• Keine Ergebnisse gefunden

GROUP SIMILARITY AS A PREDICTOR FOR GENERALIZATION AND

According to Glaser et al. (2015), a possible moderator of LAC is the strength of the association between focal and lateral objects. The stronger the association, the higher the probability of a lateral change in attitude. One way to operationalize the strength of association is the similarity of objects. Glaser et al. (2015) assumed a linear relationship between similarity and generalization effects: The more similar objects are to one another, the stronger they should be associated with one another and the more likely it is that generalization effects will occur. In the case of

displacement effects, however, Glaser et al. (2015) postulated a bell-shaped relationship instead of a linear one. Although the evaluation of the focal object is assumed to remain unchanged, the evaluation of the lateral objects is expected to change as follows: If similarity is low, the strength of the association should be correspondingly low, so that no generalization occurs. With moderate similarity, however, the similarity could go unnoticed and an automatic generalization could occur, which would then also be measurable in explicit measures. If similarity becomes too high and obvious, a propositional defense against the generalization could occur, so that no lateral changes of attitude can be found either. In summary,

126

the probability of displacement effects increases with increasing similarity, but if the similarity is too great, they decrease again (Glaser et al., 2015).

The explanations of the bell-shaped relationship between similarity and displacement effects can partly be combined with the set-reset model from Wegener and Petty (1995). According to the theory, assimilation occurs when the emotional reaction to the context (in the LAC model: lateral objects) and the reaction to the target (in the LAC model: the focal object) overlap. If this overlap is not consciously perceived, this leads, according to the model, to the objects’ being perceived as more similar than they actually are (setting). This should be the case for moderately similar objects.

Overestimated similarity can explain why generalization occurs. However, when the overlap is processed propositionally, this leads to a resetting, and focal and lateral objects are perceived as more dissimilar than they actually are. Consequently, one of the two reactions has to be corrected (Wegener & Petty, 1995). As postulated with the displacement effect, there would then be no generalization effects. This

correction of the reaction, however, could not only lead to a lack of generalization, but even to an opposite reaction if, for example, overcorrections occur (Martin, 1986;

Martin et al., 1990; Strack et al., 1993). In the study by Strack et al. (1993),

participants were asked to evaluate an ambiguous target after being primed with the respective traits in another (ostensibly “unrelated”) study. Half of the participants were reminded of the priming procedure, and that reminder resulted in an

overcorrection, a contrast effect. The other half, those who were not reminded of the priming procedure, did not correct for its possible influence and evaluated the target in line with the priming valence, showing an assimilation effect (Strack et al., 1993).

The influence of similarity as a moderator of the LAC theory can be summarized as follows: If two objects are similar, generalization effects should occur. If two attitude objects are too similar, this should lead to corrections that can lead to contrast effects. If they are dissimilar, there are no associations between them, so no

generalization effects should occur. However, in the case that they are perceived as being opposed, this leads to opposing evaluations and thus to contrasting effects.

The study presented next was designed to replicate the results of Study 3, to investigate possible explanations for the found contrast effects (i.e. which

associations are activated for lateral groups) and to analyze group similarity effects in a more detailed matter.

127

Another moderator of LAC postulated by Glaser et al. (2015) is PFC, the preference for consistency. PFC refers to the extent a person seeks consistency among his or her cognitions. Although early consistency theories already noted that all people strive for consistent attitudes, beliefs, and cognitions (e.g. Festinger, 1957; Heider, 1958), the extent varies across individuals.

According to dissonance theory (Festinger, 1957), two opposing cognitions result in an unpleasant state and therefore motivate change in the cognitive system to re-establish cognitive consistency. Glaser et al. (2015) noted that these principles can also be applied to attitude generalization: When the evaluation of the focal attitude object changes, it might become inconsistent with evaluations of lateral objects, and people should be motivated to re-establish consistency, which could be achieved by changing the lateral evaluations as well. PFC was therefore introduced as a possible moderator of LAC.

Evidence for the moderating effect of PFC on LAC were found in a study by Heitland and Bohner (2010). They asked German participants who held at least a moderate level of prejudice against Turks to generate and record Pro-arguments for integrated housing of Germans and Turks (a cognitive dissonance paradigm). Results showed that especially those participants who perceived high choice to argue for the topic (high dissonance) and who scored high on PFC showed a positive evaluation of integrated housing of Germans and Turks, and this effect generalized to the evaluation of Turks in general.

THE CURRENT STUDY

As a focal group Turks were chosen, and two pretests were run to identify four lateral groups likely to be perceived as very similar, somewhat similar, rather dissimilar and very dissimilar to Turkey and to choose items for a PFC scale. A negative stereotype activation regarding Turks was expected to influence not only the subsequent

evaluation of the focal group, but also the four lateral groups, which differ in similarity.

The positive stereotype activation was expected to have rather small positive or no effects on the evaluation of the lateral groups, but the negative stereotype activation (negative condition) was expected to lead to generalization and contrast effects, depending on group similarity.

128

1. For a group with high perceived similarity (e.g. Albanians), an explicit contrast effect was expected in the negative condition, as participants should recognize a persuasion attempt through the manipulation and correct their evaluation accordingly. After a negative stereotype activation regarding Turks, the highly similar group should be evaluated more positively than after a positive

stereotype activation. However, implicitly, the negative evaluation of Turks should generalize to the highly similar group.

2. For a moderately similar group (e.g. Kosovars), both an explicit and implicit generalization effect was expected. Participants in the negative stereotype activation condition should evaluate Kosovars less positively than in the positive stereotype activation condition.

3. A moderately dissimilar group (e.g. Slovenians) was not expected to be

affected by activation of stereotypes regarding Turks, as this group should not have any connection to Turks and has low perceived similarity.

4. A very dissimilar, antithetic group (e.g. Swedes) was expected to be explicitly and implicitly evaluated in contrast to Turks in the negative condition. After a negative stereotype activation regarding Turks, Swedes are predicted to be evaluated even more positively than in the positive condition.

5. Generally, explicit mean group evaluations in the negative condition were expected to be distributed in a U-shaped way (quadratic trend) with more positive evaluations for the highly similar and the dissimilar group, but implicit evaluations should be distributed in a linear way, with more positive ratings for less similar groups.

6. PFC was expected to moderate the evaluations such that participants high in PFC were expected to show stronger generalization effects than participants low in PFC.

7. MCPR was expected to moderate the evaluations in a way that participants low in MCPR were expected to show stronger generalization effects than participants high in MCPR.

129