Transcript Analyzing Data

Independent Studies
Resource 3: Qualitative and
Quantitative Analysis
Dr Jill Hanson
N509
[email protected]
Aims




To look at different types of qualitative
analysis
To look at thematic analysis in more depth
To introduce you to descriptive statistics
To enable you to choose the correct
inferential statistical test
QualitativeAnalysis

What is it?

It is the analysis of words or actions measured
through






Interview transcripts
Field notes (notes taken in the field being studied)
Video
Audio recordings
Images
Documents (reports, meeting minutes, e-mails)
What Is It’s Goal?


Qualitative Data Analysis (QDA) is the range of
processes and procedures whereby we move from
the qualitative data that have been collected into
some form of explanation, understanding or
interpretation of the people and situations we are
investigating.
The process of QDA usually involves two things,
writing and the identification of themes.
Writing



Involves writing about the data and what you
have found.
Often what you write may be analytic ideas.
In other cases it may be some form of précis
or summary of the data, though this usually
contains some analytic ideas.
Coding into themes


Looking for themes involves coding.
This enables researchers to retrieve and
collect together all the text and other data
that they have associated with some thematic
idea
Example
“When you move into your own home, you're alone. There is no
bustle of people around the house. I miss having someone to
chat to when I get home. I put the TV or some music so
there’s some background noise, the silence makes me feel
so alone. Sometimes I will be sat watching trash TV and
thinking I should be out doing something rather than
watching this rubbish. I read a lot but sometimes I am too
tired and just want to veg out. But it's been good to move out
of mum and dad’s as it's not healthy to rely on them as they
won't last forever. I become independent and made my own
decisions. It's good they still there when I need them. It's
good to have some distance as when I was at home I was
arguing a lot with my dad and that was what made me decide
it was time to go.”
Interpreting


It is easy to write and code in ways that are
nothing more than descriptive summaries of
what participants have said or done.
But you need to trick is to move towards
explaining why things are as you have found
them.
Organising
Researchers tend to approach this organisation in one of two
ways.
1. Manual methods
 Notes and interviews are transcribed and transcripts and images
etc. are copied. The researcher then uses folders, filing cabinets,
wallets etc. to gather together materials that are examples of
similar themes or analytic ideas.
2. Computer based
 Many analysts now also use dedicated computer assisted
qualitative data analysis (CAQDAS) packages. Can be tricky to
learn and no point unless your sample size is large

Aspects of QA that you should consider








Are you interested in interpreting the data in terms of themes / concepts / ideas /
interactions / processes? Then YOU have to do the thinking, the analysis. There
is no software that can actually do the thinking for you
Data may be messy
You need to give thought to efficient data management.
You need to find out what literature there is around your research topics.
Qualitative data usually cannot be reduced to numbers.
If you ARE just trying to reduce the data to numbers, have you properly
understood the reasons for doing qualitative research?
Will the sample size and/or sampling method be telling you anything of value at
all? (Many qualitative samples are small and not proper random samples).
If you are generating numbers then you should see the numbers only as
pointers to more thinking and researching about where and why there are
anomalies or exceptions. This may mean more data collection, more thinking,
more testing
Approaches to analysing QD









Action research
Field Research
Memory work
Analytic
InductionFramework
analysis
Mixed methods
Biographical research
Grounded theory
Narrative analysis
Ethnomethodology Matrix
Analysis/Logical Analysis







ConversationbAnalysis
Phenomenography
Comparative analysis
Interpretative
Phenomenological
Analysis (IPA)
Phenomenology
Discourse analysis
Life History QCA Qualitative Comparative
Analysis
Ethnography Life-world
analysis
Symbolic interactionism
Thematic Analysis


We will only look at a simple Thematic
Analysis which is likely to be appropriate for
the majority of you given your experience and
the nature of the data you will be collecting.
See Braun & Clarke (2006) for extensive
detail
Braun & Clarke (2006)


“Thematic analysis is a method for
identifying, analysing and reporting patterns
(themes) within data” (p.79)
It differs from other analytic methods such as
IPA and grounded theory because it does not
require the researcher to subscribe to the
implicit theoretical commitments of other
approaches
What is a theme?


“A theme captures something important
about the data in relation to the research
question, and represents some level of
patterned response or meaning within the
data set” (Braun & Clarke, 2006, p. 82)
Will be evidenced in a number of participants
responses
Focus?

