Assignment: What are data?

In this assignment, you will practice with who (units) and what (variables) we study in social
research
Read this assignment carefully and answer all questions.
Hand in the assignment via Canvas.

1. In social science research, variables characterize units. In the next sentences, please indicate
the unit (of analysis) and the variable(s):

Research topics Units Variable(s) plus

attributes/values
A sometimes-studied topic is attachment Children the behaviour of a child
of children. In this research the behaviour (unspecified),
of a child is observed when the parent the parent walking away (or
walks away. not)
In the Stanford prison experiment, Subjects (individuals) Assigned power (roles),
researchers investigated the effect of Behaviour (unspecified)
assigned power (roles) on aspects of
behaviour.

In the criticized Milgram’s Experiments, Individuals Level of obedience (low –

the level of obedience was studied high)
among individuals.

An experiment conducted by Solomon Individuals Conform to the behaviour of

Asch showed that most individuals a group (yes or no)
conform to the behaviour of a group.

In his early experiments, Ivan Pavlov Dogs Levels of salivary secretions,

discovered that dogs produce different Types of foods.
levels of salivary secretions in response to
different foods.

In the research of the Scandinavian Countries Types of welfare state

sociologist Asping-Anderson, it was (unspecified).
argued that countries can have three
different types of welfare states.

The Dutch railways keep track of train Train journeys Time (delays in minutes
delays. Each journey of a train is (probably))
observed. The delay is recorded and used
for improving services.
Seligman and his colleagues discovered Dogs Being conditioned to fear a
the phenomenon of learned helplessness tone (or not),
when dogs that had been conditioned to Trying to escape (or not)
fear a tone did not try to escape a shock
after hearing the tone, while dogs that
had not been conditioned to fear the
tone did.

1
 2. Units of observation can, but do not have to be, identical to the units of analysis. The units of
analysis are the units we are interested in. We use the units of observation to collect and
store data. Indicate the units of analysis and the units of observation for the following
examples. They may be the same.

Example Unit of analysis Unit of observation

A researcher wants to investigate the general attitude Individuals living Individuals living in
towards immigrants in Enschede. Therefore, she in Enschede Enschede
randomly selects citizens of Enschede to participate in a
survey.

A researcher wants to study if certain company Companies HR managers

characteristics can explain why some companies have a
higher percentage of employees on sick leave than other
companies. For this, the researcher holds a survey among
HR managers of a sample of companies.

A researcher wants to compare the quality of study Study programs Individual students
programs in The Netherlands. She therefore collects data (variable 1),
from students, program managers and from colleagues managers (variable
teaching in similar programs. These data are used to give 2) and colleagues
a grade for the study programs she is studying. (variable 3)

In a panel study about COVID 19, individuals (18-75) are Individuals (18- There are (at least)
asked about their opinions regarding measures reducing 75) two different ways
the spread of COVID 19. They are asked the same to UoO can be
questions at three points in time: March, August and conceptualized:
October 2020. 1. Individuals
2. Individuals x
time
In the first, each
individual in ONE
UoO and all variables
have a time
indication too
(opinion in March,
for example).
In the second, each
individual is THREE
units of observation
(March, August and
October). In that
case variables do
NOT have a time

2
 indication.

3. Failing to distinguish between units of analysis and the units of observation can lead to
problems. In general, inference should not be made to lower levels of aggregation. Doing
that is referred to as the ecological fallacy. For the following three statements about
aggregate data, formulate an alternative conclusion about the individual level that shows
that the original conclusion might be invalid.

Statement or observation Conclusion or Alternative interpretation

interpretation
In cities with higher percentages of Students drink more beer Students are causing nuisance.
student residents, more beer per than non-students. The non-studying residents have
capita is consumed than in cities with to drink more beer, just to be
lower percentages of student able to sleep at night, when the
residents. students roam the cities.
In cities with a higher percentage of Migrants vote for populist The more migrants in a city, the
immigrants, support for populist parties more the others vote for
parties is higher than in cities with a populist parties
low percentage of immigrants.

(US 1900s) In cities with a high Immigrants are more Immigrants move to cities with
percentage of immigrants, the level of literate high levels of economic growth
literacy is higher. and these are generally cities
with a highly literate population

Researchers found that death rates Individual fat You do not know this. It may be
from breast cancer were significantly consumption is that a third variable (economic
increased in countries where fat associated with breast prosperity) reduces fat
consumption was high when cancer consumption AND death rates
compared with countries where fat because of breast cancer. Thus
consumption was low. fat consumption and death
because of breast cancer are
not related on an individual
level.
Note: the interpretations in this exercise are written for educational purposes. It does not mean that
they are necessarily true. For example: we do not think students cause so much nuisance that others

have to drink beer 😉

4. The values or attributes of variables have to be exhaustive (a.k.a. complete) and mutually
exclusive. Indicate for the following examples whether they fulfil these requirements. If an
aspect is missing, make clear how you would do that.

