This session

commences
at 6.00 am.
Research Methodology

Lecture 7
 Quantitative Data Analysis
 1. Measures of Frequency
 2. Measures of Central Tendency

 3. Outliers

 4. Measures of Dispersion

 5. Probability

 6. Inferential Statistical Tests – t-tests

 Quantitative Data Analysis
 1. Measures of Frequency
 2. Measures of Central Tendency

 3. Outliers
l et ed
m p
Co of Dispersion
 4. Measures

 5. Probability

 6. Inferential Statistical Tests – t-tests

INFERENTIAL
STATISTICS
Goal of Research

Why Quantitative Data

Analysis?
Can you
contribute?

8
 Inferential Statistics
• Goal of research- not to show statistics
about the investigated group
• But to make inferences to the whole
population
• Inferential statistics allows this
generalization
 Inferential Statistics
• Goal of research- not to show statistics
about the investigated group
• But to make inferences to the whole
population
• Inferential statistics allows this
generalization l i di t y
a l va
r n
Exte
 For Example: Mean Value
Difference

Experimental 68.64
Group

Control
51.34
Group

11
 For Example: Mean Value i ficance?
i gn
ys
Difference
nce st atist i cal l

d i f f ere
val ue
ean
e m
I s t h
Experimental 68.64
Group

Control
51.34 ulation?
Group ol pop
e
e w h
d th
to
pp l i e
t be a t o r?
Can i ef ac
h anc
t o c
i t d ue
Or i s
12
 Different Statistical Tests

Independent
sample t test

Paired sample t
test
Chi square test

Pearson
correlation test

13
 Different Statistical Tests

Independent
sample t test
at a &
Paired sample t s of d
typ e
i ng t o si gn
test Accor search d
d e
re Chi square test

Pearson
correlation test

14
 Learning Outcomes

By the end of this part of the lecture, you will

be able to describe four statistical tests.

15
 Learning Outcomes
h em !
o ut t
on ab
rm at i
i nf o
d m or e
Fi n

By the end of this part of the lecture, you will

be able to describe four statistical tests.

