critical thinking and arguments about contradictions using qualitative data & nvivo. 2014
DESCRIPTION
How to analyse qualitative data, using NVIVO, to examine data that seems to hold Contradictions. Three specific types of contradictions typically occur. I show how Logical arguments (units of logic that have premises, conclusion each) can be put together to make complex arguments. I also show how 'encompassing' can help resolve issues of empirical contradiction. Finally the slides show the Bowell-Kemp and FIsher type of diagram of an 'argument' (a series of linked statements leading to a related conclusion). Wendy Olsen University of Manchester July 2014TRANSCRIPT
42510011 0010 1010 1101 0001 0100 1011
Critical Thinking About Contradictions Using
Qualitative Data (Part One)
By Wendy Olsen
2014
Methods@Manchester Workshop
Aiming at PhD Students and Researchers Who Want to Disseminate Arguments
4251
0011 0010 1010 1101 0001 0100 1011
AIM 1) To introduce ‘critical thinking’ to your academic writing. Class Exercise 1:
We look at your own samples of data, and at one of mine.
AIM 2) Set up (hand) coding on a simple transcript.
AIM 3) A simple NVIVO lecture.
WE WILL BE IN WILLIAMSON ROOM 3.59 COMPUTER CLUSTER 11-3
AIM 4) (rejoin in Room HBS 1.69) and do Exercise 2.Integrate our analysis of the sample transcript (or your own data
sample, if you bring one) with what we learned about complex social-science argumentation.
4251
0011 0010 1010 1101 0001 0100 1011
Critical Thinking• Parse the logic of a sample piece of writing.• The steps should be related, and coherent.• The conclusion should rest on the argument.• Complex arguments use data as evidence.• P= Premises• C = Conclusions• R = Reasoning and D = Data or evidence.
4251
0011 0010 1010 1101 0001 0100 1011
4
An Argument …
• Is an extended set of sentences about one thing.
• Has a coherent relationship among the sentences.
• Is coherent as a whole.
• Leads toward its own conclusion.• I have stipulated this definition.
4251
0011 0010 1010 1101 0001 0100 1011
5
Warranted Arguments
• In a warranted argument,– Conclusions are not just beliefs,– Premises are consistent and coherent,– Reasoning is sound,– Verbs used are relevant and appropriate,– Logic is used (various types), and– The conclusion would be false if any of the P’s or
R’s are false. Use Triangulation!!– * on the previous slide, ‘better impact’ rested
upon unspecified Premises and Reasoning about ‘impact’
4251
0011 0010 1010 1101 0001 0100 1011
6
Exercise 1.–In Pairs: Share textual samples.
• Write down two first-stage ‘codes’ which are theme names, so that the material in this sample could be RETRIEVED.
• Write down two second-stage ‘codes’ which are analytical, ie. Perhaps they relate to theory, such as gender roles, agency, structures of family, power, or other.
• The theme is going to develop into an argument. You are not merely descriptive in summarising your findings. Induction. This is called the second-stage coding.
• See my example coding on ethnicity.
4251
0011 0010 1010 1101 0001 0100 1011
See also:– Bazeley, books and online materials on
NVIVO.– Elliot, J (2005) Using Narrative in Social
Research, London, Sage.– Flick, U/Banks, M (2007) Using Visual Data in
Qualitative Research, London, Sage.– Gibbs, G (2007) Various books on using
NVIVO.– Lewins, A and Silver, C (2007) Using Software
in Qualitative Research, London, Sage.
7
4251
0011 0010 1010 1101 0001 0100 1011
Conclusion of Exercise 1
• Step 1 : simple, descriptive codes.
• Step 2 : analytical codes, axial codes.• These invoke theory.
• Step 3 : develop an argument and test it out, work on it. Code more…
– What are the people’s lay arguments? (See Sayer)– What is your expert argument, over-arching?
8
4251
0011 0010 1010 1101 0001 0100 1011
Conclusion of Exercise 1
• Step 4: what are the contradictions?
