Mixed Methods boston.. - University of Florida

Report
MIXED METHODS AND
INTERDISCIPLINARY RESEARCH
QUALITATIVE DOESN’T MEAN WIMPY
H. Russell Bernard
University of Florida
University of Massachusetts-Boston
November 28, 2012
What this talk is about
• 1. How social scientists contribute to interdisciplinary
research.
• It’s all about the science
• Interdisciplinary must never mean undisciplined
• 2. The mixed methods movement
• Not forcing a choice between qual and quant
• The qual in the qual-quant mix
About interdisciplinary research…
• The first thing: be highly qualified in your discipline.
• Maintain credentials in your discipline.
• Publishing
• This is not always easy to do
The ‘WHY’ questions
• Researchers in the natural sciences bring social scientists
•
•
•
•
•
onto project to address the social problems that are
associated with their research.
Why don’t government policy makers heed the advice of
scientists about how to stop pollution in the ocean?
Why do people waste water? What can we do about it?
Why do adolescents start smoking
Why don’t people in this village use their bed nets?
Why don’t people wash their hands after defecating.
• The most important contribution a scientist can make to
solving a problem is to be right about what causes it.
• Causal inference comes from so-called qualitative work.
Statistical regularities
• If a boy sees his mother beaten by his father this does not
make him violent toward woman, but it increases the odds
that he will be.
• Being a democracy does not prevent a nation from going
to war with other democracies, but it lowers the odds of it
happening.
• Still, no matter how strong the statistical association, we need a
mechanism to explain how the association comes about.
The missing link
• Nomothetic knowledge – theory – requires nonspurious
correlation, a logical time order, and a mechanism that
makes the correlation logical.
• Qualitative research is the key in the search for mechanism in
theory.
• Explaining contradictions
• Reviewing the literature
• Responding to critiques
• Ethnography
Networks and HIV/AIDS
• Network size for people living with AIDS is a third that of
homicide victims.
• The diagnosis was so stigmatizing and traumatizing,
people pulled back toward the number who could be
trusted to know.
Kalymnian sponge divers
• On Kalymnos, Greece, in 1965, young divers worked
longer under water and came up faster than did older
divers—and were at higher risk for the bends.
• Young men, everyone said, have a lot of machismo—a
need to show their manhood—and so they take risks by
staying down too long and coming up too fast
“That’s just how young men are”
• Where does machismo come from?
• The culture ratified but didn’t cause the behavior.
• The cause was the platika system.
• By the time they went to sea, the divers were broke and
their families had to go into debt for food and other
necessities.
• The price of sponge collapsed, but the diving labor supply
collapsed faster.
Captains push the divers
• Captains pressured divers to produce
to stay down longer and produce more
sponges.
• Result: more accidents on the job.
• Quantitative data: correlation; time
order
• Qualitative: Mechanism
What are mixed methods
• Mixed methods refers to the combination of qualitative
and quantitative data at all stages of research:
• Design
• Data collection
• Data analysis
• Presentation of results
• Mixed methods without labels
The great false divide
• The split in the social sciences is not just wrong, it’s
pernicious.
Learning the crude art of irony
• “real knowledge building versus story telling”
• “the plural of anecdote is not data”
• “deep understanding and the search for meaning versus
superficial, numerical exercises”
• “evidence-based research”
The first cut
• The first cut in research is not qualitative-quantitative. The
first cut is systematic-unsystematic.
Mixed Methods:
A safe space for empiricists
• The mixed methods movement is this generation’s
attempt to deal with the qual-quant wars in social science.
• It’s a safe-space where the qual-quant war is ignored.
• But it requires varsity training in methods
• More on that later, too
It’s nothing short of a movement
• Of >2500 references to mixed methods in the SSCI
(November 2012), all but 21 of them are since 2000.
• None pre-date 1990.
• Journal of Mixed Methods Research
• Conferences on MMR
• Handbook of MMR
Citations to mixed methods: 1997-2012
Qualitative-Quantitative: Data and Analysis
Data
Qualitative
Quantitative
Analysis
(Texts)
(Ordinal/Ratio Scale)
Qualitative
A
B
Interpretive text studies.
