30
About renegades and outgroup-haters: Modelling the link between social influence and intergroup attitudes Andreas Flache Presentation @ ODCD 2017: Interdisciplinary Workshop on Opinion Dynamics and Collective Decision 2017 Jacobs University Bremen, Germany July 5-7, 2017

About renegades and outgroup-hatersodcd2017.user.jacobs-university.de/wp-content/uploads/... · 2017. 7. 13. · • Zigora, T., Stark, T. H., Flache, A. In progress. The Seeds of

  • Upload
    others

  • View
    0

  • Download
    0

Embed Size (px)

Citation preview

  • About renegades and outgroup-haters: Modelling the link between

    social influence and intergroup attitudes

    Andreas FlachePresentation @ ODCD 2017:

    Interdisciplinary Workshop on Opinion Dynamics and Collective Decision 2017

    Jacobs University Bremen, Germany

    July 5-7, 2017

  • About renegades and outgroup haters: Flache - ODCD 2017

    Changing attitudes towards Muslims in the Netherlands

    Do you generally think positively or negatively about muslims in the Netherlands?

    TNS Nipo opinion panel

    Representative random sample N=1089 TNS Nipo. De Beer & de Kraker, 2014

    0

    10

    20

    30

    40

    50

    60

    20102014

    (very) positive

    neutral or no opinion

    (very) negative

    % respondentsanswering …

    2

    year

  • About renegades and outgroup haters: Flache - ODCD 2017

    Changing attitudes about political in-/outgroups in U.S.

    Data Pew Research Center (2014)

    3

    Source: Gentzkow (2016). Polarization in 2016. Stanford University

  • About renegades and outgroup haters: Flache - ODCD 2017

    • Homophily• More likely to interact with

    ingroup members

    • More open to influence from ingroup members• e.g. attraction paradigm,

    symbolic interactionism

    • “ingroup favoritism” (trust, confidence)

    Theory / empiry

    Macy Kitts Flache Benard 2003; Salzarulo 2006; Flache & Mäs 2008; Huet & Deffuant 2010;Mäs & Flache ea 2013; Feliciani ea 2017

    Individual agents differ in “fixed / demographic” traits

    Similar agents interact / influence each other more than dissimilar ones.

    “Fixed” intergroup attitudes tend to foster intergroup dissensus / polarization

    Models

    Intergroup attitudes affect social influence

    4

  • About renegades and outgroup haters: Flache - ODCD 2017

    • Contact theory• Extended contact• Ethnic threat• Research on social influence

    among adolescents in general

    • Intergroup attitudes in particular only recently:• Van Zalk ea.• Zigora, Stark & Flache (in progress)

    Theory / empiry

    • van Zalk, M. H. W., Kerr, M., Kerr, M., van Zalk, N., & Stattin, H. (2013). Xenophobia and tolerance toward immigrants in adolescence: cross-influence processes within friendships. Journal of Abnormal Child Psychology, 41(4), 627–639

    • Zigora, T., Stark, T. H., Flache, A. In progress. The Seeds of Change among Adolescents: How do Intergroup Attitudes spread within Social Networks ?

    All models of attitude / opinion dynamics:

    The attitude / opinion that is socially influenced could also be an intergroup attitude

    But …

    Models

    Intergroup attitudes are socially influenced

    5

  • About renegades and outgroup haters: Flache - ODCD 2017

    The missing link?

    6

    Intergroup attitudes

    Process and outcome of opinion dynamics

    We have models for this

    Intergroup attitudes

    Process and outcome of opinion dynamics

    … and models for that

  • About renegades and outgroup haters: Flache - ODCD 2017

    The missing link?

    7

    Intergroup attitudes

    Process and outcome of opinion dynamics

    But no models for this

  • About renegades and outgroup haters: Flache - ODCD 2017

    Why should we care?

    Social influence on intergroup attitudes mayimprove intergroup attitudesReduced risk of intergroup polarization also on

    other issues

    Social influence on intergroup attitudes may (first) foster intergroup polarization in intergroupattitudesIncreasing risk of intergroup polarization also on

    other issues

    8

  • About renegades and outgroup haters: Flache - ODCD 2017

    Towards an integrated model

    Core assumptions

    If two individuals interact, they modify theirintergroup attitude either towards or away from theintergroup attitude of the other.