You can focus on:



A rich description of themes emerging from the
ENTIRE data set
OR
Provide a detailed account of one particular
aspect (in relation to a particular question)
Inductive versus
theoretical thematic analysis

Themes can be identified in





Bottom up way (inductive)
Top down way (theoretical and deductive)
Inductive – themes are strongly linked to data
itself, may not bear a strong relation to the actual
questions asked
Deductive – themes are driven by the
researchers theoretical or analytic interest in the
area
Choice usually maps onto how and why you are
coding the data


Code for a specific research question (maps onto
theoretical approach)
Or specific research question can evolve through the
coding process (maps onto the inductive approach)
Semantic or Latent Themes




Semantic approach, the themes are identified within the explicit or
surface meanings of the data, and the analyst is not looking for
anything beyond what a participant has said or what has been
written.
In contrast, a thematic analysis at the latent level goes beyond the
semantic content of the data, and starts to identify or examine the
underlying ideas, assumptions, and conceptualizations / and
ideologies /that are theorized as shaping or informing the semantic
content of the data.
If we imagine our data three-dimensionally as an uneven blob of
jelly, the semantic approach would seek to describe the surface of
the jelly, its form and meaning, while the latent approach would
seek to identify the features that gave it that particular form and
meaning.
Thus, for latent thematic analysis, the development of the themes
themselves involves interpretative work, and the analysis that is
produced is not just description, but is already theorized
How to conduct and write up Thematic
Analysis
1.
2.
3.
4.
5.
6.
7.
Make sure you explicitly state the theoretical
position you are taking in your write up.
Familiarise yourself with your data
Generate initial codes
Search for themes
Review themes
Define and name themes
Produce report
Phase
Description of process
familiarise yourself with the data
Transcribing data (if necessary), reading and re-reading the
data, noting down initial ideas.
generate initial codes
Coding interesting features of the data in a systematic
fashion across the entire data set, collating data relevant to
each code.
search for themes
Collating codes into potential themes, gathering all data
relevant to each potential theme.
review themes
Checking if the themes work in relation to the coded
extracts (Level 1) and the entire data set (Level 2),
generating a thematic ‘map’ of the analysis.
define and name themes
On going analysis to refine the specifics of each theme, and
the overall story the analysis tells, generating clear
definitions and names for each theme.
Produce report
The final opportunity for analysis. Selection of vivid,
compelling extract
examples, final analysis of selected extracts, relating back
of the analysis to the
research question and literature, producing a scholarly
report of the analysis
Pitfalls
1.
2.
3.
4.
Failure to actually analyse the data – you must
include analytic narrative as well as extracts
and these must be directly relevant to your
objective
Using the data collection questions as the
themes (here no analysis has been done)
Weak or unconvincing analysis – themes do
not work, there is too much overlap, themes
are not internally coherent/consistent
Mismatch between data and claims made –
claims are not supported by data and report
does not consider alternative interpretations of
data
Quantitative Analysis: NUMBERS

When?


If you have used structured interviews or
questionnaires
Your data takes the form of numbers, ranks or
categorical responses
What is Quantitative Analysis

1. Descriptive statistics – used to describe
what your data looks like. NOT ENOUGH AT
MASTERS LEVEL

2. Inferential statistics

Use these to confirm differences between groups
or relationships between variables. YOU MUST
CONDUCT THESE AT MASTERS LEVEL
Types of Variables/Data Forms
Nominal/categorical/group

Where your data takes the form of groups – you
have asked the respondent to check a box
•

e.g. gender, job type
Ratio/Interval/Scale/numeric

Where your data is a number


e.g. Likert scale ratings, age in specific years, weight,
height
Ordinal/ranked


Where you have asked respondents to rank in order
of preference e.g. 1st, 2nd, 3rd
ways to describe data

