Professional Documents
Culture Documents
University of Oxford: Seminar, Belgrade University, 2l November, 2011
University of Oxford: Seminar, Belgrade University, 2l November, 2011
Beyond outgroup attitudes/prejudice as outcome variable From cross-sectional to longitudinal research Multi-level analyses A new form of extended contact
Positive contact with a member of another group (often a negatively stereotyped group) can improve negative attitudes: -- not only towards the specific member, --but also towards the group as a whole
Quantity of contact frequency of interaction with outgroup members, e.g., how often do you meet/talk to/etc. outgroup members where you live/shop/socialize, etc? Quality of contact nature of the interaction with outgroup members, e.g., how positive/negative; friendly/unfriendly, etc, is the contact?
Equal status Stereotypes are disconfirmed Cooperation Situation allows participants to get to know each other properly Norms support equality (later: cross-group friends)
Number of Studies: 515 studies Participants: 250,089 people from 38 nations Studies range across many: Disciplines Intergroup contexts Prejudice indicators Reliable effect: more contact, less prejudice
10
11
Area A1
Costa
Day
1 1 89 1
Kitchen
Area A2
Area A4
Area A3
Asian
White Black Other
Area A1
Costa
Day
1 1 89 1
Kitchen
Area A2
Area A4
Area A3
All Asian
30.91%
White/ Asian
Black / Other
0.33%
4.18%
4.73%
Day 1 1 1 1 2 2 2 2
Area 1 2 3 4 1 2 3 4
No. of Intervals 10 10 10 10 12 12 12 12
I
-1.99 -0.71 -0.29 -1.09 -1.6 -0.39 -0.44 -1.19
Lower Limit -4.36 -2.57 -1.48 -3.04 -3.82 -1.52 -3.44 -3.02
Note: I denotes aggregation index (negative values indicate more ethnic clustering/less cross-ethnic mixing than expected from random distribution). Especially, Area 1 . . .
Area A1
Costa
Day
2 7 142 4
Kitchen
Area A2
Area A4
Area A3
Day
Area
No. of intervals
Number of Whites
Number of Asians
1 1 1 1
1 2 3 4
10 10 10 10
366 16 9 46
2
2 2 2
1
2 3 4
12
12 12 12
241
278 182 381
461
16 13 36
Area A1 had a higher proportion of Asians than the other areas. We then compared area A1 to each other area, in turn, to see whether ethnicity and area were related.
Asian and White students were sig. unevenly distributed across the cafeteria
21
Studying contact over time From individuals to neighbourhoods Extended or indirect contact
Multi-level analyses
22
23
Explicit attitudes Implicit attitudes Neural processes* Trust and forgiveness Outgroup-to-outgroup generalization: the secondary transfer effect. * Behavioural outcomes: helping *
24
We compared neural mechanisms of face processing in different social contexts (different races, different facial expressions of emotion): We investigated Event-Related Potentials (ERPs) Own and other race (White/Black) faces Emotional faces: happy/angry/neutral
25
EEG Recording
26
Emotional Modulation
White Faces
10 V PO8 Neutral Happy Angry
P300
-100 0 ms 200 400 600 ms
Both components (N170, P300) modulated by race and emotion The greater the contact, the smaller the difference between own- and other-race face processing
N170
-10 V
Black Faces
Neutral Happy Angry
27
Are the effects of contact with members of one group restricted to that outgroup, or do they have knock-on or trickle-down effects on attitudes towards other groups?
Contact with and prejudice towards Primary outgroup: immigrants Prejudice towards Secondary outgroup:
Jews Gays/homosexuals Homeless people
National cross-sectional samples (N = 1,000) in 8 European countries Investigated STE controlling for ideology of inequality (SDO)
29
Anti-semitism .31***
R2 =.37
Antiimmigrant attitudes
.46***
.30***
.32***
Negative attitudes - homeless
SDO .13**
Significant indirect effects of contact (full mediation via attitude generalization) on anti-semitism ( = -.06, p =.001), and attitudes to homosexuals ( = -.06, p < .001) and homeless ( = -.06, p < .001); controlling for age, gender, education, income. Model fit: 2=243.39, df = 95, p < .001, 2/df=
2.56, CFI = .97, RMSEA = .04, SRMR = .03
30
31
32
1.76*
.43***
p < .05;
**
p < .01;
***
p < .001
33
34
Respondents from 15 European countries Data collected in the 1980s Retrospective case-control sample: Case sample of identified rescuers (N=346, recognized by Yad Vashem, The Holocaust Martyrs and Heroes Remembrance Authority, as Righteous among the Nations) Control sample matched on age, sex, education, region (N=164) Final sample = 297 rescuers, 115 non-rescuers
Pre-war friendships with Jews increase the probability of rescuing Jews (especially Jewish friends)
36
3. Empirical Application
Multinomial logistic regression (variables coded [0,1]):
12.19**
2.24**
Notes: N = 412. Coefficients are odds ratios. No control variables. + p < 0.10, * p < 0.05, ** p < 0.01.