    Direction and magnitude of change depend on “discrepancy” with other.

    → which – in turn – depends on intergroup attitude

    9

  • About renegades and outgroup haters: Flache - ODCD 2017

    Towards an integrated model

    10

    • Positive influence on intergroup attitude

    • Negative influence on intergroup attitude

    Small discrepancy attraction / assimilation⇒

    Large discrepancy repulsion / distancing⇒

    Conditions: discrepancy (aka influence weight) depends on:

    opinion disagreement (e.g. about the attitude towards group X)whether actors belong to same group or not (“fixed xenophobia”)attitude towards the group to which “the other” belongs:this attitude is one of the opinions influenced: this is the new element

    e.g. Macy ea 2003; Jager & Amblard, Flache & Mäs 2008 …

  • About renegades and outgroup haters: Flache - ODCD 2017

    ���+1 = ��� + � ���� ���� − ��� �

    ��� +1 = ��� + � ���� ���� − ��� �

    A simple version, more formally

    11

    › We have two groups, group Zero and group One.

    › One opinion / attitude o: attitude towards group One.

    › In one time step, select agents i and j randomly to interact*.

    › If they interact:

    μ scales rate of change, μ≤ 0.5, to assure opinions of i and j don’t cross.

    › The weights w represent “discrepancy” and affect direction and magnitude of influence

    *Deffuant BC interaction rule

    1 1w

    0 1o

  • About renegades and outgroup haters: Flache - ODCD 2017

    “Raw” discrepancy i towards j:

    ���� = �� ��� − ��� + �� �� − �� + �� �� 1 − ��� + (1 − ��)���

    ��+�� + �� = 1, > 0.

    ��= impact of opinion disagreement on discrepancy

    ��= impact of “demographic” difference (�� , �� {0,1} group) on discrepancy

    ��= impact of attitude towards group of j on discrepancy

    Effect of opinion towards group 1 on raw discrepancy:

    Example:

    attitude towardsgroup 1

    Modelling subjective discrepancy

    12

    �� = �

    ��� − ��� =0.5�� = 0

    �� = 1�� =

    ��

    �� =��

    �� =��

  • About renegades and outgroup haters: Flache - ODCD 2017

    “Raw” discrepancy i towards j :

    ���� = �� ��� − ��� + �� �� − �� + �� �� 1 − ��� + (1 − ��)���

    ��+�� + �� = 1, > 0.

    ��= impact of disagreement on discrepancy

    ��= impact of “demographic” difference, “fixed xenophobia” (�� , �� {0,1} group)

    ��= impact of attitude towards group of j on discrepancy

    Effect of opinion towards group 1 on raw discrepancy:

    Example:

    Modelling subjective discrepancy (2)

    13

    �� = �

    ��� − ��� =0.5�� = 0

    �� = 1�� =

    ��

    � �.�

    �� =��

    �� =��� �.�

  • About renegades and outgroup haters: Flache - ODCD 2017

    “Raw” discrepancy i towards j :

    ���� = �� ��� − ��� + �� �� − �� + �� �� 1 − ��� + (1 − ��)���

    ��+�� + �� = 1, > 0.

    ��= impact of disagreement on discrepancy

    ��= impact of “demographic” difference, “fixed xenophobia” (�� , �� {0,1} group)

    ��= impact of attitude towards group of j on discrepancy

    Effect of opinion towards group 1 on raw discrepancy:

    Example:

    Modelling subjective discrepancy (3)

    14

    �� = �

    ��� − ��� =0.5�� = 0

    �� = 1�� =

    ��

    � �.�

    �� =��

    �� =��� �.�

  • About renegades and outgroup haters: Flache - ODCD 2017

    Discrepancy d into influence weight wweight function adopted from Mäs, Flache, Kitts, 2014

    Influence weight

    i is group One

    �� =��

    �� = 0

    �� =��

    �� =��

    �� =��

    �� = �

    j is group 0 (outgroup)

    j is group 1 (ingroup)

  • About renegades and outgroup haters: Flache - ODCD 2017

    So what do we assume?