E.g., Hermeneutics, Grounded
Theory, Phenomenology
Search for and presentation of
meaning in results of
quantitative processing
C
D
Turning words into numbers.
E.g., Classic Content Analysis,
Word Counts, Free Lists,
Pile Sorts, etc.
Statistical & mathematical
analysis of numeric data
Quantitative
Galileo the Qualitative
• He noticed that the moon had lighter and darker areas.
The darker ones were large and had been seen from time
immemorial.
• “These I shall call the ‘large’ or ‘ancient’ spots”
• The lighter spots, he said, “had never been seen by
anyone before me.”
• The moon “is not smooth, uniform, and precisely
spherical” as commonly believed, but “is uneven, rough,
and full of cavities and prominences,” much like the earth.
So, what are qualitative data?
• Qualitative data are NOT phenomena.
• Data are reductions of our experience.
• When we reduce our experience of people’s
behavior, thoughts, and emotions to numbers,
we produce quantitative data.
And what are quantitative data?
• When we reduce our experience of people’s
behavior, thoughts, and emotions to words,
images, or sounds, we produce qualitative data.
Kinds of qualitative data
• Still images
• Sounds
• Moving images
• Written words
Why don’t we use qualitative data more?
• Most of the record of human thought and human behavior
is qualitative and it occurs naturally.
• Want to know about the evolution of sexual mores in the
U.S.?
• I Love Lucy (1950s)
• Two-and-a-half men (today)
Enter technology
• Two problems: collecting and analyzing qualitative data.
• As usual technology is the game changer.
• CAQDAS
• Voice recognition
• Visualization methods
Kinds of text analysis
• Hermeneutics
• Phenomenology
• Schema analysis
• Grounded theory
• Ethnographic decision modeling
• Analytic induction (QCA)
• Content analysis
• All are assisted by CAQDAS
Hermeneutics
• Solving puzzles in texts.
• What does this text really mean?
• Can we find out the meaning of a text by systematically comparing
it to others?
• Can we apply analytic rules consistently in order to tease out the
meaning of a text?
• Who wrote this text?
• In what order were these texts written?
Constitutional law
• What did the writers of each phrase in the U.S.
Constitution mean when they wrote it and how can we
interpret that meaning now?
• Slavery, abortion, women's right to vote, the government's
ability to tax income, …
Criminal investigations
• The Susan Smith case 1994
• Susan Smith: “My children wanted me. They needed
me. And now I can't help them.”
• David Smith: “They're okay. They’re going to come
home soon.”
• Signals of deception:
• The mixing of tense in two people’s stories about the
same event.
• Like switching from “I” to “we” in the middle of reporting
events leading up to a crime.
CAQDAS the new SPSS
• Text management software
• SPSS brought stats to the masses.
• Atlas/ti, Nvivo, MaxQDA, QDA Miner, Dedoose
• Coding and analyzing themes.
• But again: It takes varsity training in research methods to work with
all kinds of qualitative and quantitative data.
• This is not “mere technology” – it’s a game changer.
Systematic text analysis is used in many fields
• Medicine
• Education
• Political science
• Marketing
• Organizational studies
• Psychology
• Anthropology
Grounded Theory
• GT is a set of techniques for:
• 1) identifying categories and concepts (themes) that
emerge from text; and
• 2) linking the concepts into substantive and formal
theories to build theories to account for the facts in a
single case.
Margaret Kearney’s Study
• Sample: 60 women who used crack cocaine during pregnancy.
• Data: Semi-structured interviews about childhood, relationships,
life context, last and previous pregnancies
• Initial Coding: Read transcript as they were produced. Looked for
social psychological themes. Asked: “What is this an example of?”
• Emerging themes/categories
• VALUE: The degree to which women valued their pregnancy and baby•
•
•
•
to-be in relation to their own priorities.
HOPE: Expressed varying degrees of hope that their pregnancies
would end well and that they could be good mothers.
RISK: Women were aware that cocaine use posed risks to their fetus,
but perceived that risk differently.
HARM REDUCTION: Women tried in various ways to minimize the risk
to their fetus
STIGMA MANAGEMENT: They used various strategies to reduce social
rejection.