Odds ratio (OR) calculated shows the probability of helping (Jewish friends; other Jews) vs not-helping OR of 12.19 in previous table 1 means:
If you had pre-war Jewish friends, the probability of Helping Jewish Friends divided by the probability of not-Helping was 12.19 times higher than if you did not have pre-war Jewish friends.
The odds of helping other Jews vs. not helping increase (only) by a factor of 2.24 if respondents had Jewish friends before the war. Less technically:
Having Jewish friends before the war made potential rescuers more likely to help, especially to help Jewish friends, but also to help other Jews.
3. Empirical Application
15.41**
1.07** 18.43**
2.89**
1.05* 5.18*
Command zone
Size Jewish population Number of rooms Many neighbours
10.89**
0.98 15.28** 0.86
10.36**
1.19** 12.11** 0.39*
Notes: N = 412. Coefficients are odds ratios. Additional control variables: gender; education level; religiosity; religious confession; SS zone, Jewish Neighbours, partner/children in household, financial resources. + p < 0.10, * p < 0.05, ** p < 0.01.
40
41
Age (yrs):
T1: Mean (SD) = 14.68 (1.06) T2 (+ 6 mths): Mean (SD) = 15.31 (1.03) T3 (+ 6 mths): Mean (SD) = 15.67 (1.05)
Variables:
Predictors: cross-group friends Mediators: intergroup anxiety; empathy Outcomes: positive outgroup attitudes; outgroup variability; negative action tendencies
3-waves permit mediation analyses Time 1 predictor -> Time 2 mediator -> Time 3 Outcome
42
Outgroup Friendships
x1 x2
Outgroup Friendships
y1 y2
Outgroup Friendships
y19 y20
Intergroup Anxiety
x3 x4 x5 y3
Intergroup Anxiety
y4 y5 y21
Intergroup Anxiety
y22 y23
Empathy
x6 x7 x8 y6
Empathy
y7 y8 y24
Empathy
y25 y26
Outgroup Friendships
Outgroup Friendships
-.14** -.11**
Intergroup Anxiety
-.14**
-.11**
Outgroup Friendships
.13** -.14**
Intergroup Anxiety
Intergroup Anxiety
-.14**
Empathy
.13**
Empathy
Empathy
.23*** .23***
.23***
.23*** .15**
.15**
Perceived outgroup Variability Perceived outgroup Variability
-.15**
Negative Action Tendencies
-.27***
-.15**
-.27***
Negative Action Tendencies
44
Green paths are autoregressive. Blue paths are 'forward' paths (as predicted by contact model: contact->mediators->prejudice / mediators->prejudice / contact>mediators). Red paths are 'reverse' paths (prejudice->mediators->contact / prejudice->mediators / mediators->contact). Model equates paths from Wave 1-2, and 2-3 All paths indicated are significant.
45
Outgroup Friendships
Outgroup Friendships
-.14** -.11**
Intergroup Anxiety
-.14**
-.11**
Outgroup Friendships
.13** -.14**
Intergroup Anxiety
Intergroup Anxiety
-.14**
Empathy
.13**
Empathy
Empathy
.23*** .23***
.23***
.23*** .15**
.15**
Perceived outgroup Variability Perceived outgroup Variability
-.15**
Negative Action Tendencies
-.27***
-.15**
-.27***
Negative Action Tendencies
46
Outgroup Friendships
Outgroup Friendships
-.14**
Outgroup Friendships
-.11**
Intergroup Anxiety
.13**
Intergroup Anxiety
Intergroup Anxiety
-.14**
Empathy Empathy
Empathy
.23***
.23*** .15**
-.15**
Negative Action Tendencies Negative Action Tendencies
-.27***
Negative Action Tendencies
47
48
Percentage of Out-groupers
Higher Prejudice
Percentage of Out-groupers
Out-group friends
Lower Prejudice
49 49
50
In colloquial language, people living in ethnically diverse settings appear to hunker down that is, to pull in like a turtle. (Putnam, 2007, p. 149)
Measures of Diversity
Missing or inappropriate measures of intergroup contact
Putnam uses high-threshold measure of contact (friends) Does not test whether contact mediates diversity effect
52
+
Diversity
(= opportunity for contact)
Intergroup Contact
+
Social cohesion
- (?)
53
(PIs: M. Hewstone, A. Heath, C. Peach, S. Spencer post docs: A. Al Ramiah, N. Demireva, S. Hussain, K. Schmid)
Test of integrated model of group threat theory and contact theory, to examine relationship between macro-level diversity and individual-level interethnic attitudes Sampled respondents from neighbourhoods of varying degrees of ethnic diversity
Control for additional key macro-level variable: neighbourhood deprivation
54
Cross-sectional survey of nationally representative sample in England (CAPI) Random location quota sampling, based on stratified design (grid = ethnic density X deprivation) Sample = 1666, of which N = 868 White British respondents (418 males, 450 females, age range: 16-97, Mage = 47.74, SD = 19.14)
55
e.g., What proportion of people with an ethnic minority background live in your neighbourhood? (1- none, 5 a lot)
e.g. How often, if at all, do you mix socially with people from ethnic minority backgrounds in your neighbourhood?