16
Types of Inferential Statistical Tests;
t-test
•If mean values of two groups are
significantly different-statistically different
•Means
•Group A: 56.7,
•Group B: 64.3
•Significant?
•Perform a t-test
 Types of t-tests
• Two types:
• 1. Independent t-tests
• 2. Paired sample t-test
 Your Study
You design new teaching materials for
development of reading skills.
Hypothesis?
Newly designed reading materials increase the
second language proficiency
Variables?
reading materials & test scores
 How to investigate the effect?
• Two ways:
• 1. Two groups of learners (Control &
experimental)
• Experimental group- New materials, Control
group – General teaching
• Test – two sets of scores
• Mean difference between Experimental Group
& Control Group
•Two sets of data
•Mean difference: Control Group- 53.8,
Experimental Group 67.9
•Statistically significant?
•Can be used to generalize?
•Two data sets – independent
•Perform independent t-test
 Two ways to investigate
• 1. Two Groups – Control & Experimental
• What is the other way to investigate?
• 2. One Group only
• Pre-test is conducted & teaching is done & Post test
is conducted
• Two sets of data
• Mean value difference significant?
• Paired sample t test
 Independent Samples t Test
• Compares the means of two
independent groups
• Why? Purpose?
• determine whether there is statistical
evidence that the associated population
means are significantly different.
 Independent Samples t Test
• Independent Samples t Test -
parametric test
• Two variables used in this test?
• Dependent variable: marks
• Independent variable: Groups
 Commonly used to test?
Statistical differences between the means of
two groups
Statistical differences between the means of two
interventions
Compare the "Means"?
For two groups only.
If Means -more than two groups
ANOVA.
 Requirements of data sets
1.Dependent variable - interval or
ratio scale (marks of a test)
2.Independent variable - categorical
(Two groups)
 Requirement 3 - Independent
samples/groups
independence of observations
No relationship between the subjects in
each sample.
Subjects in the first group - cannot also be
in the second group
Subject in one group cannot influence
subjects in the other group
Violation of this assumption - inaccurate p
value
 Requirement 4 , 5 & 6 -
4. Random sample of data from the
population
5. Normal distribution (approximately)
of the dependent variable
•6. No outliers
 If requirements not met?
• Run the nonparametric Mann-
Whitney U Test
• Anyway,
• Each group should have at least 6
subjects
Hypothesis: Independent Samples
t Test
• H0: µ1 = µ2 (two Means - equal)
H1: µ1 ≠ µ2 two Means - not equal)
• In other words?
• H0: µ1 - µ2 = 0 (difference between two
Means - equal to 0)
H1: µ1 - µ2 ≠ 0 (difference between two
Means - not 0")
 Run SPSS
• click?
• Analyze > Compare Means >
Independent-Samples T Test.
 See result sheet
• Sign of the mean difference -Sign of the t
value
• Positive t value - Mean value of Group A
significantly greater than the mean for B
• If p value is printed as ".000"? (SPSS rounds p-
values to three decimal places)
• If p-value too small to round up to .001 - will
print as .000.
 Decision and Conclusions
• Look at p value
• Imagine p < 0.001
• Less than the chosen significance level α
= 0.05,
• we can reject the null hypothesis
• conclude that the that the Mean is
significantly different.
34
Paired Sample t Test
 Paired Sample t Test
• Compares ?
• Means of two measurements taken from the
same individuals
• "paired" measurements – Examples?
• Measurement taken at two different times
• pre-test and post-test scores - intervention
between the two time points
 Purpose - Paired Samples t Test
• whether there is statistical evidence?
• Mean value difference between
paired observations - significantly
different from zero
Paired Samples t Test - commonly
used to
• Statistical difference between two
time points
 Paired Samples t Test
• Only compares the Means for?
• only two related (paired) units
• normally distributed
 Paired Samples t Test – Not
Appropriate for?
• 1. unpaired data
• 2. more than two groups
• 3. not normally distributed
 If Paired Sample t Test Not
Appropriate?
• 1. To compare unpaired Means between two independent
groups ( normally distributed)?
• Independent Samples t Test
• 2. To compare unpaired Means between more than two
groups ( normally distributed)?
• ANOVA
• 3. To compare paired means -not normally distributed?
• Nonparametric Wilcoxon Signed-Ranks Test
 Requirements of Paired Sample t
Test
1. Dependent variable - interval or ratio
scale (test scores)
2. Related samples/groups (i.e., dependent
observations)
The subjects in each sample, - same.
Subjects in Group A = Subjects in Group B
3. Random sample of data from the
population
4. Normal distribution (approximately)
of the difference between the paired
values
5. No outliers
 If requirements not met?
• Run nonparametric Wilcoxon Signed-
Ranks
Hypothesis- Paired Samples t Test
• H0: µ1 = µ2 (paired means - equal)
H1: µ1 ≠ µ2 (paired means - not equal)
• In other words?
• H0: µ1 - µ2 = 0 (difference between paired
means - equal to 0)
H1: µ1 - µ2 ≠ 0 (difference between paired
means – not equal to 0")
 Wording - Hypotheses
• H0 : There is no significant Mean value
difference between the marks of Pre-test
& the marks of Post-test
• H1 : There is a significant Mean value
difference between the marks of Pre-test
& the marks of Post-test
 P-value
• whether statistically large differences - due to
chance or to actual differences.
• P-value of <0.05 ?
• ‘Scientific significance’
• Meaning?
• If an experiment was done 100 times and there
were
truly no difference between compared values,
then a larger difference would be expected in
only about 5 of these 100 repeated experiments.
 t value
• magnitude of the difference between the
two group means
 Now, we discussed
• Two t tests
• Independent sample t test
• Paired sample t test
• Common feature?
• Compare Mean value difference by
groups
51
 Three situations
 Research A: Wants to test whether
selected teaching strategies are effective.
He uses selected strategies to one group
& traditional teaching to the other group
 Researcher B: Wants to see whether
marks of Sinhala language & marks of
English language in the final examination
are related.
 Researcher C: Wants to investigate the
opinions of teachers on the use of
selected teaching strategies
 Researcher B: Wants to see
whether marks of Sinhala language
& marks of English language in the
final examination are related.
 No two groups
 Not t test
 Some other
 Further