[Kearney et al. 1994 ]
“If I ever lost my children…to
me that would be the worst
thing that could ever happen
to me”
“That’s what makes me think I
don’t need this
baby…because I’m using. I
like drugs.”
“I know if I get pregnant, I
could stop the drug.”
“I might as well smoke the
next six months if I already
have screwed him up.””
“I was really concerned that
he might have something
wrong with him, some
deformity.”
“It’s okay to use drugs, but in
that last month you better
stop or you ain’t gonna bring
your baby home.”
“I been drinking a lot of pickle
juice…I’m gonna make sure
there ain’t nothing in my
system with this one.”
“I’d lie. I’d say [that crack]
wasn’t for me, it was for
another person out of town or
something.”
“The last time I went to the
doctor, they were like looking
at me funny. So I kind of knew
something was wrong and I
didn’t go back.”
After 20 Interview
After 30 Interview
After 40 Interview
Value
Hope
Facing The
Situation
Salvaging
Self
Risk
Harm
Reduction
Stigma
Management
Evading Harm
Checking the validity of the model
• Models are not the final product of the grounded-theory
approach.
• Present the model to knowledgeable informants:
pregnant drug users, project staff, health/social service
professionals familiar with the population.
• When this step is included, grounded theory is rigorous
and produces results that are replicable and valid … at
least for emic data.
•
Kearney, M. H., S. Murphy, K. Irwin, and M. Rosenbaum. 1995. Salvaging Self—A Grounded Theory of Pregnancy on Crack Cocaine.
Nursing Research 44(4):208–213.
Content Analysis
• Content analysis: procedures to make replicable
and valid inferences from text data –
advertisements, films, or answers to open-ended
questions in surveys.
• Like grounded theory, CA reduces the information
in a set of texts to a set of themes, or variables.
• But classic CA is confirmatory research, and tests
explicit hypotheses.
The Pelley Case
• In 1942, the U.S. Department of Justice accused
William Dudley Pelley of sedition.
• Independent coders classified 1,240 items in
Pelley’s publications as belonging or not
belonging to one of 14 identified Nazi
propaganda themes
• Harold Lasswell: 96.4% of the items were
consistent with the propaganda themes.
•
Goldsen, J. M. 1947. Analyzing the contents of mass communication: A step toward inter-group harmony. International Journal of Opinion & Attitude Research 1:81–92.
Content analysis has evolved
• CA has evolved since then:
• creating a text-by-theme matrix
• sampling design
• checking inter-rater reliability
• testing hypotheses about association
Hirschman’s hypothesis: men and women seek
complemetary qualities in personal ads
Resource
Hypotheses
Men
Women
Physical Status
Seek
Offer
Money
Education
Occupational
Intellectual
Offer
Offer
Offer
Offer
Seek
Seek
Seek
Seek
Love
Entertainment
(non-sexual)
Demographic
Ethnic Info
Personality
Seek
Seek
Seek
"
"
[
Hirschman, E. C. 1987. People as Products: Analysis of a Complex Marketing Exchange. Journal of Marketing 51:98–108.
]
Offer
Offer
Offer
"
"
Hirschman’s Findings
Hypotheses
Men
Women
Confirmation
Men
Women
Physical Status
Seek
Offer
Seek
Offer
Money
Education
Occupational
Intellectual
Offer
Offer
Offer
Offer
Seek
Seek
Seek
Seek
Offer
ns
ns
ns
Seek
ns
ns
ns
Love
Entertainment
Seek
Seek
Offer
Offer
ns
ns
ns
ns
Demographic
Ethnic Info
Personality
Seek
"
"
Offer
"
"
ns
"
"
Offer
"
"
Resource
]
Hirschman, E. C. 1987. People as Products: Analysis of a Complex Marketing Exchange. Journal of Marketing 51:98–108
By 1998…things were changing
• Internet personal ads were taking over from print, but men
continued to seek a particular kind of body in women and
women continued to offer a particular kind of body.
• Men and women alike mentioned their financial status, but
women still were more likely to explicitly seek someone
who is financially secure.
• Evidence of a major shift … in Spain: Men of all ages
sought physical attractiveness in women.
