Anxiety explains why people with domain-contingent self-worth underperform on ability-diagnostic tests

Abstract

The present research examined whether test anxiety, evaluation apprehension, or lowered efficacy could explain the link between people’s domain-contingent self-worth and underperformance on ability-diagnostic tests. Results showed that, in an ability-diagnostic condition, the more students based their self-worth on academics the higher their anxiety, the lower their efficacy, and the lower their performance. In a non-diagnostic condition, the students’ academic contingencies of self-worth were unrelated to these outcomes. Moderated mediation analysis, with all three potential mediators, provided evidence that anxiety mediated the link between students’ academic contingencies of self-worth and underperformance within the ability-diagnostic condition. The discussion considered possible ways to reduce the threat experienced by people with domain-contingent self-worth in evaluative settings.

Introduction

It is difficult to imagine people succeeding in an area of their lives—whether in academics, sports, or social relationships—if they are indifferent to how well they do in that area. When people feel good after success and feel bad after failure in a domain then they will surely try hard to succeed and avoid failure (Osborne & Jones, 2011). Is it possible, however, for people to care so much about their domain successes and failures that they feel overwhelming pressure and their performance suffers? Such underperformance may be likely when people tie their self-worth to success and failure in a domain (Burhans and Dweck, 1995, Covington, 1984, Crocker et al., 2006, Deci and Ryan, 1995), especially when they perform difficult tasks in evaluative settings (Lawrence & Crocker, 2009). The present research examines why people with domain-contingent self-worth are vulnerable to underperforming in these settings.

As the label implies, when people have domain-contingent self-worth, their self-worth rises after successes and drops after failures and setbacks (Crocker, Sommers, & Luhtanen, 2002). Their self-worth is especially likely to drop when they perform below their expectations (Crocker, Karpinski, Quinn, & Chase, 2003), or after they perform poorly when they either have low self-esteem (Park, Crocker, & Kiefer, 2007) or believe intelligence is fixed (Niiya, Crocker, & Bartmess, 2004).

Given that self-worth boosts feel good and self-worth drops feel bad, it is reasonable to assume that domain-contingent self-worth increases people’s motivation and therefore their performance (Major and Schmader, 1998, Osborne, 1995, Osborne, 1997, Osborne and Jones, 2011, Steele, 1997). Indeed, the more people base their self-worth on a given domain the more they devote time and effort to that domain (Crocker, Luhtanen, Cooper, & Bouvrette, 2003; also see Crocker et al., 2006). Despite this motivation, when faced with difficulty or failure, people with domain-contingent self-worth risk impaired motivation and performance (Burhans and Dweck, 1995, Covington, 1984, Crocker et al., 2006, Deci and Ryan, 1995). The reason, according to several theorists, is that people with domain-contingent self-worth seek to validate their abilities and traits (Burhans and Dweck, 1995, Crocker et al., 2006). These self-validation goals, in turn, can impair people’s motivation and performance on difficult tasks due to processes such as anxiety, shame, effort withdrawal, and low ability attributions (Elliott and Dweck, 1988, Grant and Dweck, 2003).

The negative outcomes of domain-contingent self-worth appear to be most likely in difficult evaluative settings. Lawrence and Crocker (2009) found, for example, that the more students based their self-worth on academics the worse they performed on an ability-diagnostic test. On a non-diagnostic test, they found no link between how much students based their self-worth on academics and their performance. These findings suggest that domain-contingent self-worth is linked to performance-impairing psychological and motivational processes within ability-diagnostic but not non-diagnostic settings. Yet it is unknown which performance-impairing process could explain this underperformance.

Three performance-impairing processes seem particularly promising as mediators: test anxiety, evaluation apprehension, and low efficacy (Geen, 1985, Hembree, 1988, Multon et al., 1991, Robbins et al., 2004, Sarason, 1984, Sarason and Stoops, 1978, Seipp, 1991). As previously noted, people with domain-contingent self-worth are prone to feeling anxiety, tension, and pressure (Crocker et al., 2006, Deci and Ryan, 1995, Dykman, 1998). Given that people with domain-contingent self-worth focus on self-validation goals (Burhans and Dweck, 1995, Crocker and Knight, 2005, Crocker and Park, 2004, Lawrence et al., 2005), they may also be predisposed to worry that performing poorly in ability-diagnostic settings will signal to others that they have low ability (i.e., high evaluation apprehension). Finally, people with domain-contingent self-worth may have less belief in their ability to do well (i.e., low efficacy) within ability-diagnostic settings than within non-diagnostic settings. Indeed, the more people tend to protect their self-worth the lower their confidence in their ability (Thompson & Dinnel, 2003).

The present study examined the link between students’ academic contingencies of self-worth and test anxiety, evaluation apprehension, low efficacy, and performance within both ability-diagnostic and non-diagnostic conditions. Undergraduate students completed an academic contingencies of self-worth measure and then learned that they would take a math test—framed as either ability-diagnostic or non-diagnostic. After viewing a sample of difficult math problems, the students completed measures of test anxiety, evaluation apprehension, and efficacy, and then took a difficult math test.

As in our previous research (Lawrence and Charbonneau, 2009, Lawrence and Crocker, 2009), we only analyzed the white students’ data in order to reduce the likelihood that stereotype threat or stereotype lift, based on minority-group membership, would affect performance (Steele and Aronson, 1995, Walton and Cohen, 2003). Although it is possible that gender-based stereotype threat and stereotype lift could affect math performance in this study (Spencer, Steele, & Quinn, 1999), our research consistently finds that among white students, gender does not interact with academic contingencies of self-worth nor with Test Instructions (i.e., ability-diagnostic versus non-diagnostic) to impact math performance (Lawrence and Charbonneau, 2009, Lawrence and Crocker, 2009; also see Gonzales, Blanton, & Williams, 2002).¹