Geometric Data Analysis. Johs. Hjellbrekke & Olav Korsnes Dep. - - PowerPoint PPT Presentation

Cultural Distinctions: A Geometric Data Analysis.

• Johs. Hjellbrekke & Olav Korsnes
• Dep. of Sociology

University of Bergen

The critique(s) against Bourdieu’s ”Distinction” (1979)

• Outdated; may have been right in the 1960-ies,

but things have changed since then

• Not relevant outside the French society
• Deterministic – habitus as a ”trojan horse” for

determinism

• Individualization-processes
• Popular culture is much more autonomous from

the legitimate culture than what Bourdieu claims; it has its own hierarchies and specific forms of capital.

Some important contributions

• Beck – individualization thesis; social hierarchies

less and less important

• Lamont – moral and symbolic boundaries,

boundary work, evaluative repertoires

• Peterson – Omnivore-univore, the homology

thesis

• Bryson – patterns of rejection/dislike
• Chan & Goldthorpe – Class or status,

”strong”/strict homology or ”loose” homologies?

• Hestholm, Sakslind & Skarpenes: no moral

evalutions of culture and cultural distinctions

Varieties in Societal Perceptions. ISSP1999 Social Inequality III

Norway Sweden Germany (West) France USA

An elite at the top, few in the middle, many at the bottom

3,1 10,3 10,6 12,4 16,2

A society that looks like a pyramide, with an elite at the top, more in the middle, and most at the bottom

10,9 23,9 26,7 49,8 30,6

A pyramide, but with few people at the bottom.

19,3 27,6 25,5 23,1 17,9

A society where most people are in the middle.

56,0 33,0 23,9 12,9 25,3

Many people near the top, only very few at the bottom

7,7 1,3 2,0 0,8 2,7

Can’t choose

3,2 3,9 11,3 0,9 5,4

Total

100 100 100 100 100

Our approach

• Cultural distinctions through reported practices
• What correspondence can be found between the
• ppositions within this space, and oppositions between

positions/groups in the social space?

• I.e.: can a correspondence between positions in social

hierarchies and hierarchies of cultural attendance/practices/preferences be found for the Norwegian case?

• How many groups/clusters can be identified, and what

are their profiles with respect to cultural practices?

• Are the internal oppositions in these clusters the same

as in the global space?

Construction of the space – analytical strategy

• Data set: Culture and Media Survey 2008, Central Bureau of

Statistics

• N=1194 (respondents 24-70 yrs old, with a registered occupation)
• Personal interviews
• Active variables = 44
• Total # of active categories = 102
• Total # of passive categories = 0 (!)
• Standard Multiple Correspondence Analysis (NOT Specific MCA [cf.

Le Roux & Rouanet 2004, 2010)

• Ascending Hierarchical Cluster Analysis
• Structured Data Analysis
• Class Specific MCA of a selected subgroup

But before we start

• Let’s remind ourselves…
• ”In the analysis of questionnaires, it is not

enough to do a correspondence analysis to do analyses à la Bourdieu. The fundamental social space must be constructed from an extensive set of relevant variables and ample enough to allow the full multidimensional display of individuals.” (Rouanet & al. 2000)

Variables

• Attendance of cultural

arrangements/events last 12 months

• Types of events/music types/expositions
• Q = 14
• K= 40

Variables

• TV-preferences; channels and types of

programs/shows

• Program types: Q=9
• Channels: Q=6
• K=30
• I.e: Binary coded variables

Radio

• Channels
• Q=6
• K=12
• Binary variables

Newspapers

• Q=9
• K=20

Axes: Eigenvalues and modified inertia rates

Eigenvalue Percentage Percentage, Benzécri’s modified rates Cumulated percentages, modified rates

Axis 1 .0923 7.00 55.8 55.8 Axis 2 .0602 4.57 16.2 72.0 Axis 3 .0530 4.02 10.6 82.6

Contributions from blocks of variables

Cultural atten- dence

TV, channels & programs

Radio,

Channels & programs

Newsp Incl. internet Axis 1 86.6 1.3 3 9.1 Axis 2 10.2 74.1 3.7 12.0 Axis 3 31.0 24.8 14.6 29.6

The Cloud of Individuals, fac.plane 1-2

Contributions to axis 1

+ Active/ Engaged Highbrow +Inactive /Dis- engaged

Contributions to Axis 2

+ TV/Media

• TV/Media

The Cloud of Individuals, fac.plane 1-3

Contributions Axis 3

+ ”New” Media: Internet/Emerging + Traditional Media: TV/Newspapers, +Highbrow

The Cloud of Individuals, fac.plane 2-3

Fac.plane 2-3

Structuring factors

• Age
• Class
• Education
• Sex

Axis 1: EGP-classes

+Inactive/Disengaged + Manual Working Class + Active/Engaged Highbrow + Service Class

