Research suggests that the method of assessment affects endorsement rates of disordered eating and related attitudes. Studies examining the apparent self-report bias that occurs in eating disorder assessments have consistently reported higher rates of eating pathology found via questionnaires as compared to interviews (Fairburn & Beglin, 1994; Keel, Crow, Davis, & Mitchell, 2002). One theory for why this discrepancy exists implicates social desirability and emotions such as shame, suggesting that participants alter their responses in particular assessment formats so that others will perceive them more favorably (Richman, Kiesler, Weisb, & Drasgow, 1999; Vitousek, Daly, & Heiser, 1991). Respondents may be motivated to deny or minimize specific attitudes or behaviors, particularly when the topic is of a highly sensitive nature (Catania, Gibson, Chitwood, & Coates, 1990; Rayburn, Earleywine, & Davison, 2003). Researchers have therefore proposed that assessment formats associated with greater anonymity may yield more candid responses (Anderson, Simmons, Milnes, & Earleywine, 2007; French et al., 1998 and Keel et al., 2002).
Given research suggesting that eating disorder assessments are influenced by response format and anonymity, assessments of behaviors associated with body dissatisfaction may be subject to a similar bias. In studies comparing responses on the questionnaire and interview versions of the Eating Disorder Examination (EDE; Fairburn & Beglin, 1994; Fairburn & Cooper, 1993), significant differences in body shape concerns have emerged, providing preliminary evidence for the influence of response format. However, despite the role that body dissatisfaction often plays in the etiology of disordered eating (Cafri et al., 2005 and Stice, 1999), body image concerns and related behaviors occur independently of eating disorders and have a higher frequency in the general population (Sarwer, Thompson, & Cash, 2005). Additionally, although body image has traditionally been viewed as a “women's issue,” a growing body of research suggests that a substantial proportion of men experience body dissatisfaction (Olivardia, 2001). While these concerns may be considered normative among women (Rodin, Silberstein, & Striegel-Moore, 1985) and increasingly normative among men, many of the associated behaviors are still potentially sensitive. Studies also suggest that body dissatisfaction is associated with depression and lower self-esteem (McCreary & Sasse, 2000; Stice, Hayward, Cameron, Killen, & Taylor, 2000), as well as potentially harmful behaviors including extreme dieting, excessive exercise (Ricciardelli & McCabe, 2004; Stice, 2001), and anabolic steroid use (Blouin & Goldfield, 1995).
The purpose of the present study was to replicate the existing literature examining effects of response format and anonymity on endorsements of eating disordered behaviors and to extend the literature by examining these effects on endorsements of behaviors associated with body dissatisfaction. Endorsement patterns on a conventional self-report measure were compared to endorsement patterns on an unmatched count questionnaire. Conventional self-report measures of nominal anonymity require that participants provide no identifying information (e.g., name) on their questionnaires. However, in many studies, participants are required to return their questionnaire to a research assistant who could conceivably look at the questionnaire and link their responses to their identify. In the unmatched count technique (UCT), individual items are grouped into sets composed of non-sensitive items and a critical item of interest. Participants indicate only how many items in a set are true and thus never endorse any item directly, guaranteeing complete response anonymity. This assessment procedure, which operates under the assumption that participants will be more likely to respond honestly when provided true response anonymity, is used to establish base rates of potentially sensitive behaviors (Anderson et al., 2007; Dalton, Wimbush, & Daily, 1994). In the present study, it was hypothesized that the conventional and unmatched count assessment methods would yield significantly different endorsement rates of eating disordered and body dissatisfaction-related behaviors, consistent with participants providing higher endorsement rates on the unmatched count questionnaire.
