Cultural Consensus Analysis in Psychological Anthropology

Cultural Consensus Analysis
Bill Dressler & Kathy Oths
Society for Psychological Anthropology
2017
Basic Outline of (very brief) Workshop
Review basics of the model
Quick overview of techniques
Studying variation
Cultural consonance
Steps in Studying Cultural Models
Determine cultural domain to be studied
 
Identify salient elements of the domain
Free lists                      Open-ended interviews
Explore the structure of the domain
Pile sorts, ratings, rankings                             Narrative analysis
 
Test for Cultural Consensus
Explore distribution of cultural knowledge and
content of shared understanding
Point to remember….
Cultural consensus analysis is not the beginning
Rather, it is the culmination of ethnographic
work, usually supplemented by various forms of
structured ethnographic techniques
CCA then enables you to:
Verify that knowledge is shared within a domain
Explore and better understand the configuration of
the cultural model(s)
Review of the Basics of the Model
The Cultural Consensus Model
R
1
R
2
R
3
R
4
R
5
R
6
.
.
R
n
Degree of 
sharing of
knowledge
Determine R’s
relative degree of 
shared knowledge
Calculate a consensus set
of responses, weighted by
pattern of knowledge 
sharing
Responses of
Individuals
(Romney, Weller and Batchelder 1986)
2-dimensional array of food similarities/differences: 2010 - 2014
2-dimensional array of food similarities/differences: 2010 - 2014
Meal
s
Health
Individual representation =
[cultural model] + [personal model]
What are the features
What are the features
 that distinguish among
 that distinguish among
 the elements of the model?
 the elements of the model?
Is there a shared
 understanding of these
 features?
Adriane
.83
Greg
Why?
Pure personal biography:
                       Adriane – individual experiences 
                                      with food
     .83
                        Greg – individual experiences
                                   with food
Adriane
.83
Greg
Cultural Model
  of Food
knowledge1
knowledge2
Individual correlation = knowledge1 x knowledge 2
Shared cultural model:
Patterns of agreement:
Meals
Ratio indicates overall
 consensus, if > 3.0
Cultural competence coefficients
 indicate how well your answers
 reflect group-level answers
Average cultural competence
 indicates the strength of the
 cultural consensus
Cultural model of how
Cultural model of how
to construct a meal
to construct a meal
Kim
Robin
Adriane
Bryanna
Kelso
Diedre
Alec
Heidi
Joshua
Greg
.95
.98
.93
.98
.93
.92
.90
.76
.78
.89
Consensus ratings
of foods
Consensus ratings = a ‘cultural best estimate’
How would a reasonably culturally competent
member of this social group rate or classify these
foods?
Fitting attributes to the array
How do we know if people are using features to
define semantic structures?
Correlate the features with the distance between
the points in the map of the pile sort
If the correlation is high (> .75), then probably
people were using those features to sort the
terms
?
?
Fitting the attributes
Meal correlation = .91
Health correlation = .91
Southern correlation = .20
2-dimensional array of food similarities/differences: 2010 + 2011
2-dimensional array of food similarities/differences: 2010 + 2011
 
Is it healthy?
 
Where does  it
 go on my plate?
Quick Overview of Techniques
Doing Cultural Consensus Analysis
In Anthropac
The formal process model
The informal data model
In a standard statistical package: SPSS
Formal process model (MTTIW)
Informal data model
Studying Variation
Classic Brazilian Food
Feijão/Feijoada
Rice
Churrasco
Cerveja (beer)
Macarão (pasta)
Hearts of palm
Pintado (catfish)
Doces (sweets)
Sucos/vitaminas
Almoço (lunch)
Launches
Salgadinhos/picados
Comida mineiro
Comida baiano
Comida carioca
Comida gaúcho
The Prestige Value of Food in Urban Brazil
(Oths, Carolo, & dos Santos 2003)
Testing variation with separate model
estimates within groups 
(Oths, Carolo, & dos Santos 2003)
 