• Women under 40 sought physical attractiveness in men.
•
Gil-Burman, C., F. Peláez, and S. Sánchez. 2002. Mate choice differences according to sex and age: An analysis of personal advertisements in Spanish
newspapers. Human Nature 13:493–508.
And today …
• Today, personal ads continue to inform us about
preferences in mate selection among heterosexuals,
but also among gay men, lesbians and bisexuals.
• Obituaries of business leaders contain data about
men’s and women’s management practices and
about how people in different cultures memorialize
the dead.
•
•
•
•
•
Smith, C. A. and S. Stillman 2002a. Butch/femme in the personal advertisements of lesbians. Journal of Lesbian Studies 6:45–51.
Phua, V. C. 2002. Sex and sexuality in men’s personal advertisements. Men and Masculinities 5:178–191.
Kirchler, E. 1992. Adorable woman, expert man: Changing gender images of women and men in management. European Journal of Social Psychology 22:363–373.
Alali, A. O. 1993. Management of death and grief in obituary and in memoriam pages of Nigerian newspapers. Psychological Reports 73:835–842.
de Vries, B. and J. Rutherford 2004. Memorializing loved ones on The World Wide Web. Omega: Journal of Death and Dying 49:5–26.
Content dictionaries
• To build a coding machine: assign words to categories
according to a set of rules.
• Write a program that reads text and assigns words to
categories.
• Phillip Stone –1960: The General Inquirer and the
Harvard Psychosocial Dictionary
Stone, P. J., D. C. Dunphy, M. S. Smith, and D. M. Ogilvie, eds. 1966. The General Inquirer: A Computer Approach Tto Content Analysis. Cambridge, MA:
M.I.T. Press.
Stone’s first test
• 66 suicide notes—33 by men who had taken their own
lives, and 33 by men who produced fake suicide notes.
• The program parsed the texts and got it right 91% of the
time.
• Today’s dictionary can tell if “broke” means "fractured,"
"destitute," "stopped functioning," or (when paired with
"out") "escaped."
Content dictionaries get better
• Rosenberg: 71 speech samples from people with
psychological disorders (depression, paranoia) or cancer.
• The human coder beat the computer in diagnosing
patients who had cancer.
• The computer beat the human coder in identifying
psychological disorders.
• Today, just two decades later, every time you hear “this call
may be monitored” …
•
•
Rosenberg, Stanley D., P. P. Schnurr, and T. E. Oxman 1990. Content Analysis: A Comparison of Manual and Computerized Systems.
Journal of Personality Assessment 54(1 and 2):298–310.
Analytic induction
• Think of the difference between saying: “whenever you
see X you will see Y” and “whenever you see X, there is a
92% chance that you’ll see Y”.
• The method is based on Mill’s work on logic and the
methods of agreement and difference.
Analytic induction – Ragin’s QCA method
• Charles Ragin formalized the logic:
• With one dichotomous variable, A, there are 2
possibilities: A and not-A.
• With two dichotomous variables, A and B, there are 4
possibilities.
•
Ragin, C. C. 1987. The comparative method. Moving beyond qualitative and quantitative strategies. Berkeley: University of California Press.
Haworth-Hoeppner’s QCA of eating disorders and
body image
• 30 women, 21 either anorexics or bulimics
• Asked about body image and eating problems
• Four Themes
• (1) Constant criticism in the family
• (2) Coercive parental control
• (3) Feeling unloved by parents
• (4) Family discourse on weight
=C
=R
=U
=D
• Code transcripts for these concepts.
• Find the simplest set of features that account for
the dependent variable.
•
Haworth-Hoeppner, S. 2000. The critical shapes of body image: The role of culture and family in the production of eating disorders. Journal of Marriage and
the Family 62:212–227.