Relationships
 1. Relationship between Mean value
difference between two groups?
 t test
 2. Relationship between two
continuous variables (Not by
groups)
 Pearson correlation test
 Non-experimental
research context
 Year End results are released.
 Those who have scored more
marks in Sinhala have also
scored more marks in English.
 You want to find whether there
is a relationship between Marks
of Sinhala & Marks of English.
 Marks of English & Marks of Sinhala
 No Groups
 We want to see whether there is a
relationship
 Not a test. Why?
 No two groups
 Then?
 Pearson correlation test
 Correlation?
 Relationship between two variables
 Example?
 Relationship between?
 Marks of Sinhala & English
 Income & marks of English test
 No of days attended lectures & marks
of Research methodology
 Requirements
 1. Ratio/Scale (Interval) Variables
 2. Independent: Marks of Research
Methodology given by me & marks of thesis
given by another lecturer
 3. Normal Distribution (Each set)
 4. Random Sample
 5. No Outliers
 example analysis

 Research Problem
 Is there a relationship between
students vocabulary marks and
grammar marks?
 Two sets of marks
 Are they related?
 Analysis

Gramma Vocabular
r y
Grammar 1 0.611
Pearson Correlation Sig 0.000
2 tailed 83 78
N

Vocabulary 0.611 1
Pearson Correlation Sg 0.000
2 tailed 78 81
N
 Pearson correlation coefficient
for correlation between
Vocabulary & Grammar is
0.611, which is significant at
p < .001 for a two-tailed test
 conclusion
 Vocabulary & Grammar have a
statistically significant relationship
(r=0.611, p < 0.001).
Vocabulary & Grammar - positively correlated
Meaning?
When Vocabulary increases, Grammar
increases
Strength, of the relationship?
0.611
 Two groups taken by teacher. One
group taught in the language lab. The Pearson
A other one in normal classroom. Now correlation
compare marks

Paired
Marks of English & Marks of Science sample t
B compared by a teacher test

A teacher uses new materials designed

Independent
for teaching Writing skills. She conducts
sample t test
C a Pre-test and after two months of
teaching, a Post-test to compare
 Two groups taken by teacher. One
group taught in the language lab. The Pearson
A other one in normal classroom. Now correlation
compare marks

Paired
Marks of English & Marks of Science sample t
B compared by a teacher test

A teacher uses new materials designed Independen

for teaching Writing skills. She conducts t sample t
C a Pre-test and after two months of test
teaching, a Post-test to compare
Analysis of Survey Data -
Quantitative
You want to investigate whether
students like to learn grammar
 Questionnaire

• Do you like grammar?

• YES

• NO
 Frequency

• Do you like grammar?

• YES 73%

• NO 27%
Another Question in a Questionnaire
 Questionnaire
• Which of the following do you like most?
• Grammar
• Listening
• Reading
• Writing
• Speaking?
 Questionnaire
• Which of the following do you like most?
• Grammar 46%

• Listening 14%

• Reading 23%

• Writing 10%

• Speaking? 7%
 Another Survey Research
•Opinions of effectiveness of grammar
teaching on development of overall
language proficiency
•Opinions of two types of teachers
•1. Teachers of English 2. English medium
Teachers of Science.
•Two data sets
•Not experimental study
• Opinions –Survey
• "Do you recommend teaching
Grammar?"
• Not – Independent sample t test
• How do you analyze?
• "Do you recommend teaching
Grammar?"
• YES

