämnesproven i grundskolans årskurs 6. Stockholm: Skolverket.
Skolverket. (2014). Nationella prov & bedömningsstöd. Retrieved 06/30, 2014, from
http://www.skolverket.se/bedomning/nationella-prov-bedomningsstod

126
Stamm, M. (2008). Gifted and yet a dropout? An empirical study on the phenomen of
dropouts among gifted adolescents in Switzerland. Paper presented at the
European Conference on Educational Research in Göteborg, Sweden,
September 2008.
Stockholm University. (2013). PRIM-gruppen: Institutionen för matematikämnets och
naturvetenskapsämnenas didaktik. Retrieved 06/30, 2014, from
http://www.su.se/primgruppen/matematik
Stoeger, H. (2004). Interview with Francoys Gagné. High Ability Studies, 15(2),
167-172.
Szabo, A. (2013). Matematiska förmågors interaktion och det matematiska minnets roll
vid lösning av matematiska problem. Licentiate thesis, Stockholm: Stockholms
Universitet.
Vanderbilt University. (2014). Study of mathematical precosious youth. Retrieved
03/02, 2014, from https://my.vanderbilt.edu/smpy/
Vialle, W., Heaven, P. C. L., Ciarrochi, J., & Australian, R. C. (2007). On being
gifted, but sad and misunderstood: Social, emotional, and academic
outcomes of gifted students in the Wollongong youth study. Educational
Research and Evaluation: An International Journal on Theory and Practice, 13(6),
569-586.
Voutsina, C. (2012). Procedural and conceptual changes in young children's
problem solving. Educational Studies in Mathematics, 79(2), 193-214.
Wackerly, D. D., Mendenhall, W., & Scheaffer, R. L. (2002). Mathematical
statistics with applications (6th ed.). Pacific Grove, CA: Duxbary.
Wettbewerbsbedingungen. (2013). Känguru der mathematk. Retrieved 11/19,
2013, from http://www.mathe-kaenguru.de/
Wiliam, D. (2007). Keeping learning on track: Classroom assessment and the
regulation of learning. In F. K. Lester (Ed.), Second handbook of research on
mathematics teaching and learning (pp. 1051-1098). Charlotte, NC: IAP.

127
Appendicies

128
Appendix 1

Chi-square test comparing the sample and the population through their results in the national test in
year 3.

No. of
students, No. of Expected value,
actual value, students in 𝑉𝐴 (𝑉𝐸 − 𝑉𝐴 )2
Points 𝑉𝐸 = ∙ 247 𝑉𝐸
VA population 654

0-56 1 20 7.553517 5.685905

57-62 4 21 7.931193 1.948544
63-66 3 20 7.553517 2.745015
67-69 7 26 9.819572 0.809606
70-71 6 24 9.06422 1.03588
72-73 8 31 11.70795 1.174322
74-75 8 32 12.08563 1.381173
76 7 20 7.553517 0.040561
77 12 26 9.819572 0.484162
78 8 28 10.57492 0.626977
79 13 31 11.70795 0.142586
80 21 52 19.63914 0.094298
81 11 33 12.4633 0.171805
82 10 26 9.819572 0.003315
83 11 29 10.9526 0.000205
84 19 40 15.10703 1.003187
85 16 29 10.9526 2.326046
86 17 32 12.08563 1.998329
87 11 30 11.33028 0.009627
88 11 25 9.441896 0.257119
89 12 25 9.441896 0.69307
90 13 20 7.553517 3.927201
91 11 20 7.553517 1.572545
92-93 7 14 5.287462 0.554668
Sum 247 654 calc 28.68615
df 23
crit (0.05) 35.172
Appendix 2

Chi-square test comparing the sample and the population through their results in the national test in
year 6.