Variable Exhaustive Mutually What would have

exclusive been better?
Age, measured in years Yes Yes

Level of happiness, measured on a Yes Yes

scale from 1 (not happy at all) to 7

3
(very happy)
Religion; protestant, catholic, Yes (other) No, Protestants Religion; protestant,
Christian, other, none and Catholics are catholic, other, none
Christians too.
Level of education you followed: Yes No, it is possible Highest level of
that you’ve education, measured
primary education, lower vocational, followed several in the following
secondary education, higher categories: primary
education, other education, secondary
education, higher
education, other
Weight, No, < 3 kg is No, Weight, measures in
missing 50 kg in two the categories: 0-50
3-50 kg, categories kg, 51-100 kg, 101 -
50-100 kg, + > 150 and > 100 150 kg, >150 kg
>100 kg, overlap
>150 kg

5. We have distinguished five different measurement levels: dichotomous, nominal, ordinal,

interval and ratio. Indicate the measurement level for the following variables.

Variable Level of measurement

Marital Status Nominal (no order in married /widowed / divorced / single)
Favourite book Nominal (many possible values)
Temperature in Celsius Interval (despite the 0 degrees Celsius)
Satisfaction, measured on a Ordinal (although, often used as an interval by researchers)
scale from 1-7
Having a driver’s license Dichotomous
Student number Nominal
Height, in centimetres Ratio

6. Sometimes we cannot assign a value / attribute to an individual, although the set of

values/attributes of the variable we are using is exhaustive and the elements of that set are
mutually exclusive. Can you think of a reason why?

Because sometimes data are missing (due to non-response or a filter question for example).
Think of the variable: which party did you vote for (after the filter question: did you cast a
valid vote in the most recent election).

7. Let us assume that we are studying residents of the UTwente campus. The collected data is
stored in a data matrix. The following table is the codebook that belongs to the data file. A
codebook contains all kinds of ‘meta data’ about the data: which data were collected, the
meaning of the variables, etc. Can you add the measurement level for each of the variables in
the codebook below.

Variable name Question Values Measurement

4
 level
Student Are you a student? Yes / Dichotomous
No
Employee Are you an employee? Yes / Dichotomous
No
Age What is your age in years? 0 - 76 Ratio

DurationofResidence How long have you live on campus in total? 0 - 30 Ratio

Please give your answer in years and round
of to the nearest whole number.
ApprovalRate How do you rate your campus living 1-7 Interval
experience on a scale of 1 (terrible) to 7
(excellent)?

8. Given the codebook above, can you provide an example of what a dataset might look like. Fill
each cell of the matrix below.

Participant Student Employee Age DurationofResidence ApprovalRate

1 Yes Yes 22 4 6
2 Yes No 18 0 2
3 No No 76 30 7
4 No No 1 1 99
5 No Yes 54 5 5

Make sure you add the variables in the top row. Each following row contains the data on one
individual. You can add one variable that has an identifier for each participant. Note that the
code -99- is used for a missing variable.

9. In the datamatrix above, what do we find in the columns?

In each column we find a variable and all observations done on that variable.

10. In the datamatrix above, what do we find in the rows?

In each row we find information about one campus resident.

11. In the datamatrix above, what do we find in the cells?

In the cells we find values that we observe for each variable and each unit (campus residents). In
other words, we can read facts about the campus inhabitants that we observed in the cells.

12. Below we have added some excerpts of information from a (simulated) study about student
experiences with different types of student housing in Enschede. Can you use this
information to construct a codebook? What other information do you need to complete the
codebook?

Questions from a questionnaire given to campus residents.

5
  Do you live in shared living or independent living?
 How many roommates do you have?
 If you focus on the past month, how do you rate your experience of living here?

Questions from a questionnaire given to city residents.

 Do you live in shared living or independent living?

 How many roommates do you have?
 If you focus on the past month, how do you rate your experience of living here?

Dataset

ID Campus Shared Roommates CampusRating CityRating Age

1 0 Yes 2 17 20
2 0 No 4 17
3 1 Yes 7 7 21
4 1 Yes 9 18 19
5 1 No 12 18
6 1 No 5 18
7 0 Yes 3 9 18

Variable name Question/Description Values Measurement

level
ID Participant ID number (not relevant,
but technically
nominal)
Campus Lives on campus versus lives in the 0 = city Dichotomous
city 1 = campus
Shared Do you live in shared living or Yes/ No Dichotomous
independent living?

Roommates How many roommates do you Ratio

have?

CampusRating If you focus on the past month, 1 – 20? Interval

how do you rate your experience
of living here?

CityRating If you focus on the past month, 1 – 20? Interval

how do you rate your experience
of living here?

Age Participant age In years Ratio

6
You can guess that the rating of housing experiences was done on a scale of 0 to 20 and that higher
numbers represent better ratings. However, based on the material here, you cannot be certain. It
would be important to look at the original questionnaire material. For example, you may need to
access the survey environment online.

<< End of the assignment >>