Axis 1: Educational level

+ Active/ Highbrow Higher educ. levels +Inactive/ Disengaged + Lower educ levels

Axis 2: (Minor) Sex (Differences) Internal opposition at lower secondary educational levels

• TV/Media

+ TV/Media

Axis 3: Age

+ Active/ Highbrow +Inactive Younger respondents Older respondents + Traditional Media: TV/Newspapers, +Highbrow + ”New” Media: Internet/Emerging

Subgroups (clusters) within this space

• Ascending hierarchical cluster analysis
• Ward’s method
• Done on total inertia, i.e. all the axes in the

analysis (all the dimensions in the space)

• Identified on the basis of similarities/dis-

similiarities across the active set of variables

• Calculated on the basis of the individuals’

axis coordinates on all dimensions

How many clusters?

• How many clusters?
• What are their profiles?
• How large are they?
• Do they intersect in the factorial planes?
• 6 retained for interpretation

Cluster 1: 26%. Actives/Engaged.

Over-represented categories, active variables Over-represented categories, supplementary variables

1-4 art exhib. 12 months Artium – highest gen educ One type of art exhibition Women 1-4 theater visits 12 mnths HED 3-4 years One type of concerts EGP 1 (Higher service) 1-4 concerts 12 mnths

Underrepresented:

1-4 museum visits 12 mnths No univ. educ Men

Cluster 2: 10,5%. Hyper- actives/hyper-engaged

Over-represented categories, active variables Over-represented categories, supplementary variables

5+ theater Artium – gen educ level 5+ art exhibitions HED 5-6 yrs 5+ museums NOK 1 mill+ HH income 5+ concerts HED 7yrs+ 5+ ballet/dance EGP 1 Higher service 5+ cinema

Underrepresented

National newspapers P2 – national radio No university educ.

Cluster 3: 4,7%. TV - traditional

Over-represented categories, active variables Over-represented categories, supplementary variables

TV Nature No university education TV Other NRK 1 Underrepresented: TV News HED 3-4 yrs NRK 2

Cluster 4: 8%. TV-/Media cluster

Over-represented categories, active variables Over-represented categories, supplementary variables

NRK2 Educ level: Realskole TV debate Age group 3 – 45-64 yrs old NRK1 TV news TV info soc Underrepresented: P2

Age group 2 – 24-43 yrs old

2 types of concerts 1-4 museum visits 12 mnths

Cluster 5: 25.1%. Music listeners, radio/concerts

Over-represented categories, active variables Over-represented categories, supplementary variables

No art exhib/types Age group 2 1 type of concert Men 1-4 concerts No TV debate No NRK2 Underrepresented:

Age group 3 – 44-64 yrs

Cluster 6: 25,8%. Inactives

Over-represented categories, active variables Over-represented categories, supplementary variables

No concerts/types No art exhibitions/types Oldest age group: Age group 4 – 65 yrs + No theater No cinema No university education No museums No and Low incomes: - 200’ 1 festival Underrepresented: No library/library books HED 3-4 yrs, 5-6 yrs 7yrs+

6 clusters in fac.plane 1-2

4 of 6 clusters in fac.plane 1-3

Step 3: Class Specific Analysis (CSA)

• Specific MCA: Developed by Brigitte Le Roux,

restriction of analysis to the categories of interest (Le Roux & Rouanet 2004)

• Class Specific MCA: Developed by Brigitte Le

Roux (see Le Roux & Rouanet 2004, 2010) in

• rder to study a class of indivuals/cases with

reference to the complete sample/population, i.e. an analysis of a subcloud with reference to the global cloud.

• What are the principal axes of the subcloud?

How are they interpreted, compared to the the axes in the global cloud?

CSA

• CSA of cluster 2: The Hyper-Active/Hyper-

Engaged

• Internal oppositions at the top of the

capital hierarchy

• High vs. Low frequentation?
• Omnivores vs. Univores?
• Univores vs. Univores?

Factorial plane 1-2, cloud of individuals

Contributing points – axis 1 10 categories – 91.6% of the contribution

+ Theater + Museum + Art exhib

Contribution points axis 2 18 categories – 88.1% of the contribution

Omnivorousness – art & music Univorousness – art & music ?

Opera - NO

Factorial plane 1-3

Factorial plane 2-3

Contributing points Axis 3 17 categories – 83.2% of the contribution

+ Museum visits + Radio/Internet

• Thank you for your attention!

Basic references

• Bourdieu, Pierre (1979). La Distinction. Critique

Sociale du Jugement. Paris: Ed. De Minuit.

• Le Roux, Brigitte & Rouanet, Henry (2004).

Geometric Data Analysis. From Correspondence Analysis to Structured Data Analysis. Dordrecht: Kluwer - Springer.

• Le Roux, Brigitte & Rouanet, Henry (2010).

Multiple Correspondence Analysis. #163, QASS-

• series. Thousand Oaks: Sage.