• NO
 English Teachers
• "Do you recommend teaching
Grammar?"
• YES 63%

• NO 37%
 English medium science Teachers
• "Do you recommend teaching
Grammar?"
• YES 27%

• NO 73%
 "Do you recommend
teaching Grammar?"
• Is there a relationship between how English
Teachers & English Medium Science Teachers
respond?
• Example
• Science Teachers – NO But English teachers –
YES
• Is there a relationship?
 When chi-square?
Statistical relationship between two categorical
variables
Only?
Categorical variables
Example?
Gender, Economic state, Type of Teachers
(English/ English medium)
Not?
Continuous variables (marks of a test)
 "pre-test" and "post-test"?
• chi-square test?
• Not
•
 Example situation
• Survey
• "Do you recommend Teaching
Grammar?"
• Two groups?
• English teachers & Non-English teachers
 Requirements
1. Two categorical variables
2. Independence of observations
No relationship between the subjects in
each group
Example?
English teachers' opinions & English
medium teachers' opinions
3. Large sample size
 Hypothesis
• H0: "[Variable 1] is independent of
[Variable 2]"
• H1: "[Variable 1] is not independent of
[Variable 2]"
• OR
• H0: "[Variable 1] is not related to
[Variable 2]"
• H1: "[Variable 1] is related to [Variable 2]"
 Hypothesis
• H0: "[Variable 1] is independent of [Variable
2]"
• H0: "[Variable 1; teacher category] is
independent of [Variable 2; opinion of
teachers]"
• H0: There is no relationship between the
opinions of teachers
•
 Hypothesis
• H1: "[Variable 1] is related to [Variable 2]"
• H1: "[Variable 1; teacher category] is
related to [Variable 2; teacher opinion]"
• H1: There is a relationship between the
teacher category and teacher opinion.
Reg No Teacher Grammar
Category (Yes/No)
2019/023 ENG YES
2019/034 ENG NO
2019/041 ENG MEDIUM YES
2019/045 ENG MEDIUM NO
2019/056 ENG NO
2019/071 ENG MEDIUM YES
• Respondents were asked?
• Their teaching category (English teacher
or English medium teacher) and
• whether or not they think grammar
teaching is important
• We want
• a relationship?
• Between teacher category & opinion
• How to decide a relationship between
two categorical variables?
• Chi square test
• BUT
• Before that?
• Bar chart
 Category 1- YES
Category 2 - NO
Teacher Opinions on Teaching Grammar

4.5

3.5

2.5

1.5

0.5

0
Category 1 Category 2
 Relationship unlikely
Category 1- YES
Category 2 - NO
Teacher Opinions on Teaching Grammar

4.5

3.5

2.5

1.5

0.5

0
Category 1 Category 2
• If you perform chi square test?
• The corresponding p-value?
• p = 0.205
• Decision?
• p-value is greater than our chosen
significance level (α = 0.05)
• We cannot reject the null hypothesis
• Conclusion?
• Not enough evidence to suggest a
relationship between teacher category &
opinion
•
 Category 1- English Tr
Category 2 – Science TR
 Relationship likely
Category 1- English Tr
Category 2 – Science TR
• If you perform a Chi Square?
 Conclusion

• p-value is less than α = 0.05

• We can reject the null
hypothesis
• That means?
• There is a relationship between
teacher category and opinion on
grammar teaching
 Different Statistical Tests

Independent
sample t test

Paired sample t
test
Chi square test

Pearson
correlation test

101
Various Statistical Tests are
performed to identify
statistical significance of the
data
 ear ch
to r es
or di ng t
Acc co n t ex

Various Statistical Tests are

performed to identify
statistical significance of the
data
Read more
& enrich
your
knowledge!