    • More positive opinion about outgroup

    →more open to influence from outgroup

    →less open to influence from ingroup

    • If outgroup attitude is important and positive

    → possible to be more open to influence fromoutgroup than from ingroup

    16

  • About renegades and outgroup haters: Flache - ODCD 2017

    Experiment 1:

    “Fixed xenophobia” vs influenced outgroup attitude

    Scenario N=100, (g0:50/g1:50)

    Initial “ingroup favoritism”: Experiment:

    17

    �� = 0.4�� = �.��� = �.�

    �� = 0.4�� = �.��� = �.�

    Distribution initial opinion about group 1(two symmetric beta distributions)

  • About renegades and outgroup haters: Flache - ODCD 2017

    Scenario N=100, (g0:50/g1:50). Typical runs.

    Experiment:

    18

    Intergrouppolarization

    Consensus

    �� = 0.4�� = �.��� = �.�

    �� = 0.4�� =.�

    �� = �.�

    Experiment 1:

    “Fixed xenophobia” vs influenced outgroup attitude

  • About renegades and outgroup haters: Flache - ODCD 2017

    Scenario N=100, (g0:50/g1:50). 100 runs per cond.

    Full experiment:

    19

    �� = 0.4�� = �.�

    �� = �

    �� = 0.4�� = �

    �� = �.�

    0.1 0.2 0.3 0.4 0.5 0.6betaA

    0.2

    0.4

    0.6

    0.8

    1.0

    Distance group means:

    Signed distance 1 0

    Absolute distance

    g1-g0

    Intergrouppolarization

    Consensus

    Distribution distancebetween groups in final round

    Experiment 1:

    “Fixed xenophobia” vs influenced outgroup attitude

  • About renegades and outgroup haters: Flache - ODCD 2017

    Why?

    • Initial differences between groups are moderate

    With “mainly fixed xenophobia” even that is enoughfor most agents to be negatively influenced by most outgroup members.

    With “mainly socially influenced” intergroup attitude only agents who think very negatively of outgroup(“outgroup haters”) are negatively influenced bymost outgroup members. But they are the minority.

    20

  • About renegades and outgroup haters: Flache - ODCD 2017

    What if there are more “renegades”?

    21

    Can renegades serve as “linkage” between groupsand prevent intergroup polarization?

  • About renegades and outgroup haters: Flache - ODCD 2017

    What if there are more “renegades”?

    Renegade = “hates ingroup, loves outgroup”

    “Usually, people have a tendency for ingroup bias, that is, they prefer their own group over other groups (e.g. Social Identity Theory)

    Sometimes, however, people have a strong dislike for their own group, e.g., because these groups are stigmatized, … because … an ingroup violates important personal moral standards … labelled … "ingroup hate", "outgroup favoritism", or "disidentification".Becker, J. C., & Tausch, N. (2014). When group memberships are negative: The concept, measurement, and behavioral implications of psychological disidentification. Self andIdentity, 13(3), 294–321.

    22

    Can renegades serve as “linkage” between groupsand prevent intergroup polarization?

  • About renegades and outgroup haters: Flache - ODCD 2017

    Experiment 2: add renegades(make tails of initial distribution longer)

    Scenario N=100, (g0:50/g1:50)

    Increase initial “ingroup favoritism”:

    23

    �� = 0.4�� = 0.2�� = 0.4

  • About renegades and outgroup haters: Flache - ODCD 2017

    Experiment 3: add some “extra renegades”

    Scenario N=100, (g0:50/g1:50)

    “Rewire” per group 10 from “outgroup hater” to

    “renegade”

    24

    �� = 0.4�� = 0.2�� = 0.4

  • About renegades and outgroup haters: Flache - ODCD 2017

    Experiment 4: push it to the extreme

    Scenario N=100, (g0:50/g1:50)

    initially uniformly distributed intergroup attitudes

    (50% “renegades”)

    25

    �� = 0.4�� = 0.2�� = 0.4

    “Regular”intergroup polarization

    “Reversed”intergroup polarization

  • About renegades and outgroup haters: Flache - ODCD 2017

    0.4 0.5 0.6 0.7 0.8 0.9 1.0betaA

    1.0

    0.5

    0.5

    1.0

    How does effect of “renegades” change if we gradually increase impact of intergroup attitude?