No. of
students, No. of Expected value,
actual students in 𝑉𝐴 (𝑉𝐸 − 𝑉𝐴 )2
𝑉𝐸 = ∙ 264 𝑉𝐸
Points value, VA population 611

0-32 3 21 9.07365 4.065533

33-40 2 19 8.209493 4.696733
41-45 6 21 9.07365 1.041182
46-51 9 24 10.36989 0.180965
52-55 9 23 9.937807 0.088499
56-58 3 21 9.07365 4.065533
59-60 7 21 9.07365 0.473902
61-62 8 22 9.505728 0.238511
63-65 8 26 11.23404 0.931012
66-67 6 24 10.36989 1.841476
68-69 9 19 8.209493 0.076119
70-71 8 26 11.23404 0.931012
72-73 9 29 12.53028 0.99462
74-75 18 36 15.55483 0.384374
76-78 22 29 12.53028 7.156715
79-80 12 24 10.36989 0.256249
81-82 18 33 14.25859 0.981733
83-84 16 32 13.82651 0.341665
85-86 12 22 9.505728 0.654489
87-88 13 29 12.53028 0.017608
89-90 13 21 9.07365 1.699011
91-92 15 23 9.937807 2.578617
93-94 12 21 9.07365 0.94378
95-97 17 28 12.0982 1.986051
98-106 9 17 7.345336 0.372742
Sum 264 611 calc. 36.99813
df 24
crit 36.415
(0.05)
Appendix 3
Efter underskrift, lämna till matematikläraren på skolan så snart som möjligt.

Samtyckesformulär för elev

Forskare på matematikavdelningen vid Karlstads Universitet vill jämföra ditt

resultat på Kängurutävlingen med dina tidigare resultat på nationella prov i
matematik.

När jämförelsen görs är ditt namn och personnummer kodat. Ingen, inte ens den
som gör jämförelsen, kommer veta vem som hör ihop med resultaten. Ditt
personnummer kommer genom ett dataprogram bli en kod som ser ut ungefär så
här:

8f3645edc7852a51cd251c9fda56d682

Du tar själv ställning till om du tycker det är okej att använda ditt resultat eller
inte. Enligt personuppgiftslagen har du rätt att en gång varje år ta kontakt med
oss på universitetet för att ta reda på vilka uppgifter vi har om dig.

JA. NEJ.

Jag samtycker till att Karlstads Jag samtycker inte till att Karlstads
universitet behandlar personuppgifter universitet behandlar personuppgifter
om mig i enlighet med det ovanstående. om mig i enlighet med det ovanstående.

Ort och datum Ditt personnummer

Underskrift

Klass och namnförtydligande

Karlstads universitet, 651 88 Karlstad

Universitetsgatan 2. Telefon 054-700 10 00 , Telefax 054-700 14 60 .
Postgiro 78 81 07-1 Org. nr. 202100-3120. www.kau.se
Appendix 4
Efter underskrift, lämna tillbaka till skolan med underskrift så snart som möjligt, tack.

Samtyckesformulär för vårdnadshavare

Vi bekräftar härmed att vi tagit del av informationen om deltagande i en

forskningsstudie där resultatet på Kängurutävlingen kommer att användas.
Resultatet jämförs med resultat på ämnesprovet i matematik för årskurs 3 och
årskurs 6.

Du har enligt 26 § personuppgiftslagen (1998:204) rätt att, en gång per kalenderår, efter
skriftligt undertecknad ansökan ställd till oss, få besked om vilka personuppgifter om
dig som vi behandlar och hur vi behandlar dessa. Du har också rätt att enligt 28 §
personuppgiftslagen begära rättelse i fråga om personuppgifter som vi behandlar om
dig.

JA. NEJ.

Jag samtycker till att Karlstads Jag samtycker inte till att Karlstads
universitet behandlar personuppgifter universitet behandlar personuppgifter
om mitt barn i enlighet med det om mitt barn i enlighet med det
ovanstående. ovanstående.