 
Cultural Modeling - The Domain of Family Life
“Family organization
“Affective climate
Violence/addiction
“Lack of education
Importance for having
   a family
Distribution of Cultural Competence in
Family Life – urban Brazil
Eigenvalue ratio = 
8.49
Recommended cut-off points for the S.D. of
cultural competence 
(Hruschka and Maupin 2013)
Relationship of years in Nairobi and cultural
competence in  the cultural model of HIV+
management 
(Copeland 2011)
Cultural competence in family life and
psychological distress in urban Brazil 
(n = 52)
Residual Agreement
Concept introduced by Boster in 1986
Refers to agreement among respondents
beyond
 an overall cultural consensus
“Multicentric domain”
Defined by Caulkins and Hyatt
Multicentric domains have multiple centers of
agreement:
Subcultural domains, in which there are two or more centers
of agreement that are different but not oppositional
Contested domains, in which some individuals take a
perspective opposite to that expressed by others in the same
population
Boster’s (1986) original approach
Works from the agreement matrix
Run consensus analysis and then generate the
predicted agreement
Subtract the predicted agreement matrix from
the observed agreement matrix to generate a
residual agreement matrix
Use QAP (quadratic assignment procedure) to
test for correlation of the observed and residual
agreement matrices
If matrices are correlated, then there is residual
agreement not exhausted by the consensus
Boster’s original approach (cont.)
This can be thought of as an “omnibus” test of
residual agreement
Just tells you it’s there
Not what it is
Utility of this test may not be great, especially if
there is always going to be some kind of residual
agreement—a reasonable assumption
Norbert Ross & Doug Medin
In essence, they use the same approach as
Boster to test for residual agreement
They calculate residual matrices within groups and
compare them
Then, they identify which items from the
consensus analysis differ between groups
This actually confounds overall cultural
consensus with whatever differences exist
Hruschka, et al. (2008)
First, they confirm that there is an overall
cultural consensus
Second, using three different techniques, they
demonstrate that there are different patterns of
response within groups
Third, they compare cultural answer keys
calculated within groups
Does not account for overall cultural consensus
Boster and Johnson (1989)
Use the second factor
This captures agreement left over after the
overall consensus has been accounted for
Plotting respondents by cultural competence
coefficients and second factor loadings—or
“residual agreement coefficients”—shows the
shape of divergence from the overall consensus
Distribution of cases by cultural competence
and residual agreement
Eigenvalue ratio = 6.7
Mean competence = 0.70
s.d. = 0.11
(First Factor)
(Second Factor)
Distribution of cases by cultural competence and residual agreement
Blue = 2001
Red = 2011
p < .001 for residual agreement
Residual agreement 
Residual agreement 
(Dressler et al. 2015)
(Dressler et al. 2015)
OVERALL
CULTURAL
CONSENSUS
Shared deviation
 from consensus
 within subgroup
1
Shared deviation
 from consensus
 within subgroup
2
“Residual agreement”
Calculating deviation from the overall
Calculating deviation from the overall
cultural consensus
cultural consensus
Dev1 = individual rating of item 1 – consensus rating of item 1
Dev2 = individual rating of item 2 – consensus rating of item 2
…..
Dev
N = 
individual rating of item 
N
 – consensus rating of item 
N
Deviations averaged
 over individuals in
subgroup
1
Deviations averaged
over individuals in
 subgroup
2
compute d1 = ls1 - 3.07.
compute d2 = ls2 - 3.23.
compute d3 = ls3 - 1.64.
compute d4 = ls4 - 3.02.
compute d5 = ls5 - 3.9.
compute d6 = ls6 - 2.6.
compute d7 = ls7 - 3.30.
compute d8 = ls8 - 3.92.
compute d9 = ls9 - 2.06.
compute d10 = ls10 - 3.11.
compute d11 = ls11 - 3.84.
compute d12 = ls12 - 3.6.
compute d13 = ls13 - 2.96.
compute d14 = ls14 - 2.57.
compute d15 = ls15 - 3.3.
compute d16 = ls16 - 3.82.
compute d17 = ls17 - 2.38.
compute d18 = ls18 - 2.78.
compute d19 = ls19 - 2.04.
compute d20 = ls20 - 3.34.
compute d21 = ls21 - 1.91.
compute d22 = ls22 - 3.88.
compute d23 = ls23 - 2.14.
compute d24 = ls24 - 3.77.
compute d25 = ls25 - 2.84.
compute d26 = ls26 - 3.65.
compute d27 = ls27 - 2.6.
compute d28 = ls28 - 2.56.
compute d29 = ls29 - 2.45.
compute d30 = ls30 - 3.65.
compute d31 = ls31 - 3.66.
compute d32 = ls32 - 2.73.
Results of the analysis of residual agreement
Results of the analysis of residual agreement
I
tems rated more
 important in 2011
Items rated more
 important in 2001
<
>
I
tems rated more
 important in 2011
Items rated more
 important in 2001
Respondents in 2011 rated as
   more important than the overall
   consensus items associated
   with information technologies,
   especially cell phones and 
   internet access
Respondents in 2001 rated as
  more important than the overall
  consensus items associated with
  traditional Brazilian sociality, 
  especially spending time in 
  venues associated with social
  interaction
Cultural Consonance
Cultural consonance is
the degree to which
individuals approximate,
in their own beliefs and
behaviors, the
prototypes for belief
and behavior encoded
in shared cultural
models
Cultural models
    Cultural consonance
          Health outcomes
More important
Less important
Calculating Cultural Consonance in Lifestyle
1.
 No      0
2.
 Yes    1
3.
 No      0
4.
 No      0
5.
 Yes    1
6.
 No      0
7.
 No      0
8.
 Yes    1
9.
 Yes    1
10.
 Yes    1
11.
 Yes    1
12.
 Yes    1
13.
 Yes    1
14.
 No      0
15.
 No      0
16.
 No      0
17.
 Yes    1
18.
 No      0
Total =        9
9/18=       .50
Responses of Respondent 10042
Pile sort - 2001
Pile sort - 2011
Organization’
‘Affect’
‘Violence’
‘Education’
Organization’
‘Affect’
‘Violence’
‘Education’
‘A good family’
‘A good family’
Constructing and scaling survey items to measure
cultural consonance in family life
Each of core concepts 
from cultural domain
Group discussion of possible
   survey research items
   within research staff
18 Likert-response
  items gauging the
  respondent’s appraisal
  of her own family
Scale scores = sum of responses
  