Data Matrix for Haworth-Hoeppner’s Study
Source: Susan Haworth-Hoeppner (personal communication)
Critical
family
environment
Coercive
parental
control
Unloving
parent-child
relationship
Main discourse on
weight
Suffers
from
eating
disorder
1
1
0
0
0
0
2
0
1
0
0
0
3
1
0
1
0
0
4
0
0
0
0
0
5
0
0
0
0
0
6
0
0
0
0
0
7
0
0
0
0
0
8
0
0
0
0
0
9
0
0
0
0
0
10
1
1
1
0
1
11
1
1
0
0
1
12
0
0
0
1
1
13
0
0
0
1
1
14
1
0
0
1
1
15
1
0
0
1
1
16
1
1
0
0
1
17
1
1
0
0
1
18
1
1
0
0
1
19
1
1
0
0
1
20
1
0
1
1
1
21
1
0
1
1
1
22
1
1
1
1
1
23
1
1
1
1
1
24
1
1
1
1
1
25
1
1
1
1
1
26
1
1
1
1
1
27
1
1
1
1
1
28
1
1
1
1
1
29
1
1
1
1
1
30
1
1
1
1
1
Case
Eating disorders = CR + CD + r u D
• Eating disorders are caused by the simultaneous
presence of C AND R (Constant criticism in the family and
Coercive parental control), AND by the simultaneous
presence of C AND D (Constant criticism in the family and
Family discourse on weight), AND by the presence of D
(Family discourse on weight) in the absence of R and U
(Coercive parental control and Feeling unloved by
parents).
•
Haworth-Hoeppner, S. 2000. The critical shapes of body image: The role of culture and family in the production of eating disorders. Journal of Marriage and
the Family 62:212–227.
Visualization methods
• Make quantitative data qualitative so we
can understand them.
• Relational data are very, very
complicated… pile sorts, for example.
• Here’s just one:
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
1 2 3
- - 1 0
1
0
0 0
0 0 0
0 0 0
1 1 0
0 0 0
0 0 0
0 0 0
0 0 1
0 0 0
0 0 0
0 0 0
0 0 0
0 0 0
0 0 0
0 0 0
1 1 0
0 0 0
0 0 0
4
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
1
0
5
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
1
6
1
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
1
0
0
7
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
8
0
0
0
0
0
0
0
0
0
0
1
0
0
0
0
0
0
0
0
9
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
1
0
0
0
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
1
1
0
0
0
0
0
0
0
0
0
0
0
0
0
1
0
0
0
0
0
1
2
0
0
0
0
0
0
0
1
0
0
0
0
0
0
0
0
0
0
0
1
3
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
1
4
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
1
5
0
0
0
0
0
0
0
0
0
0
1
0
0
0
0
0
0
0
0
1
6
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
1
7
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
1
8
1
1
0
0
0
1
0
0
0
0
0
0
0
0
0
0
0
1
9
0
0
0
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0 0
0 0 0
0 0 0 0
2
0
0
0
0
0
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
An Individual similarity matrix for a 20-item pile sort
Bernard, H. R. 2012. Social Research Methods: Qualitative and Quantitative Approaches, 2 nd edition. Newbury Park, CA:
Sage. P. 410
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
CA
FLU
CLD
DIA
AID
SCZ
MLR
POX
HD
HIV
MON
ART
TB
POL
MEA
PNE
MUM
SYP
MEN
DEP
1
CA
---1.00
0.08
0.08
0.24
0.16
0.08
0.22
0.05
0.22
0.11
0.00
0.22