Ort och datum Elevens namn och personnummer

Elevens klass

Underskrift vårdnadshavare 1 Underskrift vårdnadshavare 2

Karlstads universitet, 651 88 Karlstad

Universitetsgatan 2. Telefon 054-700 10 00 , Telefax 054-700 14 60 .
Postgiro 78 81 07-1 Org. nr. 202100-3120. www.kau.se
Appendix 5

Information till lärare om forskningsstudie om elevers prestationer i matematik

Denna informations riktar sig till dig som undervisar i matematik i årskurs 7 i X kommun
och avser en förfrågan om att bistå med utdelning av informationsblad och
samtyckesblankett till dina elever samt att bistå med genomförandet av Kängurutävlingen.

Bakgrund och syfte

Denna studie vill jämföra hur elever presterar i matematik när matematikkunskaper mäts
på olika sätt. Studien avser att jämföra resultat på de svenska nationella ämnesproven i
matematik i årskurs 3 och 6 med resultat på det internationella matematikprovet
”Kängurutävlingen”. Målet är att bidra med kunskap om elevers behov av olika
matematikundervisning, en del i vägen att ge alla elever matematikundervisning på den
nivå respektive elev behöver.

Deltagande

Förfrågan om deltagande i studien går till samtliga elever som gick i årskurs 3 i skola i X
kommun läsåret 2008/2009 och som nu går i årskurs 7.

Kängurutävlingen

Kängurutävlingen genomförs under mars–juni månad 2013 under skoltid av elevernas

ordinarie lärare. Resultatet på tävlingen ska inte påverka elevens betyg eller bedömning av
elevens skolprestation. De vårdnadshavare eller elever som inte vill att deras resultat ska
tas med i studien kommer inte att ingå i studien.

Lärarnas insats

Din insats består i att dela ut informationsblad och samtyckesblanketter till eleverna samt
att ta in dessa. Vi hämtar dem när de är insamlade. Du låter sedan dina elever delta i
Kängurutävlingen under lektionstid. Vi distribuerar tävlingsmaterial till skolan.

Rättning av provet görs av oss. När de är rättade lämnar vi tillbaka proven till dig. Om du
inte har möjlighet att genomföra provet under lektionstid så kan vi medverka. Tag då
kontakt med oss.

Hur går studien till?

Proven distribueras och rättas av forskarna på Karlstads universitet. Resultatet jämförs

sedan med resultaten på ämnesproven i matematik i årskurs 3 och årskurs 6.
Appendix 5

Hantering av data och sekretess

Alla personuppgifter kommer att kodas av forskargruppen innan databearbetning och

analys görs. Ingen obehörig kommer att kunna ta del av uppgifterna. Allt material
kommer att förvaras inlåst i dokumentskåp på universitetet.

Studien förväntas vara avslutad under hösten 2014. Materialet och kodnyckeln kommer
att förstöras enligt riksarkivets föreskrifter i statliga myndigheters forskningsverksamhet.
Enligt Personuppgiftslagen (1998:204) har vårdnadshavare rätt att en gång per kalenderår
få besked om vilka uppgifter som finns lagrade om sitt barn.

Hur får jag information om studiens resultat?

Studiens resultat kommer att publiceras i en licentiatavhandling under hösten 2014.

Frivillighet

Allt deltagande i studien är frivilligt och eleven och/eller dess vårdnadshavare har rätt att
när som helst och utan särskild anledning välja att avbryta deltagandet i studien. Elevens
resultat kommer då att raders från materialet. Kontakta i sådana fall någon av forskarna.

Ansvariga

Karlstads universitet är forskningshuvudman och personuppgiftsansvarig. Ansvarig för

genomförandet av studien är docent Arne Engström. Insamling och bearbetning av alla
data genomförs av forskarstuderande Elisabet Mellroth. Undrar ni över något är ni
välkomna att kontakta oss.