weighted by
 
the importance of
  the concept from cultural 
  consensus analysis
Cronbach’s alpha = .87
Description of Brazilian families
   using free listing and focus groups
Exploration of dimensions
   of meaning using pile sorts and focus groups
Confirmation of consensus
  around principal dimension
  of value
Development of scale of 
   cultural consonance based 
   on consensus meaning 
   of terms
 
A straight line from natural speech acts
to measurement = 
emic validity
r = - .65
Cultural consonance in life goals
r = - .65
The End
Slide Note
Embed
Share

Explore the essence of Cultural Consensus Analysis (CCA) as a significant aspect of psychological anthropology. Through ethnographic work and structured techniques, CCA helps verify shared knowledge within cultural domains and enhance the comprehension of cultural models. The model and steps involved in studying cultural domains are outlined, emphasizing the importance of CCA in exploring cultural configurations and shared understanding.

  • Cultural Consensus Analysis
  • Psychological Anthropology
  • Ethnographic Work
  • Shared Knowledge
  • Cultural Models

Uploaded on Sep 19, 2024 | 1 Views


Download Presentation

Please find below an Image/Link to download the presentation.

The content on the website is provided AS IS for your information and personal use only. It may not be sold, licensed, or shared on other websites without obtaining consent from the author.If you encounter any issues during the download, it is possible that the publisher has removed the file from their server.

