• Keine Ergebnisse gefunden

Heterogeneity in behavior Heterogeneity in discriminationHeterogeneity in discrimination

evidence from Colombia

5.6 Data and empirical strategy

5.7.4 Heterogeneity in behavior Heterogeneity in discriminationHeterogeneity in discrimination

People may perceive ex-combatants differently and act differently towards demobi-lized individuals depending on their conflict experiences and personal preferences.

In order to investigate heterogeneity in participant behavior, we focus on three di-mensions that are likely correlated with perceptions towards the demobilized: degree of victimization, socio-economic background as measured by private vs. public uni-versity attendance, and political orientation of the participant. Table 5.5 reports the degree of discrimination by each subgroup on the four relevant measures.

Chapter 5. Discrimination and inter-group contact in Colombia

TABLE5.5:Heterogeneouseffectsinbehaviorandattitudes BehaviorandattitudesNoContactImpactofContact DonationDonationamountLikingindexIAT-testDonationDonationamountLikingindexIAT-test PanelA:Victimization(1)(2)(3)(4)(5)(6)(7)(8) NoVictimized0.0720.197-0.419***0.069***0.0110.2720.167**-0.028 (0.047)(0.236)(0.022)(0.013)(0.050)(0.287)(0.057)(0.032) Victimized0.0510.592-0.449***0.0150.1581.0440.1330.115 (0.114)(0.815)(0.045)(0.044)(0.097)(0.816)(0.094)(0.109) StronglyVictimized-0.205-0.579-0.324**0.028-0.032-0.749-0.2610.040 (0.193)(1.593)(0.097)(0.074)(0.149)(1.229)(0.245)(0.143) PanelB:TypeofUniversity Private0.1270.444-0.416***0.060**-0.031-0.1440.1390.026 (0.059)(0.341)(0.037)(0.021)(0.056)(0.311)(0.083)(0.049) Public-0.030-0.024-0.424***0.0610.0870.7120.136**-0.033 (0.058)(0.322)(0.029)(0.030)(0.066)(0.467)(0.043)(0.044) PanelC:PoliticalOrientation Left0.049-0.313-0.392***0.082**0.0251.0860.0600.035 (0.076)(0.458)(0.038)(0.027)(0.063)(0.503)(0.090)(0.054) Center0.0520.472-0.420***0.050**0.018-0.0630.151**-0.023 (0.059)(0.315)(0.025)(0.018)(0.051)(0.326)(0.045)(0.046) Right0.2551.062-0.665**-0.0020.189-0.1710.6630.010 (0.180)(0.712)(0.212)(0.096)(0.203)(0.728)(0.427)(0.174) N10581058105810571058105810581057 Note:Thetablereportsthetotaldifferentialeffectbypersonalcharacteristicsindiscrimination.TheregressionincludecontrolsasshownintableC.2.All modelsincludesessionandstudysubjectfixedeffects.Clusteredstandarderrorsatthesessionlevelareinparenthesis,***,**,*denotesignificanceat1,5and10% includingthemultiplehypothesistestingcorrectionbySidak.

The first four columns report the differences for each of the outcomes of interest when there is no contact, while columns 5 to 8 report the impact of contact on discrimination.

Panel A reports the different effects by victimization status. Panel B reports by uni-versity type and Panel C depicts the heterogeneous effects by the political orientation of the participants. All results are adjusted for multiple hypotheses testing. We focus on total discrimination, only and do not distinguish between the SES treatment, both for simplicity and to ensure we have enough participants in each subgroup. Overall, there are only marginal differences concerning SES treatments.

Referring to the results in Panel A, we find no significant differences in charitable behavior towards demobilized beneficiaries compared to non-demobilized beneficia-ries depending on a participant’s experiences of violence. Neither the likelihood to donate (column 1) nor the amount donated (column 2) are significantly different for demobilized recipients compared to non-demobilized recipients for any subgroup.

Surprisingly, the results show that experiences of victimization do not affect attitudes towards demobilized persons. All three groups prefer poor persons to the demobi-lized, and the extent of preference does not vary by subgroup. Participants with no exposure to violence have a positive and significant IAT-test, indicating some prefer-ence for demobilized individuals. This result does not hold for participants exposed to low or intense violence. However, the implicit discrimination is not statistically different across subgroups.

Neither private university students nor public university students displayed dis-criminatory behavior towards demobilized trainees in the crowdfunding campaign.

Regarding attitudes, private and public university students report to like poor indi-viduals more than demobilized persons to a similar extent. Similarly, they achieve equal positive implicit discrimination scores.

Concerning political orientation, the results in Panel C show no discrimination in charitable behavior by any subgroup. Nevertheless, right-oriented persons report disliking demobilized individuals relative to poor people to a more substantial degree.

This finding indicates that attitudes and behaviors do not necessarily need to align with each other. Moreover, right-oriented participants have significantly lower IAT-test scores compared to the left and center-oriented participants.

We conclude that the results only partly support Hypothesis 1. We find that experi-ences of violence are not correlated with a higher degree of discrimination towards demobilized individuals. Private university students donate as much to demobilized trainees as to other SENA trainees and express the same degree of dislike as public university students. Right-wing participants express more negative attitudes toward demobilized.

Chapter 5. Discrimination and inter-group contact in Colombia

Heterogeneous treatment effects of contact

In order to check whether sub-groups react differently to the contact treatment, we estimate Equation 5.3. Persons with negative experiences of violence or right-wing political attitudes may watch the video, focusing more on the negative aspects of the reintegration process. The indirect contact with demobilized might, therefore, strengthen their existing opinion (confirmatory bias). If this is the case, our results will show an adverse effect for prejudiced groups. Results in columns 5 and 6 in Panel A of Table 5.5 show that indirect contact with demobilized individuals does not increase nor decrease discriminatory behavior. The same is true for implicit discrimination for participants with no experience or low or high experience of victimization. Nevertheless, the results show a positive change in attitudes towards demobilized for non-victimized participants with contact.

In Panel B, we investigate the heterogeneity between public and private university students, which can be seen as an indicator of different socio-economic backgrounds, since students from wealthier households are more likely to attend a private univer-sity than students from poorer households. The indirect contact does not significantly change discriminatory behavior toward demobilized persons for any of the subgroups.

We find that after watching the video, the attitudes of public university students towards the demobilized relative to poor persons are more favorable. IAT-test scores do not change with the contact.

For political preferences, we see a similar pattern. The contact does not change the difference in the likelihood to donate, the amount donated to demobilized compared to non-demobilized trainees, nor the IAT-test scores of left, center, or right-oriented participants. Center-oriented participants show more positive attitudes towards demobilized persons after they have watched the video.

We conclude that indirect contact with demobilized persons increases donations to the same extent as contact with a non-demobilized person. Non-victimized participants, public university students, and center-oriented persons show more positive attitudes towards demobilized persons relative to poor persons after the contact treatment.

Hence, indirect contact has a positive effect on attitudes towards the demobilized among specific population groups. We find no adverse effects of contact on dis-crimination towards the demobilized. Hence, our results partly confirm Hypothesis 3.