Hälsningar

Arne Engström Elisabet Mellroth

Docent Arne Engström Elisabet Mellroth

Institutionen för matematik Institutionen för matematik
och datavetenskap och datavetenskap
Karlstads universitet Karlstads universitet
Universitetsgatan 1 Universitetsgatan 1
561 88 KARLSTAD 561 88 KARLSTAD
Telefon: 054-700 24 67 Telefon: 054-700 24 35
Epost Epost
[email protected] [email protected]
Appendix 6

Competency analysis on Task no. 3 in the mathematical kangaroo Benjamin 2013

A reasonable and required solution is that a student tries different tracks and come to a
conclusion of which one that is the shortest.
Competency Question to ask in the analysis of the task. Classifi- Classifi-
For all competencies, if the answer is; cation cation
Yes – the classification is ‘1’ ‘0’ ‘1’
No – the classification is ‘0’

Applying It is not necessary to involve a sequence of mathematical actions. 0

procedure
(App)
Reasoning It is necessary to justify or argue mentally that the chosen track is the 1
(Rea) shortest.

Communication The words in the task are every-day language, but the sentence is 1
(Com) mathematically constructed, “How many times must he at least turn” It is
necessary for the student to correctly interpret the mathematical language to
solve the task.

Representation The picture is a representation of a map (a map is seen as a mathematically 1

(Rep) entity), the task cannot be solved without the map.

Connection No connections between mathematical entities or representations are 0

(Con) necessary to do for solving the task.

Problem According to the teachers this was a task were the solution process was not 1
solving known in advanced.
(Pro)

The analyse results in a reduced matrix for task 3.

Task 3 Class
App 0
Rea 1
Com 1
Rep 1
Con 0
Pro 1
 Studies in Science and Technology Education
ISSN 1652-5051

1. Margareta Enghag (2004): MINIPROJECTS AND CONTEXT RICH PROBLEMS –

Case studies with qualitative analysis of motivation, learner ownership and
competence in small group work in physics. (licentiate thesis) Linköping University
2. Carl-Johan Rundgren (2006): Meaning-Making in Molecular Life Science Education –
upper secondary school students’ interpretation of visualizations of proteins. (licentiate
thesis) Linköping University
3. Michal Drechsler (2005): Textbooks’, teachers’, and students´ understanding of
models used to explain acid-base reactions. ISSN: 1403-8099, ISBN: 91-85335-40-1.
(licentiate thesis) Karlstad University
4. Margareta Enghag (2007): Two dimensions of Student Ownership of Learning during
Small-Group Work with Miniprojects and context rich Problems in Physics. ISSN:
1651-4238, ISBN: 91-85485-31-4. (Doctoral Dissertation) Mälardalen University
5. Maria Åström (2007): Integrated and Subject-specific. An empirical exploration of
Science education in Swedish compulsory schools. (Licentiate thesis) Linköping
university
6. Ola Magntorn (2007): Reading Nature: developing ecological literacy through
teaching. (Doctoral Dissertation) Linköping University
7. Maria Andreé (2007): Den levda läroplanen. En studie av naturorienterande
undervisningspraktiker i grundskolan. ISSN: 1400-478X, HLS Förlag: ISBN 978-91-
7656-632-9 (Doctoral Dissertation, LHS)
8. Mattias Lundin (2007): Students' participation in the realization of school science
activities.(Doctoral Dissertation) Linköping University
9. Michal Drechsler (2007): Models in chemistry education. A study of teaching and
learning acids and bases in Swedish upper secondary schools ISBN 978-91-7063-112-
2 (Doctoral Dissertation) Karlstad University
10. Proceedings from FontD Vadstena-meeting, April 2006.
11. Eva Blomdahl (2007): Teknik i skolan. En studie av teknikundervisning för yngre
skolbarn. ISSN: 1400-478X, HLS Förlag: ISBN 978-91-7656-635-0 (Doctoral
Dissertation, LHS)
12. Iann Lundegård (2007): På väg mot pluralism. Elever i situerade samtal kring hållbar
utveckling. ISSN:1400-478X, HLS Förlag: ISBN 978-91-7656-642-8 (Doctoral
Dissertation, LHS)
13. Lena Hansson (2007): ”Enligt fysiken eller enligt mig själv?” – Gymnasieelever,
fysiken och grundantaganden om världen. (Doctoral Dissertation) Linköping
University.
 Studies in Science and Technology Education
ISSN 1652-5051