• You might expect 3 types. – Resistance to authority / norms / doxa.– Agreeing with and yet also disagreeing
with a norm.– A text contradicting itself, in the sense of its
intended meanings changing over time.
9
4251
0011 0010 1010 1101 0001 0100 1011
10
Lecture 1 continued
• Better and worse ARGUMENTS.– An argument is a theme, and it has to have a counter-
theme (the ‘antithesis’).
• Good arguments might have:– Better ethics than worse arguments, OR– More consistent premises, OR– Consideration of data that might falsify a claim, OR– Coverage of things that are very well known to the
writer.– See Bowell and Kemp, Critical Thinking, London:
Routledge, 3rd ed., 2010. pg 96.
4251
0011 0010 1010 1101 0001 0100 1011
11
Parsing Arguments
• To parse means to break up into small chunks.TO PARSE . MEANS . TO BREAK . UP .
INTO . SMALL . CHUNKS.
Verb definition synonym-verb object
• Break up arguments into P, R, D, E, C’s– Fisher, A. (1988). The Logic of Real Arguments.
Cambridge, NY and Sydney, Cambridge Univ. Press.
4251
0011 0010 1010 1101 0001 0100 1011
12
Warranted Arguments About Contradictions
– Reasoning is sound, and it links together elements into a complex chain of reasoning.
– Example:
• Resistance to authority / norms / doxa.
4251
0011 0010 1010 1101 0001 0100 1011
More applications…
• Agreeing with and yet also disagreeing with a norm.
• A text contradicting itself, in the sense of its intended meanings changing over time.
– Please work on such things in the NVIVO practical.
13
4251
0011 0010 1010 1101 0001 0100 1011
14
Data in Warranted Arguments
• Premises• Data … which verifies findings . . .
– After the findings are rewritten back into the data sections! But you did not know in advance what you would find. Discoveries. Retroduction.
• Reasoning . . . Which uses data! Depends on it! Needs it! Develops / analyses it!
• Conclusion(s) (Danermark et al 2001)• Beware of verificationism. Hence use hypotheses.• Or use claims e.g. ‘it is claimed that X Y’
4251
0011 0010 1010 1101 0001 0100 1011
15
The Duhem-Quine Paradox
• Willard Van Orman Quine (June 25, 1908 – December 25, 2000) – Edgar Pierce Chair of Philosophy at Harvard,
1956–78
• Two Dogmas of Empiricism (1951)• Argues that much of science does tend to
verify itself, and that we cannot describe ALL of the world, only a TINY part of it.
• As a result, descriptions often contradict each other.
fallibilism. Epistemological fallibilism.
4251
0011 0010 1010 1101 0001 0100 1011
The Implications of Quine for our Study of Contradictions
• We have to holistically develop a theoretical framework that encompasses the material we have.
• We can get more material after we do retroduction.• That means asking why these data showed up a
contradiction… ask again? Get texts/symbols? Look at visual evidence?
• Use your own cultural resources. Cite sources.
• Then build the complex argument.
16
4251
0011 0010 1010 1101 0001 0100 1011
Help From Bowell & Kemp
• A complex argument can be built up in stages.
• Each stage has Premises, Reasoning, Data and Conclusion.
• The Conclusion of one is a premise of another. (OR, using Fisher, a supposition)
17
4251
0011 0010 1010 1101 0001 0100 1011
How Bowell & Kemp and Fisher Illustrate Arguments
18
4251
0011 0010 1010 1101 0001 0100 1011
How We Will Summarise Ours
19
4251
0011 0010 1010 1101 0001 0100 1011
How We Will Summarise Ours
20
Suppose P1 a taboo on marrying a UK born person
Evidence is that some marry UK born people and some don’t
Falsifies
Respondent says she marries someone of whom family will approve
Weakness of a single case as evidence
After falsifying--- Then what?
4251
0011 0010 1010 1101 0001 0100 1011
How We Will Analyse Mixed Data Types
21
Activity of Retroduction
What would explain what we have observed, in all its variety?
Gather more evidence, e.g. create a Census table of Country of Birth of person and of their marriage partner using the couples in the UK Census. (! EASY! USES CENSUS MICRODATA – UK BEST SOURCE COMPARED WITH OTHER COUNTRIES) What does this Table mean? What does it imply about the UK world?
