Qualitative Data Analysis Ch 9 13 Mr Jitipol

9. Matrix Displays: Some Rules of Thumps
A. Building matrix displays
Matrix elements - there are aspects that you can make choices during the design process. We are considering how to partition the data.
1. Descriptive VS explanatory detail
2. Partially order VS well-ordered
3. Time-ordered VS not
4. Categories VS variable
5. two-way, three-way, or N-way
6. Cell entries
7. Single case VS multiple case data
As we can see, it is not an easy choice to build a matrix displays. Consequently, you can write matrix roughly, get a colleague to look at your format, set up a revision matrix, don’t include more than dozen variables, think about your next data, expected transpose, and stay open mind.
B. Entering the Matrix Data
1. Be clear about your level of data that you want to enter.
2. Remember that your matrix is displaying only very small percentage of data.
3. Use codes to locate key material.
4. Keep explicit record of the decision rules.
5. The decision rules should be explained clearly.
6. Show the missing data.
7. Don’t lock up your format
8. If you use number, keep words with the number.
9. Order the matrix.
10. Get colleague to review.
C. Drawing conclusion form matrix data
The traditional of presenting basic data is deeply ingrained in reports of qualitative data analysis, we must think carefully about the data the reader will need and data need to be checked and verified always.

10. Making Good Scenes
This chapter shows 13 tactics drawing and verifying conclusion.
1. Noting patterns, themes
We can expect patterns of variables involving similarities and differences among categories, and patterns of processes involving connections in time and space with a context.
2. Seeing plausibility
During documentation of our analysis efforts, we often found ourselves giving “plausibility” basis for conclusions we draws. Plausibility in this sense was a sort of pointer, drawing the analyst’s attention to a conclusion that looked reasonable and sensible on the fact of it.
3. Clustering
Sorts data into groups, such that the degree of natural association is high among members of the same group and low between members of different groups.” In other words, a cluster analysis is a statistical method for grouping “like” things together.
4. Making metaphors (คำอุปมา)
Try on various metaphors and see how well they fit what is observed. Can also ask participant for metaphors and listen for spontaneous metaphors. "Hallway as a highway.” Like highway in many ways: traffic, intersections, etc. Best to check validity of metaphor with participants - "member check".
5. Counting
It is also important to know (a) that we are sometime counting and 9b) when it is a good idea to work self-consciously with frequencies and when it’s not. There are three goods reasons to number
1. Seeing what you have
2. Verifying a hypothesis
3. Keeping yourself analytically honest
6. Making contrasts/comparisons
Comparison is a time-honored, classic way to test a conclusion; we draw a contrast or make a comparison between two set of things – persons, roles, activities, cases as a whole – that are known to differ in some other important respect.
7. Partitioning variables
You need the courage to question what might be called “premature parsimony”. Divide variable in early stage to avoid monolithism and data blurring.
8. Subsuming particulars into the generals
Clustering involves clumping together things that “go together” by using single or multiple dimensions.
9. Factoring
Factoring come form factor analysis, a statistical technique for presenting a large number of measured variables of a smaller number of unobserved, usually hypothetical, variables. What we do in qualitative are (a) reduce the bulk of data (b) find the patterns in them. After clustering, what they do or are is the “factor”, and the process by you generate it is “factoring”.
10. Nothing relations between variables
This process bridge the gap of the relation between variables. Thus, we might have:
 A+, B+ (positive correlated)
 A+, B- (negative correlated)
 A up, B up (positive correlated)
 A up , B down (negative correlated)
 A up, then B up (verge toward casualty)
 A up, then B down (verge toward casualty)
 A up, then B up, then A up (sequential verge toward casualty)

11. Finding intervention variables
It often happens during analysis that two variables that “ought” to go together according to your conceptual expectation. Another puzzle is why the go together in dept? For example,

A -> Q -> B

This process is to find Q, the tactic here is also to be examine a series of other candidate variables that may be “depressing” or confusing the relationship between A and B.