You are allowed to download the files provided on this website for personal or commercial use, subject to the condition that they are used lawfully. All files are the property of their respective owners.

The content on the website is provided AS IS for your information and personal use only. It may not be sold, licensed, or shared on other websites without obtaining consent from the author.

E N D

Presentation Transcript


  1. Cultural Consensus Analysis Bill Dressler & Kathy Oths Society for Psychological Anthropology 2017

  2. Basic Outline of (very brief) Workshop Review basics of the model Quick overview of techniques Studying variation Cultural consonance

  3. Steps in Studying Cultural Models Determine cultural domain to be studied Identify salient elements of the domain Free lists Open-ended interviews Explore the structure of the domain Pile sorts, ratings, rankings Narrative analysis Test for Cultural Consensus Explore distribution of cultural knowledge and content of shared understanding

  4. Point to remember. Cultural consensus analysis is not the beginning Rather, it is the culmination of ethnographic work, usually supplemented by various forms of structured ethnographic techniques CCA then enables you to: Verify that knowledge is shared within a domain Explore and better understand the configuration of the cultural model(s)

  5. Review of the Basics of the Model

  6. The Cultural Consensus Model (Romney, Weller and Batchelder 1986) Responses of Individuals Determine R s relative degree of shared knowledge R1 R2 R3 R4 R5 R6 . . Rn Degree of sharing of knowledge Calculate a consensus set of responses, weighted by pattern of knowledge sharing

  7. 2-dimensional array of food similarities/differences: 2010 - 2014

  8. Meals

  9. Health

  10. Individual representation = [cultural model] + [personal model] What are the features that distinguish among the elements of the model? Is there a shared understanding of these features?

  11. Pure personal biography: Adriane individual experiences with food .83 Adriane Greg individual experiences with food .83 Why? Shared cultural model: Greg Adriane knowledge1 Cultural Model of Food .83 Greg knowledge2 Individual correlation = knowledge1 x knowledge 2

  12. Patterns of agreement: Meals Ratio indicates overall consensus, if > 3.0 Cultural competence coefficients indicate how well your answers reflect group-level answers Average cultural competence indicates the strength of the cultural consensus

  13. Kim .95 Robin .98 Adriane .93 .98 Bryanna Cultural model of how to construct a meal .93 Kelso .92 Diedre .90 Alec .76 Heidi .78 .89 Joshua Greg

  14. Consensus ratings of foods

  15. Consensus ratings = a cultural best estimate How would a reasonably culturally competent member of this social group rate or classify these foods?

  16. Fitting attributes to the array How do we know if people are using features to define semantic structures? Correlate the features with the distance between the points in the map of the pile sort If the correlation is high (> .75), then probably people were using those features to sort the terms

  17. ? Food Meal Health Southern TEA MASHP STEAK BURGERS SANWICH HOTDOG CORN CATFISH CHEESE APPLE FF COOKIE FCHICK RICE WING BEER ICREAM ORANGE SALAD PIE SHRIMP GRITS MNCHEES BEANS PEAS BBQ PIZZA BANANA OKRA TOMATO 4.00 2.00 1.00 1.00 1.00 1.00 2.00 1.00 4.00 5.00 2.00 3.00 1.00 2.00 1.00 4.00 3.00 5.00 2.00 3.00 1.00 2.00 2.00 2.00 2.00 1.00 1.00 5.00 2.00 4.00 1.76 2.64 2.71 1.77 2.76 1.49 3.48 3.06 2.71 3.93 1.21 1.06 1.31 3.16 1.49 1.51 1.17 4.00 3.76 1.27 3.58 2.64 1.85 3.68 3.79 1.87 1.65 3.82 3.78 3.82 4.00 3.86 3.47 3.31 2.66 2.38 3.84 3.76 3.56 2.58 2.94 2.69 3.97 3.08 2.78 3.25 2.85 1.80 2.63 3.88 3.23 4.00 3.67 3.63 3.32 3.90 2.06 2.03 3.87 3.46

  18. Fitting the attributes Meal correlation = .91 Health correlation = .91 Southern correlation = .20

  19. 2-dimensional array of food similarities/differences: 2010 + 2011 Where does it go on my plate? Is it healthy?

