For those concerned with the quality of global intergroup relationships, the 21st century has not begun auspiciously. As we enter only its fifth year, we have already seen unprecedented incidents of international conflict and terrorism (e.g., Afghanistan, 2002; Iraq, 2003; Spain, 2004; USA, 2001), a rising tide of Islamophobia and anti-Semitism in Europe and elsewhere (e.g., Allen & Nielson, 2002), and increasingly hostile immigration receptions experienced by refugees (e.g., MORI, 2003). The task of ameliorating these problematic intergroup relationships will be arduous and complex, undoubtedly necessitating fundamental economic and political transformations, aided by the coordinated efforts of several social science disciplines. In this article, we consider the contribution that social psychology might make to such an endeavor, focusing particularly on the role of intergroup contact in reducing prejudice and in generating more favorable intergroup attitudes and stereotypes. We begin, appropriately enough in this the half centenary of its publication, with a brief reassessment of Allport's (1954) classic 'Contact Hypothesis' (see also Dovidio, Gaertner, & Kawakami, 2003). Although Allport was by no means the first to advocate bringing groups together as a means of prejudice reduction (Watson, 1947; Williams, 1947), his theorizing in The Nature of Prejudice has justly acquired landmark status because of its lucid exposition, penetrating insights, and careful analysis of the social conditions likely to facilitate the beneficial effects of contact on intergroup attitudes and behavior. As we shall show, there is now ample evidence to support his basic contentions. However, the past 50 years have also seen important theoretical and empirical developments of the Contact Hypothesis-in particular, Brewer and Miller's (1984) decategorization model, Gaertner and Dovidio's (2000) common in-group identity model, Hewstone and Brown's (1986) intergroup contact model, and Pettigrew's (1998) theory of longitudinal contact. In this article, we aim to provide a critical assessment of these developments, focusing mainly on our own model, which provided the point of departure for our own broad research program on contact over nearly 20 years. Although our original model was conceived partly as a riposte to Brewer and Miller's (1984) principal hypothesis, and it contrasts in key respects with Gaertner and Dovidio's (2000) subsequent alternative, the evidence that has accumulated, both from our own research and that of others, now suggests to us that some rapprochement of these apparently contradictory accounts is not only possible but also highly desirable. Later in the article, we discuss how this might be achieved. We will start with a brief re-presentation of Allport's (1954) classic hypothesis and show, with reference to recent cross-sectional and longitudinal surveys, laboratory experiments, and meta-analysis, that many of his original propositions have capably withstood the test of time. We then examine Brewer and Miller's (1984) and Gaertner and Dovidio's (2000) attempts to extend the Contact Hypothesis, in both of which categorization processes play a key role. This approach sets the stage for own model, first published in 1986 (Hewstone & Brown, 1986) and subsequently modified (Hewstone, 1996; Vivian, Hewstone, & Brown, 1997). In that model, we were concerned with identifying the conditions that would allow the generalization of attitudes and behavior change beyond the specific context in which the contact occurs. We hypothesized that group salience-broadly speaking, the extent to which group memberships are psychologically 'present' during contact-would play a key role in encouraging such generalization, essentially suggesting that it moderates the effects of contact on prejudice reduction. We are now in a position to evaluate that hypothesis in the light of the extensive evidence that has accumulated from a wide variety of contact settings. A major portion of the article is dedicated to a review of that evidence. A further goal of our theorizing has been to understand the processes that mediate the link between contact and improved intergroup attitudes. In this review, we highlight what seems to be one of the most important of those mediating factors, intergroup anxiety. However, as we shall show, other emotions-both positive and negative-also play an important role in determining the outcomes of contact. The empirical work we present is eclectic in its variety: It includes experimental work in laboratory settings, correlational and longitudinal research in a diversity of field contexts, and studies in more applied domains such as education and acculturation. We conclude with our suggestion as to how these different models might be reconciled, for which Pettigrew's (1998) similar integration provides an instructive comparison. The article is organized into nine sections. The first briefly reviews the research inspired by the Contact Hypothesis. In Section III, we introduce the three developments of Contact Theory that occurred in the 1980s, including our own model. Sections IV-IX review the empirical research instigated by the Hewstone-Brown model. In the final Section, X, we review progress to date and attempt a theoretical integration of these models in the light of the large volume of research that they have stimulated. For convenience of exposition, we have organized the material more or less historically. Our early work in this domain was focused on the processes of stereotype change in the face of disconfirming information, such as might be encountered in a contact situation. This is the subject of Section IV. We then move to experimental studies of contact that have attempted to manipulate intergroup salience directly to test the central hypothesis of the Hewstone-Brown model (Section V), that it is only when some group salience is maintained that one will find the generalized attitude change that is the goal of contact interventions. Necessarily, such laboratory studies are limited in the range of intergroup relationships they can investigate. In Section VI, therefore, we present the results from several correlational studies conducted in a wide range of different cultural contexts. The common theme of these studies was to investigate whether group salience, as experienced and reported by participants, moderates the effects of contact on various indicators of intergroup attitude. In the course of this work, it emerged that an occasional negative by-product of too much salience could be heightened intergroup anxiety, which is usually associated with unfavorable intergroup attitudes, as Stephan and Stephan (1985) had warned some years ago. This finding led to the realization that our original model had rather neglected the affective processes implicated in intergroup contact, an oversight that we addressed in a revised version (Vivian et al., 1997). This instigated a series of studies that investigated the mediating role of anxiety and other emotions in determining the outcome of contact and that also extended the range of outcome measures employed, including measures of affect and implicit attitude. These are presented in Section VII, together with some studies that have simultaneously tested moderation and mediation. In Section VIII, we present research that has attempted to apply the Hewstone-Brown model in educational settings. In Section IX, we discuss work on acculturation of ethnic groups in pluralist societies that links directly to models of intergroup contact. Section X presents our theoretical integration. © 2005 Elsevier Inc. All rights reserved.