12. Building a logic chain of evidence
With these tactics, discrete bits of information come together, building a logic chain of evidence try to make sense of that relation and create a chain between.
13. Making conceptual/theoretical coherence
Form the field of concepts, the step of making conceptual/theoretical coherence are: (a) establishing the discrete findings (b) relating the findings to each others (c) naming the patterns (d) identifying a corresponding construct.

Tactic for testing or confirming findings

Data quality can be assessed through checking for the representative (1); checking for researcher effect (2) on the case and vise versa; and triangulating (3) across the data sources and methods. These checks also may involve weighting the evidence (4), deciding which kinds of data most trustable.

Looking at “unpatterns” can tell us a lot. Checking the meaning of outliers (5), using extreme cases (6), following up surprises (7), and looking for negative evidence (8) are all tactics that test a conclusion about a “pattern” by saying what it is not like.

Making if then test (9), ruling out spurious relations (10), replicating a finding (11), and checking out rival explanations (12) are all ways of summiting our beautiful theories to the assault of brute facts, or to a race with someone else’s beautiful theory.

11. Ethical Issues in Analysis
Ethical frameworks and aspects of research (Flinders, 1992)
Utilitarian Deontological Relational Ecological
Recruitment Informed consent Reciprocity Collaboration Cultural Sensitivity
Fieldwork Avoidance of harm Avoidance of wrong Avoidance of imposition Avoidance of detachment
Reporting Confidentially Fairness Confirmation Responsive communication

 Utilitarian: Pragmatic approach judges actions according to their specific consequences
 Deontological: Viewing invokes one or more universe rules.
 Relationship: Emphasizing issues of attachment, caring and respect, more than agreements made.
 Ecological: Basis of ethic decisions, emphasizing the impact of action on a complete possible context.

12. Producing Reports
Qualitative research is not just for fun exercise for our private enjoyment. Compare with quantitative research, which have format like this

Statement of the problem
Conceptual framework
Research questions
Data analysis

We could follow that format, but it looks Procrustean, forced. Normally, we’d have other expectation for qualitative report. Zeller(1991) suggests that qualitative studies don’t report “data” but “scenes”. So, we might consider audiences and effects.

Type of the readers
Local respondents: the prople who provide the data
Program operators: people running and/or deciding about the program being looked at.
Practitioners: people engage in same source of work
Other research: colleagues in our setting
Policymakers: governing board
General readers: purchaser of trade books
Mass readers: purchaser of magazines and newspapers

Type of effect
Aesthetic: arouse feeling
Scientific: heighten insight
Moral: sharpen moral issue
Activist: enable improved decisions

Voice, Genres, and Stances
Realist: matter-of-fact portrait
Confessional: written in fieldwork viewpoint, “what happened?”
Impressionist: personalized, disorderly reality

Formats and structures
As we have said before, good qualitative research requires an interactive mix in two basic views of the world. Then, It can be your choice. For example, Lieblich’s (1993) headings for a narrative report of the life of young woman hwo had immigrated form Russia to Israel:

Part I: life in transition
The present situation
Natasha’s past in Moldavia
The beginning of transition: Natasha’s years as a student in Moldavia
Departure and arrival

Part II: what was lost?
Parental authority
Friends and social network
Loss status
Loss of clarity of norms
Loss of self-confidence
Loss of sense of cultural belonging
Clarity of career path

Part III: Changing and not changing
First signs of changing
Establishing new friendships
A boy friend
New career
New gender expectations

Can we speak about a newly acquired identity?

1. The report should tell us what the study is all about.
2. It should communicate a clear sense of the social and historical context.
3. It should provide us see clearly what was done, by whom, and how.
4. A good report should provide data being focused.
5. Researches should expressive their conclusion.

13. Concluding Remarks
Overview of qualitative data analysis processes

Qualitative Data Analysis, Ch.13, p.308

Unless otherwise stated, the content of this page is licensed under Creative Commons Attribution-ShareAlike 3.0 License