  20. Quick Overview of Techniques

  21. Doing Cultural Consensus Analysis In Anthropac The formal process model The informal data model In a standard statistical package: SPSS Formal process model (MTTIW) Informal data model

  22. Studying Variation

  23. Classic Brazilian Food Feij o/Feijoada Rice Churrasco Cerveja (beer) Macar o (pasta) Hearts of palm Pintado (catfish) Doces (sweets) Sucos/vitaminas Almo o (lunch) Launches Salgadinhos/picados Comida mineiro Comida baiano Comida carioca Comida ga cho

  24. The Prestige Value of Food in Urban Brazil (Oths, Carolo, & dos Santos 2003) Sample Eigenvalue ratio Mean ( sd) Total sample 1.8:1 .40 ( .35) Upper-middle class 10:1 .69 ( .06) Middle class Single factor .54 ( .04) Lower middle class Single factor .81 ( .03) Lower class 1.8:1 .50 ( .28)

  25. Testing variation with separate model estimates within groups (Oths, Carolo, & dos Santos 2003)

  26. Cultural Modeling - The Domain of Family Life false 1.5 exploit selfish disrespect unfaithful violent addiction 1.0 irresponsibile bad manners fights Violence/addiction Lack of education .5 critical Importance for having a family 0.0 firm organized -.5 communication workers good relations happy help outgoing understanding union honest -1.0 religion love Family organization Affective climate -1.5 -.6 -.4 -.2 -.0 .2 .4

  27. Distribution of Cultural Competence in Family Life urban Brazil Eigenvalue ratio = 8.49

  28. Recommended cut-off points for the S.D. of cultural competence (Hruschka and Maupin 2013)

  29. Relationship of years in Nairobi and cultural competence in the cultural model of HIV+ management (Copeland 2011) 0.9 0.85 CULTURAL OMPETENCE 0.8 0.75 0.7 0.65 0.6 0.55 0.5 0.45 0.4 < 5 yrs 5-9 yrs 10-19 yrs > 19 yrs YEARS IN NAIROBI

  30. Cultural competence in family life and psychological distress in urban Brazil (n = 52) Step 1 -.073 -.163 -.343* Step 2 -.033 -.161 -.288* Step 3 .035 -.056 -.252* Age Sex Socioeconomic status Cultural competence in family life Cultural consonance in family life Multiple R - -.282* -.197* - - -.461* .378* .467* .635*

  31. Residual Agreement Concept introduced by Boster in 1986 Refers to agreement among respondents beyond an overall cultural consensus Multicentric domain Defined by Caulkins and Hyatt Multicentric domains have multiple centers of agreement: Subcultural domains, in which there are two or more centers of agreement that are different but not oppositional Contested domains, in which some individuals take a perspective opposite to that expressed by others in the same population

  32. Bosters (1986) original approach Works from the agreement matrix Run consensus analysis and then generate the predicted agreement Subtract the predicted agreement matrix from the observed agreement matrix to generate a residual agreement matrix Use QAP (quadratic assignment procedure) to test for correlation of the observed and residual agreement matrices If matrices are correlated, then there is residual agreement not exhausted by the consensus

  33. Bosters original approach (cont.) This can be thought of as an omnibus test of residual agreement Just tells you it s there Not what it is Utility of this test may not be great, especially if there is always going to be some kind of residual agreement a reasonable assumption

  34. Norbert Ross & Doug Medin In essence, they use the same approach as Boster to test for residual agreement They calculate residual matrices within groups and compare them Then, they identify which items from the consensus analysis differ between groups This actually confounds overall cultural consensus with whatever differences exist

  35. Hruschka, et al. (2008) First, they confirm that there is an overall cultural consensus Second, using three different techniques, they demonstrate that there are different patterns of response within groups Third, they compare cultural answer keys calculated within groups Does not account for overall cultural consensus