14. Christel Persson (2008): Sfärernas symfoni i förändring? Lärande i miljö för hållbar
utveckling med naturvetenskaplig utgångspunkt. En longitudinell studie i
grundskolans tidigare årskurser. (Doctoral Dissertation) Linköping University
15. Eva Davidsson (2008): Different Images of Science – a study of how science is
constituted in exhibitions. ISBN: 978-91-977100-1-5 (Doctoral Dissertation) Malmö
University
16. Magnus Hultén (2008): Naturens kanon. Formering och förändring av innehållet i
folkskolans och grundskolans naturvetenskap 1842-2007. ISBN: 978-91-7155-612-7
(Doctoral Dissertation) Stockholm University
17. Lars-Erik Björklund (2008): Från Novis till Expert: Förtrogenhetskunskap i kognitiv
och didaktisk belysning. (Doctoral Dissertation) Linköping University.
18. Anders Jönsson (2008): Educative assessment for/of teacher competency. A study of
assessment and learning in the “Interactive examination” for student teachers. ISBN:
978-91-977100-3-9 (Doctoral Dissertation) Malmö University
19. Pernilla Nilsson (2008): Learning to teach and teaching to learn - primary science
student teachers' complex journey from learners to teachers. (Doctoral Dissertation)
Linköping University
20. Carl-Johan Rundgren (2008): VISUAL THINKING, VISUAL SPEECH - a Semiotic
Perspective on Meaning-Making in Molecular Life Science. (Doctoral Dissertation)
Linköping University
21. Per Sund (2008): Att urskilja selektiva traditioner i miljöundervisningens
socialisationsinnehåll – implikationer för undervisning för hållbar utveckling. ISBN:
978-91-85485-88-8 (Doctoral Dissertation) Mälardalen University
22. Susanne Engström (2008): Fysiken spelar roll! I undervisning om hållbara
energisystem - fokus på gymnasiekursen Fysik A. ISBN: 978-91-85485-96-3
(Licentiate thesis) Mälardalen University
23. Britt Jakobsson (2008): Learning science through aesthetic experience in elementary
school science. Aesthetic judgement, metaphor and art. ISBN: 978-91-7155-654-7.
(Doctoral Dissertation) Stockholm university
24. Gunilla Gunnarsson (2008): Den laborativa klassrumsverksamhetens interaktioner -
En studie om vilket meningsskapande år 7-elever kan erbjudas i möten med den
laborativa verksamhetens instruktioner, artefakter och språk inom elementär ellära,
samt om lärares didaktiska handlingsmönster i dessa möten. (Doctoral Dissertation)
Linköping University
25. Pernilla Granklint Enochson (2008): Elevernas föreställningar om kroppens organ och
kroppens hälsa utifrån ett skolsammanhang. (Licentiate thesis) Linköping University
26. Maria Åström (2008): Defining Integrated Science Education and putting it to test
(Doctoral Dissertation) Linköping University
27. Niklas Gericke (2009): Science versus School-science. Multiple models in genetics –
The depiction of gene function in upper secondary textbooks and its influence on
students’ understanding. ISBN 978-91-7063-205-1 (Doctoral Dissertation) Karlstad
University
 Studies in Science and Technology Education
ISSN 1652-5051