0.08
0.11
0.08
0.08
0.05
0.05
0.16
0.08
2
FLU
---0.08
1.00
0.92
0.00
0.03
0.03
0.24
0.32
0.05
0.05
0.49
0.08
0.30
0.22
0.19
0.59
0.22
0.08
0.24
0.08
3
CLD
---0.08
0.92
1.00
0.00
0.03
0.03
0.24
0.35
0.05
0.05
0.49
0.08
0.30
0.22
0.19
0.59
0.22
0.08
0.24
0.08
4
DIA
---0.24
0.00
0.00
1.00
0.03
0.16
0.08
0.05
0.35
0.05
0.00
0.35
0.05
0.05
0.05
0.00
0.05
0.11
0.05
0.22
5
AID
---0.16
0.03
0.03
0.03
1.00
0.00
0.16
0.03
0.22
0.84
0.14
0.05
0.16
0.08
0.08
0.11
0.05
0.43
0.14
0.00
6
SCZ
---0.08
0.03
0.03
0.16
0.00
1.00
0.03
0.03
0.00
0.00
0.03
0.11
0.05
0.00
0.03
0.03
0.05
0.00
0.14
0.76
7
MLR
---0.22
0.24
0.24
0.08
0.16
0.03
1.00
0.32
0.14
0.16
0.27
0.11
0.35
0.46
0.38
0.22
0.35
0.11
0.32
0.03
8
POX
---0.05
0.32
0.35
0.05
0.03
0.03
0.32
1.00
0.05
0.11
0.38
0.08
0.11
0.46
0.70
0.27
0.73
0.19
0.05
0.08
9
HD
---0.22
0.05
0.05
0.35
0.22
0.00
0.14
0.05
1.00
0.22
0.03
0.24
0.14
0.08
0.08
0.08
0.05
0.03
0.11
0.05
10
HIV
---0.11
0.05
0.05
0.05
0.84
0.00
0.16
0.11
0.22
1.00
0.24
0.05
0.22
0.14
0.14
0.11
0.14
0.54
0.11
0.00
11
MON
---0.00
0.49
0.49
0.00
0.14
0.03
0.27
0.38
0.03
0.24
1.00
0.03
0.30
0.35
0.27
0.41
0.41
0.27
0.27
0.08
12
ART
---0.22
0.08
0.08
0.35
0.05
0.11
0.11
0.08
0.24
0.05
0.03
1.00
0.11
0.11
0.08
0.11
0.05
0.11
0.05
0.08
13
TB
---0.08
0.30
0.30
0.05
0.16
0.05
0.35
0.11
0.14
0.22
0.30
0.11
1.00
0.22
0.24
0.54
0.19
0.11
0.41
0.05
14
POL
---0.11
0.22
0.22
0.05
0.08
0.00
0.46
0.46
0.08
0.14
0.35
0.11
0.22
1.00
0.43
0.16
0.46
0.11
0.24
0.03
15
MEA
---0.08
0.19
0.19
0.05
0.08
0.03
0.38
0.70
0.08
0.14
0.27
0.08
0.24
0.43
1.00
0.19
0.81
0.14
0.14
0.05
An aggregate similarity matrix for a 20-item pile sort
Bernard, H. R. 2012. Social Research Methods: Qualitative and Quantitative Approaches, 2 nd edition. Newbury Park, CA: Sage. P. 418
16
PNE
---0.08
0.59
0.59
0.00
0.11
0.03
0.22
0.27
0.08
0.11
0.41
0.11
0.54
0.16
0.19
1.00
0.16
0.11
0.30
0.08
17
MUM
---0.05
0.22
0.22
0.05
0.05
0.05
0.35
0.73
0.05
0.14
0.41
0.05
0.19
0.46
0.81
0.16
1.00
0.16
0.16
0.05
18
SYP
---0.05
0.08
0.08
0.11
0.43
0.00
0.11
0.19
0.03
0.54
0.27
0.11
0.11
0.11
0.14
0.11
0.16
1.00
0.03
0.05
19
MEN
---0.16
0.24
0.24
0.05
0.14
0.14
0.32
0.05
0.11
0.11
0.27
0.05
0.41
0.24
0.14
0.30
0.16
0.03
1.00
0.14
20
DEP
---0.08
0.08
0.08
0.22
0.00
0.76
0.03
0.08
0.05
0.00
0.08
0.08
0.05
0.03
0.05
0.08
0.05
0.05
0.14
1.00
Multidimensional scaling
• We can reduce complexity with factor analysis,
but this may still be too complex to understand.
• MDS produces a graphic display of the relation
among any set of items.
• The items might be people, or objects, or ideas,
or attitudes.
• MDS turns numbers that represent relations into
a picture, which pattern-seeking animals like
humans can easily understand.
What does it mean to be green?
• Free list produced 85 items – “wear sweaters in
the house during the winter to save energy,”
“teach kids to respect the environment”
• Pile sort these little texts –what-goes-with-what?