Perceived threat (5 items, 1-strongly disagree, 5-strongly agree; see Stephan & Stephan, 2000)
E.g., The more political and economic power ethnic minorities have in this country, the more difficult it is for White British people. E.g., People with ethnic minority backgrounds threaten White British peoples way of life.
How do you feel about White British people/people from ethnic minority backgrounds?
Analysis
Data hierarchically ordered in a two-level structure (respondents nested within neighbourhoods)
Multilevel structural equation modeling to account for both within-level and between-level variance of constructs
57
.45***
Intergroup contact
WITHIN
-.35***
Perceived diversity
ns
Perceived threat
1.13***
Ingroup bias
Intergroup contact
.97***
BETWEEN
-.57*
% Non-White British
.86***
ns
Perceived threat
1.84***
Ingroup bias
IMD
.01***
Model fit: 2 (61) = 105.84, p =.00, CFI = .98, RMSEA = .03, SRMRwithin = .04, SRMRbetween = .09; all parameters are unstandardized.
59
EXTENDED CONTACT Indirect/Vicarious contact, via family or friends, e.g., How many of
your family members/friends have outgroup friends?
60
Some of my friends have friends who are . . . (outgroup members) Extended contact is second-hand, rather than involving the participants in direct intergroup contact themselves
Just knowing other people in your group who have out-group friends might improve attitudes to the out-group (Wright et al., 1997)
61
Extended Contact in Northern Ireland (Results for Catholic and Protestant students; N = 316)
-.18***
.79
.53***
- .03
.17**
.89
In-group bias
Longitudinal analysis of the effects of extended contact at time 1 on direct contact at time 2 (Swart, Hewstone, Tausch et al., in prep.)
Neighbourhood Contact Quantity (Time 2) .15***
Extended Contact (Time 1)
.23***
.21***
Controlling for direct contact scores at Time 1
Meaningful contact is crucial for integration Just living together is not enough (re-segregation problem in cafeteria) Direct contact has reliable effects Longitudinal and multi-level evidence Evidence of mediators Contact has multiple outcomes Secondary transfer; helping Extended contact has reliable effects
Especially for those with no/low direct contact It promotes take up of direct contact But is typically weaker than direct contact
65
Funding
Leverhulme Trust Community Relations Unit (N.I.) Economic and Social Research Council Nuffield Foundation Russell Sage Foundation, U.S.A. Templeton Foundation, U.S.A. Max-Planck Gesellschaft (Uncle Steve)
Research collaborators
Prof. Ed Cairns (University of Ulster) Dr Oliver Christ (University of Marburg, Germany) Prof. Joanne Hughes (University of Ulster) Dr Jared Kenworthy (University of Texas) Clemens Kronenberg (University of Mannheim) Dr Katharina Schmid (University of Oxford) Dr Alberto Voci (University of Padua, Italy) MPI-team, Goettingen!
(ex) Graduate students Maria Ioannou Dr Ananthi al-Ramiah Dr Hermann Swart Dr Nicole Tausch Dr Rhiannon Turner Dr Christiana Vonofakou Dr Pamela Walker
Undergraduate students Eleanor Baker Christina Floe Caroline Povah Elisabeth Reed Anna Westlake
66
APPENDIX SLIDES
67
Putnam (2007); Lancee & Dronkers (2008); Fieldhouse & Cutts (2010)
Marschall & Stolle (2004; Black sample); Fieldhouse & Cutts (2010; ethnic minority sample in UK); Morales & Echazarra (forthcoming)
Marschall & Stolle (2004; White sample); Gesthuizen, van der Meer &
Scheepers (2008); Hooghe et al. (2008)
68
Cross-sectional survey of nationally representative sample in England, using computer assisted personal interviewing techniques Random location quota sampling, based on stratified design (grid = ethnic density X deprivation) Primary sampling unit (PSU): middle layer super output areas (average PSU size: 2250 addresses) 224 PSUs sampled Total N = 1666, of which N = 868 White British respondents (418 males, 450 females, age range: 16-97, Mage = 47.74, SD = 19.14) (Ethnic minority respondents not analysed yet)
69
Chairs Table
= White
= Contiguous groups/isolates (3) = White/Asian adjacencies (2)
= Asian
Example of a colour-coded table. Each chair is coded for the ethnicity of the student, and White-Asian adjacencies and contiguous groups circled and counted for each time interval.
Kappa Coefficient
Total number of people in areas Number of Whites in area Number of Asians in area .86
Significance level
p=<.001
.90
.90
p=<.001
p=<.001