28. Per Högström (2009): Laborativt arbete i grundskolans senare år - lärares mål och hur
de implementeras. ISBN 978-91-7264-755-8 (Doctoral Dissertation) Umeå University
29. Annette Johnsson (2009): Dialogues on the Net. Power structures in asynchronous
discussions in the context of a web based teacher training course. ISBN 978-91-
977100-9-1 (Doctoral Dissertation) Malmö University
30. Elisabet M. Nilsson (2010): Simulated ”real” worlds: Actions mediated through
computer game play in science education. ISBN 978-91-86295-02-8 (Doctoral
Dissertation) Malmö University
31. Lise-Lotte Österlund (2010): Redox models in chemistry: A depiction of the
conceptions held by upper secondary school students of redox reactions. ISBN 978-
91-7459-053-1 (Doctoral Dissertation) Umeå University
32. Claes Klasander (2010): Talet om tekniska system – förväntningar, traditioner och
skolverkligheter. ISBN 978-91-7393-332-2 (Doctoral Dissertation) Linköping
University
33. Maria Svensson (2011): Att urskilja tekniska system – didaktiska dimensioner i
grundskolan. ISBN 978-91-7393-250-9 (Doctoral Dissertation) Linköping University
34. Nina Christenson (2011): Knowledge, Value and Personal experience – Upper
secondary students’ use of supporting reasons when arguing socioscientific issues.
ISBN 978-91-7063-340-9 (Licentiate thesis) Karlstad University
35. Tor Nilsson (2011): Kemistudenters föreställningar om entalpi och relaterade begrepp.
ISBN 978-91-7485-002-4 (Doctoral Dissertation) Mälardalen University
36. Kristina Andersson (2011): Lärare för förändring – att synliggöra och utmana
föreställningar om naturvetenskap och genus. ISBN 978-91-7393-222-6 (Doctoral
Dissertation) Linköping University
37. Peter Frejd (2011): Mathematical modelling in upper secondary school in Sweden An
exploratory study. ISBN: 978-91-7393-223-3 (Licentiate thesis) Linköping University
38. Daniel Dufåker (2011): Spectroscopy studies of few particle effects in pyramidal
quantum dots. ISBN 978-91-7393-179-3 (Licentiate thesis) Linköping University
39. Auli Arvola Orlander (2011): Med kroppen som insats: Diskursiva spänningsfält i
biologiundervisningen på högstadiet. ISBN 978-91-7447-258-5 (Doctoral
Dissertation) Stockholm University
40. Karin Stolpe (2011): Att uppmärksamma det väsentliga. Lärares ämnesdidaktiska
förmågor ur ett interaktionskognitivt perspektiv. ISBN 978-91-7393-169-4 (Doctoral
Dissertation) Linköping University
41. Anna-Karin Westman (2011) Samtal om begreppskartor – en väg till ökad förståelse.
ISBN 978-91-86694-43-2 (Licentiate thesis) Mid Sweden University
42. Susanne Engström (2011) Att vördsamt värdesätta eller tryggt trotsa.
Gymnasiefysiken, undervisningstraditioner och fysiklärares olika strategier för
energiundervisning. ISBN 978-91-7485-011-6 (Doctoral Dissertation) Mälardalen
University
43. Lena Adolfsson (2011) Attityder till naturvetenskap. Förändringar av flickors och
pojkars attityder till biologi, fysik och kemi 1995 till 2007. ISBN 978-91-7459-233-7
(Licentiate thesis) Umeå University
 Studies in Science and Technology Education
ISSN 1652-5051