– and you get an 85x85 relational matrix
• This is hopeless
• Turn to MDS
Multidimensional scaling of 85 items in two dimensions (44 informants)
*B
Cut grass high
Plant shrubs
Plant garden
Compost
Plant trees
Restore buildings
Pick up litter
Paper bags
Encourage others to recycle
Don’t litter
Organize drives for recyclables
Encourage recycled products
Teach kids about recycling
Save wetlands
*A
Political activities
Write congressperson
“Save the Earth” t-shirts
*A
Mulch grass clippings
Water off while shaving
Full loads in dishwasher
Cold-water detergent
Lowflow shower
Rinse w/ cold water
Short dishwasher cycles
Water lawn in morning/evening
Recyling bins
Water-saving toilets
Salvation Army
Use things longer
Cloth diapers
Reuse towels
Cool leftovers
Wear sweaters
Clothes line
Double-pane windows
Gas heat
Insulate heating ducts
Convection oven
Both sides paper Clean lint filter
Use own grocery bags
Recylce toxic prods.
*B
Copper & brass
Redeem cans
Put bins in office
Buy recycled prods.
Overpackaged foods
No aerosol
Remove CFC in old refrig.
Reduce meat consumption
Walk or bike
Dolphin safe tuna
Carpool
Inflate tires properly
Public transport
Gas mileage on new car
Use ethanol
Assure car runs well
Join environmental groups
Teach kids about endangered species
Show kids by example
Teach about gains from environment
Teach kids to preserve planet
Support world population organizations
Tell others not to do bad things
Ride Motorcycle
Close shades
Turn off lights
Air off when leave
Fans
Dishwasher w/ built-in heater
Insulate home
Weatherstrip
Automatic timers for house temp.
Frig. seal
Dryer with moisture sensor
Oven door seal
Freezers on top
Fluorescent bulbs
Low-watt bulbs
Dishwasher w/ airdry
Photocells
Furnace tune-up
Regulate thermostat
Buy Electric Car
Bernard, H. R., G. W. Ryan, G. W. and S. P. Borgatti, S. P. 2010. Green cognition and behavior: A cultural domain analysis.
In: W. Kokot, ed. Papers in Honor of Hartmut LangNetworks, Resources and Economic Action. Ethnographic Case Studies
in Honor of Hartmut Lang, C. Greiner and W. Kokot, eds. Berlin, Dietrich Reimer Verlag. Pp. 189-215.
Analysis: a dialectic from qual to quant to qual …
• We can use cluster analysis on the same matrix
to identify groups of items.
• The next slide shows the clusters superimposed
on the MDS output, using colors. We’ve named
the chunks.
• This is qualitative analysis (naming chunks) of a
picture (qualitative data) derived from a matrix
(quantitative data), derived from texts (qualitative
data).
Multidimensional scaling and cluster analysis of 85 items in two dimensions (N= 44)
*B
Plant shrubs
Plant garden
Compost
Plant trees
Garden
Water off while shaving
Full loads in dishwasher
Cold-water detergent
Lowflow shower
Rinse w/ cold water
Short dishwasher cycles
Water lawn in morning/evening
Close shades
Turn off lights
Air off when leave
Fans
Dishwasher w/ built-in heater
Insulate home
Weatherstrip
Automatic timers for house temp.
Frig. seal
Dryer with moisture sensor
Copper & brass
Oven door seal
Redeem cans
Freezers on top
Put bins in office
Fluorescent bulbs
Buy recycled prods.
Low-watt bulbs
Overpackaged foods
No aerosol
Dishwasher w/ airdry
Remove CFC in old refrig.
Photocells
Reduce meat consumption
Walk or bike
Furnace tune-up
Dolphin safe tuna
Carpool
Inflate tires properly Regulate thermostat
Public transport
Gas mileage on new car
Use ethanol
Assure car runs well
Restore buildings
Pick up litter
Paper bags
Encourage others to recycle
Don’t litter
Organize drives for recyclables
Encourage recycled products
Teach kids about recycling
Save wetlands
Rhetoric
Cut grass high Mulch grass clippings
Water-saving toilets
Recyling bins
Salvation Army
Use things longer
Cloth diapers
Reuse towels
Cool leftovers
Wear sweaters
Clothes line
Double windows
Gas heat
Insulate heat ducts
Convection oven
Both sides paper Clean lint filter
Use own grocery bags
Recylce toxic prods.