Measures of central tendency


Measures of dispersion


Mean, mode, median
SD, variance, skewness, kurtosis
You should always report both measures but
careful to choose the statistic that is correct
for YOUR form of data
Descriptive Statistics
Measure for
numeric data
Non-mean
based measure
Center
Mean
Mode, median
Spread
Variance
(standard
deviation)
Range,
Interquartile
range
Skew
Skewness
--
Peaked
Kurtosis
--
Mean
n
x
 i
i 1
n
X
AAARRRGGGGHHHH! All that says is add up all the
scores in the sample and divide by the number of cases!
i.e. work out the average….
Variance, Standard Deviation
( xi   )
2
 ,

n
i 1
2
n
( xi   )


n
i 1
n
2
Again, ignore the silly sum.
Variance means looking at
the way the scores in your
sample vary around the
mean. The standard
deviation is a sum which
calculates that. The bigger
the standard deviation, the
further away your scores
are from the mean. What
does that mean?
The normal curve for numeric data
34% 34%
47%
49%
47%
49%
The z-score or the
“standardized score”
z 
x x
x
A calculation you perform when you want your scores to be
normalised and put into a format where they can be
compared to scores that have been calculated using different
scales. A z score distribution has a mean of 0 and a standard
deviation of 1
Skewness
Symmetrical distribution


IQ
SAT
Frequency



Value
“No skew”
“Zero skew”
Symmetrical
Skewness
Asymmetrical distribution

Frequency

Value
“Negative skew”
“Left skew”
Skewness
(Asymmetrical distribution)

Frequency

Value
“Positive skew”
“Right skew”
Kurtosis
k>3
leptokurtic
Frequency
k=3
k<3
Value
mesokurtic
platykurtic
How do you present descriptive statistics?




Tables
Graphs:
 Line
 Histogram
 Bar
 Scatterplots
 Box plots
Make sure you use tables and graphs only to ILLUSTRATE
WHAT YOU HAVE WRITTEN
Always give them a Table or Figure number and a title
Three words about pie charts:
don’t use them
So, what’s wrong with them

Hard to get a comparison among groups the eye is very bad in judging relative size of
circle slices
example
The worst graph ever published
Conventions for using graphs and tables
1.
2.
3.
4.
ALWAYS give the graph or table a clear title
and table or figure number
Make sure the axes in graphs are clearly
labelled
Don’t make them over complicated
Never use them as analyses in their own
right. They are there to illustrate what you
say in your write up only.
Inferential statistics




Descriptive statistics and the graphs you can
draw using them will often suggest that there are
meaningful differences or relationships in your
data.
However, how can you tell if those differences
are real or if the relationship really exists?
Perhaps it is just an artefact.
Inferential tests are used to confirm that what you
think you see is real, or not.
At UG and PG level it is not enough to just use
descriptive statistics
What can you do with inferential statistics?




Identify significant relationships between
variables (e.g. see if age is related to weight)
Compare groups for significant differences (e.g.
compare males and females on commitment…)
Identify groups of similar cases (e.g. can you
cluster people based on their personality and
IQ?)
Identify groups of similar variables (e.g. do the
items in a scale only represent one factor or do
they actually represent two or three?)
Deciding which test to use
How is your objective worded?
What kind of data have you collected?
How many variables do you have?
How many samples do you have?
Does your data meet the test requirements?
1.
2.
3.
4.
5.
•
•
Parametric tests require your data to meet certain
requirements (e.g. normally distributed, numeric data)
Non-parametric tests do not. (There is always a nonparametric equivalent)
Identifying Significant Relationships
Type of Data
Statistic
Numeric/Scale/Continuous
(and normally distributed)
Pearson’s Correlation
Ordinal / Ranked (or nonnormal continuous)
Spearman’s Rho
Categorical/nominal/grouped
Phi, Cramer’s V
Regression


An extension of correlation where you work
out the extent to which one or more variables
can predict changes in another variable
Not too difficult if you use the right book!
Comparing Groups for Differences



T-tests (where you have parametric data, 1 sample,
or two groups)
ANOVA (where you have parametric data and you
want to compare more than 2 groups)
Non-parametric equivalents of the above (when your
data does not meet requirements)



Mann-whitney/wilcoxon/kruskal-willis/friedman’s ANOVA
Chi-square (for categorical data)
See your SPSS guide for more information
Identifying similar cases/variables

Similar cases – cluster analysis

Similar variables – Factor analysis
Most likely for you?





Correlations
Maybe regression if you are model testing
T-tests or ANOVA (to compare groups on a
numeric/scale/interval variable e.g. do females have a higher
verbal IQ than men?)
Chi square (to compare groups on a nominal/categorical variable
e.g. do more men than women smoke?)
Good news is that these are all easy to run using SPSS and easy
to interpret using Pallant’s fab book!