    Scenario N=100, (g0:50/g1:50). Initial uniform, 100 runs / c

    Full experiment:

    26

    �� = 1/3�� = 1/3�� = 1/3

    �� = 0�� = 0�� = 1

    g1-g0

    Reversedintergroup polarization

    Consensus

    Distribution distancebetween groups in final round

    Distance group means:

    Signed distance 1 0

    Absolute distance

  • About renegades and outgroup haters: Flache - ODCD 2017

    Do renegades bridge groups?

    Tentative answer based on this model: NO

    Instead, renegades can foster “reversed” intergroup polarization.

    Does this make sense?

    27

  • About renegades and outgroup haters: Flache - ODCD 2017

    Tentative conclusions about this model & what’s next?

    • Socially influenced intergroup attitudes change dynamics

    → can foster intergroup polarization jointly with xenophobia

    • “Renegades” do not bridge groups. Instead:

    → they switch sides, or they foster intergroup polarization

    Some things to do next:

    - Compare effects of negative influence to BC

    - Would you really totally give up your ingroup favoritism underpressure from renegades?

    Introduce some effect of initial position (Friedkin)

    - Interplay intergroup attitudes and other attitudes?

    28

  • About renegades and outgroup haters: Flache - ODCD 2017

    Tentative conclusions about this model & what’s next?

    • Socially influenced intergroup attitudes change dynamics

    → can foster intergroup polarization jointly with xenophobia

    • “Renegades” do not bridge groups. Instead:

    → they switch sides, or they foster intergroup polarization

    Some things to do next:

    - Compare effects of negative influence to BC

    - Would you really totally give up your ingroup favoritism underpressure from renegades?

    Introduce some effect of initial position (Friedkin)

    - Interplay intergroup attitudes and other attitudes?

    29

    Thus: it’s complex … but alsointriuging … to be continuedThank you for your attention

  • About renegades and outgroup haters: Flache - ODCD 2017

    • Flache, Mäs, Feliciani, Chattoe-Brown, Deffuant, Huet, Lorenz. Forthcoming. Models of Social influence: towards the next frontiers. JASSS.

    • Feliciani, Flache & Tolsma. 2017. How, When and Where Can Spatial Segregation Induce Opinion Polarization? Two Competing Models. JASSS. http://jasss.soc.surrey.ac.uk/20/2/6.html

    • Leszczensky, Flache, Stark, & Munniksma. 2017. The Relation between Ethnic Classroom Composition and Adolescents' Ethnic Pride.” GPIR. doi: 10.1177/1368430217691363.

    • Takács, Flache, & Mäs. 2016. Discrepancy and Disliking Do Not Induce Negative Opinion Shifts. PLoS ONE 11(6): e0157948. doi:10.1371/journal.pone.0157948.

    • Munniksma, Verkuyten, Flache, Stark & Veenstra, 2015. Friendships and outgroup attitudes among ethnic minority youth. IJIR, 44 , 88-99.

    • Stark, Mäs & Flache, 2015. Liking and disliking minority-group classmates. SSR 50:164-176.

    • Mäs & Flache. 2013. Differentiation without Distancing. Explaining Bi-Polarization of Opinions without Negative Influence. PLoS ONE 8(11): e74516.

    • Munniksma, Stark, Verkuyten, Flache & Veenstra. 2013. Extended intergroup friendships within social settings. GPIB. 16(6) 752–770.

    • Stark, Flache & Veenstra 2013. Generalization of positive and negative attitudes towards individuals to outgroup attitudes. PSPB 39: 608-622.

    • Mäs, Flache, Takács & Jehn 2013. In the short term we divide, in the long term we unite: Organization Science24. 3: 716–736.

    • Stark & Flache, 2012. The Double Edge of Common Interest. SoE 85.2:179-199.

    • Flache & Macy, 2011. Local Convergence and Global Diversity. JCR 55.6: 968 - 993.

    • Flache & Macy 2011. Small Worlds and Cultural Polarization. JMS 35.1: 146-176.

    • Mäs, Flache & Helbing, 2010. Individualization as Driving Force of Clustering Phenomena in Humans. PLoSComputational Biology 6(10): e1000959.

    • Flache & Mäs 2008. How to get the timing right? CMOT 14.1:23-51.

    • Flache & Mäs 2008. Why do faultlines matter? SimPat 16.2: 175-191.30

    (Some) related published work:

    [email protected]://flache.gmw.rug.nl/