  36. Boster and Johnson (1989) Use the second factor This captures agreement left over after the overall consensus has been accounted for Plotting respondents by cultural competence coefficients and second factor loadings or residual agreement coefficients shows the shape of divergence from the overall consensus

  37. Distribution of cases by cultural competence and residual agreement Eigenvalue ratio = 6.7 Mean competence = 0.70 s.d. = 0.11 (Second Factor) (First Factor)

  38. Distribution of cases by cultural competence and residual agreement Blue = 2001 Red = 2011 p < .001 for residual agreement

  39. Residual agreement (Dressler et al. 2015) Shared deviation from consensus within subgroup1 OVERALL CULTURAL CONSENSUS Residual agreement Shared deviation from consensus within subgroup2

  40. Calculating deviation from the overall cultural consensus Dev1 = individual rating of item 1 consensus rating of item 1 Dev2 = individual rating of item 2 consensus rating of item 2 .. DevN = individual rating of item N consensus rating of item N Deviations averaged over individuals in subgroup1 Deviations averaged over individuals in subgroup2

  41. compute d1 = ls1 - 3.07. compute d2 = ls2 - 3.23. compute d3 = ls3 - 1.64. compute d4 = ls4 - 3.02. compute d5 = ls5 - 3.9. compute d6 = ls6 - 2.6. compute d7 = ls7 - 3.30. compute d8 = ls8 - 3.92. compute d9 = ls9 - 2.06. compute d10 = ls10 - 3.11. compute d11 = ls11 - 3.84. compute d12 = ls12 - 3.6. compute d13 = ls13 - 2.96. compute d14 = ls14 - 2.57. compute d15 = ls15 - 3.3. compute d16 = ls16 - 3.82. compute d17 = ls17 - 2.38. compute d18 = ls18 - 2.78. compute d19 = ls19 - 2.04. compute d20 = ls20 - 3.34. compute d21 = ls21 - 1.91. compute d22 = ls22 - 3.88. compute d23 = ls23 - 2.14. compute d24 = ls24 - 3.77. compute d25 = ls25 - 2.84. compute d26 = ls26 - 3.65. compute d27 = ls27 - 2.6. compute d28 = ls28 - 2.56. compute d29 = ls29 - 2.45. compute d30 = ls30 - 3.65. compute d31 = ls31 - 3.66. compute d32 = ls32 - 2.73.

  42. Results of the analysis of residual agreement internet 0.30 Items rated more important in 2011 < washmachine cellphone Mean deviations from the consensus answer key - 2011 0.20 listenmusic stereo microwave car read useinternet refigerator 0.10 sports church computer theater house rest study 0.00 stove tv lunchout converse -0.10 school movies parties telephone -0.20 watchtv Items rated more important in 2001 bar > shopping club furniture -0.30 -0.40 -0.30 -0.20 -0.10 0.00 0.10 0.20 Mean deviations from the consensus answer key - 2001

  43. Respondents in 2011 rated as more important than the overall consensus items associated with information technologies, especially cell phones and internet access Items rated more important in 2011 internet 0.30 washmachine cellphone Mean deviations from the consensus answer key - 2011 0.20 listenmusic stereo microwave car read useinternet refigerator 0.10 sports church computer theater house rest study 0.00 stove tv Respondents in 2001 rated as more important than the overall consensus items associated with traditional Brazilian sociality, especially spending time in venues associated with social interaction lunchout converse -0.10 school movies parties telephone Items rated more important in 2001 -0.20 watchtv bar shopping club furniture -0.30 -0.40 -0.30 -0.20 -0.10 0.00 0.10 0.20 Mean deviations from the consensus answer key - 2001

More Related Content

giItT1WQy@!-/#giItT1WQy@!-/#giItT1WQy@!-/#giItT1WQy@!-/#giItT1WQy@!-/#giItT1WQy@!-/#