Author:

Abstract:

For many decades, attitudes were typically measured directly by relying on self-report. However, introspective limitations or self-presentational strategies were found to sometimes threat the validity of these self-report measures (Greenwald et al., 2002). To avoid these self-presentational effects, new measures have been developed to assess attitudes indirectly, that is, without having to ask the individual for a verbal report (Fazio & Olson, 2003). The two most frequently used indirect attitude measures are the Implicit Association Test (IAT; Greenwald, McGhee, & Schwartz, 1998) and affective priming tasks (Fazio, Jackson, Dunton, & Williams, 1995).In this dissertation, we examined the ability of indirectly measured attitudes to predict behaviour. Predicting behaviour relying on indirect attitude measures has not been consistently successful. Especially for affective priming measures, predictive validity was often disappointing. Although in some studies consistency between the indirectly measured attitude and the behavioural measure(s) was observed, many studies also reported to find no substantial correlations between affective priming effects and behaviour (e.g., Bosson, Swann & Pennebaker, 2000). We aimed toexplain why previous studies sometimes failed to find substantial predictive validity for affective priming measures and to improve their predictive validity. Two main hypotheses were put forward. First, we hypothesized that affective priming effects (APE’s) obtained using a naming procedure (i.e. by having participants name the targets as quickly as possible) are more adequately suited to assess individuals’ attitude indirectly and subsequently to predict behaviour compared to APE’sresulting from the typically used evaluative decision procedure, which requires participants to categorise targets based on their valence. Second, the strength with which people hold an attitude in memory is often not taken into account when investigating attitude-behaviour consistency. However, differences in attitude strength between participants or conditions might explain why some studies report finding meaningful consistency between attitudes and behaviour and others do not.In this dissertation, we pursued two major directions. In a methodological approach, we relied on the findings of Spruyt and colleagues to develop an affective priming task using a naming procedure assessing people’s self-attitude indirectly. Primes were facial pictures or self-related words. We questionned whether this naming version of the affective priming task is able to predict behaviour. And if so, whether it can do so over and above direct attitude measures? We hypothesized that itsresulting APE’s would successfully predict behavioural consequences of the self-attitude. In line with the dual process models, we predicted that this indirect self-esteem measure would primarily guide spontaneous behaviour which is not easy to consciously control over and above direct self-esteem measures, which are assumed to predominantly predict deliberative behaviour. In a theoretical approach, we focused on attitude strength as a moderating variable influencing theattitude-behaviour relationship. We investigated whether holding an attitude more strongly in memory results in indirectly measured attitudes predicting behaviour more successfully. We hypothesized that more strongly held self-attitudes would result in stronger attitude-behaviour consistency. Apart from laboratory studies, we also aimed to put these hypotheses to the test in a clinical setting.Attracted by its clinical relevance, we chose to frame the research questions within research on self-esteem. That is, an individual’s global evaluation or liking of oneself in affective terms (Rosenberg, 1986). In the first part of the dissertation, we provide an integrative review and a meta-analysis of studies relying on an affective priming task to predict self-reported measures or observed behavioural measures. Subsequently, we report on the psychometric properties and the underlying factor structure of the Self-Liking/Self-Competence Scale Revised (Tafarodi & Swann, 2001), now validated inDutch.In a second part of the dissertation, we report on the findings of five experimental studies. When using facial pictures as primes, we found that higher levels of indirectly measured self-esteem as measured by the picture-picture naming task (PPNT) were related to higher levels of directly measured self-esteem if participants’ self-attitude was strong. In addition, the correspondence was mediated by the degree to which individuals perceived themselves as physically attractive, provided that participants value physical attractiveness as an important issue in their lives. These findings suggest that attitude strength indeed is relevant to observe consistency and that affective priming effects captureinterindividual differences in self-relevant constructs. In accordance with the dual process models, we also found that the APE predicted spontaneous gaze behaviour over and above a direct self-esteem measure. Similarly, the level of selfreported self-esteem at admission for eating disordered inpatients was unable to predict patient outcome at follow-up six months after discharge. In contrast, higher levels of indirectly measured self-esteem at admission were associated with poorer outcome at follow-up. These findings suggest that a larger discrepancy between directly and indirectly measuredself-esteem at admission is less favourable for future outcome. We also investigated the influence of attitude strength on attitude-behaviour consistency by examining the effect of attitude accessibility. Self-attitude accessibility was increased by having participants repeatedly evaluate self-related words, thereby strengthening the self-evaluation association. We hypothesized that increased self-accessibility would allow the APE’s to predict gaze behaviour even better. However, the attitude-behaviour correlations reversed. Evaluating attitude concepts prior to assessing individuals’ attitude indirectlytemporarily increased activation of the self-evaluation associations. This increased activation interferes with assessing the true strength of the self-evaluation associations in memory. Subsequently, the resulting indirect attitude effect does not reflect the individuals’ true attitude. Therefore the ability of the indirect attitude’s effect to predict behaviour can be distorted.