44. Anna Lundberg (2011) Proportionalitetsbegreppet i den svenska gymnasie-

matematiken – en studie om läromedel och nationella prov. ISBN 978-91-7393-132-8
(Licentiate thesis) Linköping University
45. Sanela Mehanovic (2011) The potential and challenges of the use of dynamic software
in upper secondary Mathematics. Students’ and teachers’ work with integrals in
GeoGebra based environments. ISBN 978-91-7393-127-4 (Licentiate thesis)
Linköping University
46. Semir Becevic (2011) Klassrumsbedömning i matematik på gymnasieskolans nivå.
ISBN 978-91-7393-091-8 (Licentiate thesis) Linköping University
47. Veronica Flodin (2011) Epistemisk drift - genbegreppets variationer i några av
forskningens och undervisningens texter i biologi. ISBN 978-91-9795-161-6
(Licentiate thesis) Stockholm University
48. Carola Borg (2011) Utbildning för hållbar utveckling ur ett lärarperspektiv –
Ämnesbundna skillnader i gymnasieskolan. ISBN 978-91-7063-377-5 (Licentiate
thesis) Karlstad University
49. Mats Lundström (2011) Decision-making in health issues: Teenagers’ use of science
and other discourses. ISBN 978-91-86295-15-8 (Doctoral Dissertation) Malmö
University
50. Magnus Oscarsson (2012) Viktigt, men inget för mig. Ungdomars identitetsbygge och
attityd till naturvetenskap. ISBN: 978-91-7519-988-7 (Doctoral Dissertation)
Linköping University
51. Pernilla Granklint Enochson (2012) Om organisation och funktion av människo-
kroppens organsystem – analys av elevsvar från Sverige och Sydafrika.
ISBN 978-91-7519-960-3 (Doctoral Dissertation) Linköping University
52. Mari Stadig Degerman (2012) Att hantera cellmetabolismens komplexitet –
Meningsskapande genom visualisering och metaforer. ISBN 978-01-7519-954-2
(Doctoral Dissertation) Linköping University
53. Anna-Lena Göransson (2012) The Alzheimer Aβ peptide: Identification of Properties
Distinctive for Toxic Prefibrillar Species. ISBN 978-91-7519-930-6 (Licentiate thesis)
Linköping University
54. Madelen Bodin (2012) Computational problem solving in university physics education
- Students’ beliefs, knowledge, and motivation. ISBN 978-91-7459-398-3 (Doctoral
Dissertation) Umeå University
55. Lena Aretorn (2012) Mathematics in the Swedish Upper Secondary School Electricity
Program: A study of teacher knowledge. ISBN 978-91-7459-429-4 (Licentiate thesis)
Umeå University
56. Anders Jidesjö (2012) En problematisering av ungdomars intresse för naturvetenskap
och teknik i skola och samhälle – Innehåll, medierna och utbildningens funktion.
ISBN 978-91-7519-873-6 (Doctoral Dissertation) Linköping University
57. Thomas Lundblad (2012) Simulerad verklighet i gymnasieskolans fysik: en
designstudie om en augmented reality simulering med socio-naturvetenskapligt
innehåll. ISBN 978-91-7519-854-5 (Licentiate thesis) Linköping University
58. Annie-Maj Johansson (2012) Undersökande arbetssätt i NO-undervisningen i
grundskolans tidigare årskurser. ISBN 978-91-7447-552-4 (Doctoral Dissertation)
Stockholm University
 Studies in Science and Technology Education
ISSN 1652-5051

