A guide for Data Analysis for Graduate Studies

1. ANALYSING DATA

2. Key issues in qualitative data analysis
 How should I prepare my qualitative data for analysis?
 What are the main strategies for analysing qualitative
data?
 How can I identify concepts and conceptual frameworks
in my data?
 What qualities should I aim for in my analysis?

3. Managing your qualitative data
Qualitative data presents a number of
challenges compared to quantitative data:
– It is usually ‘raw’ – that is, not processed or
transformed into a format useful for analysis
– It will take many different forms
– It can be voluminous,
… and yet still must be traceable, reliable and
complete.

4. Ensuring traceability

5. Student experience example
Our example of 4 written commentaries by 2nd year students on their experience
of the first year is relatively straight forward:
Gender Living situation Code
Female Home FH
Female University FU
Male Home MH
Male University MU
This coding takes place before data collection and is vital to subsequent
analysis. The codes can be used in the analysis to identify patterns and
relationships within the data.

6. Extracting concepts from
your data
 Open coding/indexing – identify concepts
 Classification – grouping the concepts together into categories
 Conceptual framework - proposing relationships between
categories
Techniques include mind-maps, influence diagrams and logic
diagrams to name but a few, but the researcher is free to use
whatever approach they think is suitable.

7. Themes in the example
ACADEMIC
Difference to
school
Teaching
styles/lecturers
Organisation
Resources and
support
SOCIAL
Making
friends/integrating
Partying
Personal
development
MAIN THEMES
SUB-THEMES

8. Quality of your analysis

9. ANLYSING QUANTITATIVE DATA

10.  How can I record and manage
quantitative data?
 How can I describe my quantitative
data using statistics?
So I have my data-what now?

11.  Decide how to record and tabulate data
 By hand
 In a word-processor
 In a srpeadsheet
 Uusing statistical software e.g. spss, Excel.
 Coding data-representing data as a numerical code or other
“shorthand”-e.g. Female =1, Male =2.
 Cleaning data means finding and correcting errors in the
data
Coding, cleaning and preparing
data

12. Entering data: Example
Gender
Male Female
Age group
40 and under Over 40
To what extent do you agree/disagree with the following
“Autocratic management is the most effective”
Strongly agree Agree Disagree Strongly disagree
4 3 2 1

13. Frequency counts
Cross-tabular tables (Cross-tabs)
T-tests and Chi-squared test
Distribution and central tendency
Correlation
Basic descriptive statistics
The approaches you take to analysing your data will depend on the type
of data you collect which in turn is determined by question you set in the
survey. Clearly you can use many types of analysis if you have several
different types of data from you survey.

14. Frequency count
This technique simply counts the number of occurrences in a category
and presents the results in a table-another popular way to represent this type
of data is through a pie chart. When presenting results a good rule of thumb
is to be consistent with the types of tables/graphs/charts you use.
The table below is for data from the example.
Age Under 40 40 +
Total 64 52 116
Gender Male Female
Total 69 47 116

15. Cross tabulating frequencies
A cross tabulation essentially brings together the frequencies.
Male 41 28
Female 23 24
Under 40 Over 40
69
47
Total
These background statistics would normally appear early in the
analysis and become important when analysing later data from
questions for example comparing how males and females respond or
how age impacts on response to a question.

16. Presenting question responses
Total
Male 8 27 25 9 69
Female 2 10 24 11 47
Total 10 37 49 20 116
Strongly
agree
Agree disagree
strongly
disagree Total
Male 12% 39% 36% 13% 69
Female 4% 21% 51% 23% 47
Total 10 37 49 20 116
disagree
strongly
disagree
Strongly agree Agree
The table bottom left presents the results of the responses to the question about management style
preference. It shows how many people responded in each category. Bottom right shows the same data
but represented as percentages. This is more accurate because note the “disagree” category-on the
left it looks like more males responded whereas On the right we can see actually a greater proportion
of females responded.

17. T-tests and chi-square tests
Both these tests look at variation in data sets.
T-tests look at the differences in means for two sets of data. For example
the difference in average student performance in an assignment for males
and females. A t-test will identify whether the difference is statistically
Significant-i.e. whether it is caused by chance.
Chi-square tests look at the difference between actual outcomes and expected
outcomes. This test is particularly useful for analysing the difference in
responses to a question. In our question example about attitudes to autocratic
management we could test whether the differences between male and female
are statistically significant.
N.b. We could look at the difference between gender and age but this would
reduce the sample sizes and render results less reliable. Remember with
quantitative data the bigger the sample size the better.

18. Preparing to use chi-square
We use chi-square to test whether the actual outcomes (in our example the
responses to the autocratic management question) differ significantly from
what might be expected. The first step is to calculate the expected values-
Total
Male 8 27 25 9 69
Female 2 10 24 11 47
Total 10 37 49 20 116
Strongly
agree
Agree disagree
strongly
disagree
Total
Male 5.9 22.0 29.1 11.9 69
Female 4.1 15.0 19.9 8.1 47
Total 10 37 49 20 116
Agree disagree
strongly
disagree
Strongly
agree
Total
Male a x e/g b x e/g c x e/g d x e/g e
Female a x f/g b x f/g c x f/g d x f/g f
Total a b c d g
Strongly
agree
Agree disagree
strongly
disagree
We need to calculate what the expected values
are for these 8 cells
The table below indicates how expected
values are calculated
The table below shows the results of the
calculations for each of the 8 cells.

19. Preparing to use chi-square
Actual observed
Total
Male 8 27 25 9 69
Female 2 10 24 11 47
Total 10 37 49 20 116
Expected
Total
Male 5.9 22.0 29.1 11.9 69
Female 4.1 15.0 19.9 8.1 47
Total 10 37 49 20 116
Strongly
agree
Agree disagree
strongly
disagree
Agree disagree
strongly
disagree
Strongly
agree
Normally we will have a hypothesis which is another way of saying we believe something to
be true-in this case we believe that there is no difference between the responses of males
and females. This is called the null hypothesis and is stated as follows:
H0-There is no difference between males and females in their preference for autocratic
management
Of course if our null hypothesis is proven to be false then it holds that the opposite is the
case, i.e. that there is a difference between males and females
A chi-square test is carried out on the data
In these tables. The result is a value known
as the p value which is a measurement of the
likelihood that any difference occurred by
chance. If the result of the chi-square test is
not above 0.05 it means that the difference
between males and females is significant
(There is a less than 5% chance the difference
was caused by chance)
The p value in this case is 0.051-very close,
but not considered to be significant. The null
hypothesis is proven. Try the test on the data file
In Unilearn.

20. Measures of central tendency
 Measures of central tendency
provide information about the
shape of your data, in
particular the central point
 This chart shows that the
central point for this distribution
is 5 women on the corporate
board, with 120/500
companies reporting this
number
0
20
40
60
80
100
120
Number
of
boards
1 2 3 4 5 6 7 8 9 10
Number of women on
corporate board

21. Correlation
Positively correlated
0
20
40
60
80
100
0 20 40 60 80 100 120
Variable 1
Variable
2
Uncorrelated
0
20
40
60
80
100
120
0 20 40 60 80 100 120
Variable 1
Variable
2
Negatively correlated
0
20
40
60
80
100
120
0 20 40 60 80 100 120
Variable 1
Variable
2
Perfectly correlated
0
20
40
60
80
100
120
0 20 40 60 80 100 120
Variable 1
Variable
2
Correlation explores the relationship between variables. For instance there is a positive
correlation between class attendance and student grade. There is a negative correlation
between the amount of alcohol you drink and the ability to speak cohesively. But alcohol
consumption and the probability of having a fight is positively correlated.