*B
Recycle
House
*A
Political activities
Write congressperson
“Save the Earth” t-shirts
*A
Join environmental groups
Teach kids about endangered species
Show kids by example
Teach about gains from environment
Teach kids to preserve planet
Support world population organizations
Tell others not to do bad things
Ride Motorcycle
Buy Electric Car
Bernard, H. R., G. W. Ryan, G. W. and S. P. Borgatti, S. P. 2010. Green cognition and behavior: A cultural domain analysis.
In: W. Kokot, ed. Papers in Honor of Hartmut LangNetworks, Resources and Economic Action. Ethnographic Case Studies
Multidimensional scaling of 33 home-based items (N=44)
Insulate heating ducts
Furnace tune-up
Insulate home
Convection oven
Freezers on top
Frig. Seal
Weatherstrip
Gas heat
Close shades
Fluorescent bulbs
Double-pane windows
Cool leftovers
Automatic
timers
for house temp.
Sweaters
Regulate thermostat
Low-watt bulbs
Fans
Photocells
Oven door seal
Turn off lights
Water-saving toilets
Air off when leave
Heat & Light
Dishwasher w/ air dry
Clothes line
Dryer with moisture sensor
Water off while shaving
Dishwasher w/ built-in heater
Clean lint filter
Cold-water detergent
Low-flow shower
Water
Rinse w/ cold water
Short dishwasher cycles
Full loads in dishwasher
Bernard, H. R., G. W. Ryan, G. W. and S. P. Borgatti, S. P. 2010. Green cognition and behavior: A cultural domain analysis.
In: W. Kokot, ed. Papers in Honor of Hartmut LangNetworks, Resources and Economic Action. Ethnographic Case Studies
in Honor of Hartmut Lang, C. Greiner and W. Kokot, eds. Berlin, Dietrich Reimer Verlag. Pp. 189-215.
Jang and Barnett’s study of CEOs’ letters
Goal: Compare American and Japanese business practices
Data: CEOs’ yearly letters to stockholders from 35 firms (1992)
Calculate
Word Frequencies
Read
All Texts
Amer. 1 Text …
Amer. 2 Text …
Amer. 3 Text …
…
Japan. 1 Text …
Japan. 2 Text …
Japan. 3 Text …
...
Eliminate
Stop Words
a
an
and
because
also
else
here
was
will
etc.
Jang, Ha-Yong., and George. Barnett. 1994. Cultural Differences in
Organizational Communication: A Semantic Network Analysis. Bulletin de
Méthodologie Sociologique 44 (Septem-ber):31–59.
we
our
business
products
new
company
market
billion
world
…
great
image
...
armed
garage
655
788
180
172
185
170
113
103
82
10
10
1
1
Identify
Top 94 words
1 we
2 our
3 business
4 products
5 new
6 company
7 market
8 billion
9 world
…
94
Multidimensional scaling of company-by-company matrix
Respondents
(Companies)
Respondents
(Companies)
A
B
C
D
A
-
.9
.3
.4
B
.9
-
.4
.2
C
.3
.4
-
.8
D
.4
.2
.8
-
J
J
J
J
J
J J
J
J
J J
A JA J A
AJ AA A
A
A A J
J
A
A
A
A
A U.S. Company
J Japanese Company
Jang, H-Y., and G. Barnett. 1994. Cultural Differences in Organizational Communication: A Semantic Network Analysis. Bulletin de
Méthodologie Sociologique 44 (Septem-ber):31–59.
]
Visualizing complex data
• There are new methods for taking all this further –
visualizing very complex interactions, like those in
social networks.
• We can add dimensions, color and even motion
as aids to visualizing the complex relations in
matrices.
• These are all qualitative aids to understanding
numerical data.
The Freemans’ EIES data
Freeman, L. C. 2000. Visualizing social networks. Journal of Social Structure.
http://www.cmu.edu/joss/content/articles/volume1/Freeman.html
These are not your mother’s qualitative
methods
• Social scientists who can do it all – or work with teams
that, collectively, can do it all -- will be in demand.
• In fact, there is no shortage of jobs for social scientist.
• There are exciting times ahead.

similar documents