59. Anna Jobér (2012) Social Class in Science Class. ISBN 978-91-86295-31-8 (Doctoral
Dissertation) Malmö University
60. Jesper Haglund (2012) Analogical reasoning in science education – connections to
semantics and scientific modeling in thermodynamics. ISBN 978-91-7519-773-9
(Doctoral Dissertation) Linköping University
61. Fredrik Jeppsson (2012) Adopting a cognitive semantic approach to understand
thermodynamics within science education. ISBN 978-91-7519-765-4 (Doctoral
Dissertation) Linköping University
62. Maria Petersson (2012) Lärares beskrivningar av evolution som undervisningsinnehåll
i biologi på gymnasiet.ISBN 978-91-7063-453-6 (Doctoral Dissertation) Karlstad
University
63. Henrik Carlsson (2012) Undervisningsform, klassrumsnormer och matematiska
förmågor. En studie av ett lokalt undervisningsförsök för elever med intresse och
fallenhet för matematik. ISBN 978-91-86983-89-5 (Licentiate thesis) Linnaeus
University)
64. Anna Bergqvist (2012) Models of Chemical Bonding. Representations Used in School
Textbooks and by Teachers and their Relation to Students’ Understanding. ISBN 978-
91-7063-463-5 (Licentiate thesis) Karlstad University
65. Nina Kilbrink (2013) Lära för framtiden: Transfer i teknisk yrkesutbildning. ISBN
978-91-7063-478-9 (Doctoral Dissertation) Karlstad University
66. Caroline Larsson (2013) Experiencing Molecular Processes. The Role of
Representations for Students’ Conceptual Understanding. ISBN 978-91-7519-607-7
(Doctoral Dissertation) Linköping University
67. Anna-Karin Carstensen (2013) Connect Modelling Learning to Facilitate Linking
Models and the Real World through Labwork in Electric Circuit Courses for
Engineering Students ISBN 978-91-7519-562-9 (Doctoral Dissertation) Linköping
University
68. Konferensproceeding: 10-year Anniversary Meeting with the Scientific Committee
69. Marie Bergholm (2014) Gymnasieelevers kommunikativa strategier i
matematikklassrummet. En fallstudie av ett smågruppsarbete om derivata ISBN 978-
91-7519-306-9 (Licentiate thesis) Linköping University
70. Ingrid Lundh (2014) Undervisa Naturvetenskap genom Inquiry – En studie av två
högstadielärare. ISBN 978-91-7519-285-7 (Licentiate thesis) Linköping University
71. Nils Boman (2014) Personality traits in fish - implications for invasion biology
ISBN:978-91-7601-097-6 (Licentiate thesis) Umeå University
72. Torodd Lunde (2014) När läroplan och tradition möts - lärarfortbildning och syften
med undersökande aktiviteter inom den laborativa NO-undervisningen i grundskolans
senare del. ISBN: 978-91-7063-577-9 (Licentiate thesis) Karlstad University
73. Martin Eriksson (2014) Att ta ställning - gymnasieelevers argumentation och
beslutsfattande om sociovetenskapliga dilemman. ISBN 978-91-7063-588-5
(Licentiate thesis), Karlstad University
74. Annalena Holm (2014) Mathematics Communication within the Frame of
Supplemental Instruction. Identifying Learning Conditions. ISBN 978-91-7623-112-8
(Licentiate thesis) Lund University
 Studies in Science and Technology Education
ISSN 1652-5051

75. Daniel Olsson (2014) Young people’s ‘Sustainability Consciousness’ – Effects of

ESD implementation in Swedish schools.. ISBN 978-91-7063-594-6. (Licentiate
thesis) Karlstad University
76. Marlene Sjöberg (2014) Möjligheter I kollegiala samtal om NO-undervisning och
bedömning. https://gupea.ub.gu.se/handle/2077/24063 (Licentiate thesis) Gothenburg
University.
77. Teresa Berglund (2014) Student ‘Sustainability Consciousness’ and Decision-Making
on Sustainability Dilemmas. Investigating effects of implementing education for
sustainable development in Swedish upper secondary schools. ISBN 978-91-7063-
599-1 (Licentiate thesis) Karlstad University
78. Elisabet Mellroth (2014) High achiever! Always a high achiever? A comparison of
student achievements on mathematical tests with different aims and goals. ISBN 978-
91-7063-607-3 (Licentiate thesis) Karlstad University
 Elisabet Mellroth | High achiever! Always a high achiever? | 2014:67
High achiever! Always a high achiever?
This thesis describes a study based on teacher observations of students who
achieve highly on the international competition ‘the mathematical kangaroo’
although they do not in the national test. The aim with the study was to investigate
students’ relative achievement in mathematics over time and how mathematical
competencies can be used to explore differences between groups of students on
a non-curriculum based test in mathematics. The study was divided in two parts.
Study 1 compared students’ (n=568) relative achievement in two national tests
in mathematics (years 3 and 6), changes in relative achievement between the
two tests as well as differences in relative achievement between the national test
in year 6 and the mathematical kangaroo in year 7 (age 13) were explored. The
study identified two groups of students with high achievements, in only one of
the tests, from a sample (n=264) of study 1. Study 2 explored how differences
between those students’ relative achievement on the mathematical kangaroo
could be explained through activation of mathematical competencies.

ISBN 978-91-7063-607-3

ISSN 1403-8099

LICENTIATE THESIS | Karlstad University Studies | 2014:67