Combine this with close study of what must be true for some to conform, and others to not conform, with the norm
4251
0011 0010 1010 1101 0001 0100 1011
How We Will Summarise Ours
22
Suppose P1: a contested taboo on marrying a UK born person
Evidence is that some marry UK born people and some don’t
Revised
Some respondents say they marry someone of whom family will approve
Some respondents say they make a personal choice and/or fall in love.
Evidence on marital patterns is mixed by ethnicity and by religion and COB
These women have to navigate roles as active agents
4251
0011 0010 1010 1101 0001 0100 1011
There are many situations…
23
4251
0011 0010 1010 1101 0001 0100 1011
Help From Bowell & Kemp
• A complex argument can be built up in stages. Example: – We did a deductive test of whether the pakistani-
origin couple have a traditional division of labour. – Next we inductively seek evidence whether the
woman is comfortable with her traditional division of household labour.
• We also note she has a paid parttime job.• We also note her husband works.• And she travels for work.
24
4251
0011 0010 1010 1101 0001 0100 1011
Keep the argument developing…
• We interpret her contradictory statements as accepting that there are two sets of pressures upon them:
• Field of work: be a dual-earner household.• Field of family: contribute money to HH income. Be a hard
worker. • Field of Household: woman should be a double-burden
bearer without complaint.• Field of work: get help with domestic work to enable
ourselves to get to do paid work.
– Retroduce using Bourdieuvian social theory.25
4251
0011 0010 1010 1101 0001 0100 1011
Introduce a novel concept
• We need to resolve a problem that we switched meanings in our own terminology ‘traditional division of household labour’.– A) new term, ‘conceptual ideal of the
TDHL’ vs. ‘practical reality of the pragmatic sharing of the HL’
– B) Now re-write the whole argument.
26
4251
0011 0010 1010 1101 0001 0100 1011
COMMENTARY
• I SOLVED MY INTERPRETATION PROBLEM.
• IN FISHER, WE ALSO SEE THIS AS THE LOGIC OF CHALLENGING A SUPPOSITION: suppose there is a norm that all pakistan-origin households have a strict gender division-of-labour…
27
Matrix search– These are complex searches on the database. They usually result in a
tabular output. They may correspond with ‘complex arguments’ that summarise and resolve contradictions.
• To mix Survey Data or a face-sheet with Textual Data, use Import and then Analyse in NVIVO. You will need to create Case Nodes.
• These then allow a Classification sheet showing the attributes of each person/case.– You can use a Matrix Search on the sub-nodes of a
demographic node like CLASS with the sub-nodes of another substantive node like ATTITUDES.
– Results are Some/None coded; and linkages to text.
13/04/23 28
Conclusions
• So far I introduced complex reasoning• I said that NVIVO can be used to code (at Stage
2) the themes that help you to resolve any contradictions.
• Sometimes these relate to power, where people or other agents have an ambiguous and changing position vis-a-vis power/hegemonic-discourses/norms.
– Actually setting up a matrix search is beyond a one-day course, but is possible in NVIVO to illustrate contradictions.
13/04/23 29
4251
0011 0010 1010 1101 0001 0100 1011
30
Very Concise Sources:
• Cook, S. (1999). "Methodological aspects of the encompassing principle." Journal of Economic Methodology 6: 61-78.
• AND chapters 2-3 of:• Sayer, A. (1992 (orig. 1984)). Method in
Social Science: A Realist Approach. London, Routledge.
• OR two chapters from Smith, M., ed. 1998, Social Science in Question.
42510011 0010 1010 1101 0001 0100 1011
Thank you.P.S. Something to read by Wendy Olsen on
ethics . . . Olsen, Wendy, (2009) “Moral Political Economy and
Moral Reasoning About Rural India: Four Theoretical Schools Compared”, Cambridge
Journal of Economics, http://cje.oxfordjournals.org/cgi/reprint/33/5/875.pdf,
33:5, 875-902.