The identification of the physical attractiveness stereotype – initially encapsulated as “what is beautiful is good” (Dion et al. Reference Dion, Berscheid and Walster1972) – and a seminal qualitative review of the then emerging physical attractiveness literature (Berscheid & Walster Reference Berscheid, Walster and Berkowilz1974) engendered a tsunami of multidisciplinary research on physical appearance over the past four decades. Unsurprisingly, meta-analysis has been commonly used to review findings from the thousands of studies of physical attractiveness. Meta-analyses have examined the attractiveness stereotype (Eagly et al. Reference Eagly, Ashmore, Makhijani and Longo1991; Feingold Reference Feingold1992a), correlations between attractiveness and other characteristics (Feingold Reference Feingold1992a; Jackson et al. Reference Jackson, Hunter and Hodge1995; Langlois et al. Reference Langlois, Kalakanis, Rubenstein, Larson, Hallam and Smoot2000), and the effects of attractiveness on decision making in mate selection (Feingold Reference Feingold1988; Reference Feingold1990), employment (Hosoda et al. Reference Hosoda, Stone-Romero and Coats2003), and the courtroom (Mazzella & Feingold Reference Mazzella and Feingold1994).
In these meta-analyses, many of which were cited in the target article, investigators began by specifying methods for study retrieval (e.g., databases searched) and study inclusion criteria. Although these steps should be taken for narrative (qualitative) and for meta-analytic reviews, they were rarely used for the former until their importance was highlighted by pioneering meta-analysts critical of traditional reviews. In addition, meta-analyses focus on effect sizes (including moderators of effect sizes) and their confidence intervals for findings, rather than on null hypothesis significance tests. The former approach is now widely recognized as preferable and is supplanting the latter, even in empirical research (Cumming Reference Cumming2013; Feingold Reference Feingold2015).
The target article is an old-school qualitative review in the mold of Berscheid and Walster (Reference Berscheid, Walster and Berkowilz1974). The authors do not provide the criteria used to locate or include studies for their review, although one evident criterion for inclusion was that a study had to have found a positive effect of physical attractiveness on social treatment. However, when a meta-analysis found an overall effect of attractiveness to be nonsignificant, there were often empirical studies included in that review that had found significant effects. Hence, with a narrative review approach that allows for cherry picking of studies based on statistical significance (and capitalizing on type I errors) to be cited, physical attractiveness effects may appear to be more pervasive and robust than they actually are. Consequently, researchers who conduct qualitative reviews of the attractiveness literature may draw tendentious conclusions that are inconsistent with findings from past or future meta-analyses conducted with a comprehensive and objectively selected set of studies and addressing issues with publication bias, outliers, and among-study variations in effects.
The discussion in the target article (and its abstract) adheres to the classical view of the physical attractiveness stereotype, that is, that all desirable attributes are more likely to be ascribed to the attractive than to the unattractive, and ignores moderation of attractiveness effects by type of attribute. The review also concluded that the stereotype is strongest for perceptions of women and of the opposite sex. But are these broad conclusions supported by meta-analyses of the attractiveness stereotyping literature?
Two independent meta-analyses that have synthesized findings from experiments on the physical attractiveness stereotype (Eagly et al. Reference Eagly, Ashmore, Makhijani and Longo1991; Feingold Reference Feingold1992a) suggest otherwise. Feingold (Reference Feingold1992a) found observers were most likely to attribute social skills and extraversion (sociability and dominance) to physically attractive targets. The attractive were seen as only moderately more mentally healthy and intelligent. Most important, given the frequent use of the term “good” to describe perceptions of the attractive, this meta-analysis found that attributions of character (e.g., honesty and trustworthiness) were unaffected by target attractiveness, and the attractive were viewed as less modest (e.g., vainer). The moderation of effect sizes by target gender was also examined, and no significant differences on attributions of social competence, intelligence, and personality traits were found. Most important, Target Gender × Target Attractiveness × Participant Gender interactions were all nonsignificant.
Eagly et al. (Reference Eagly, Ashmore, Makhijani and Longo1991) also examined the moderation of effects of attractiveness by attribute. The abstract of their meta-analysis noted, “The differences in subjects' perception of attractive and unattractive targets were largest for indexes of social competence; intermediate for potency, adjustment, and intellectual competence; and near zero for integrity and concern for others” (p. 109), which is consistent with the findings from Feingold (Reference Feingold1992a). The Eagly et al. meta-analysis also found no significant differences in effect sizes between male and female targets for the different attributes examined, which was also in accord with Feingold (Reference Feingold1992a). However, the findings from the meta-analyses are inconsistent with the conclusion of the target article that the physical attractiveness stereotype is more potent in perceptions of women than of men, and for opposite-sex rather than same-sex evaluations.
The target article also cited Mazzella and Feingold's (Reference Mazzella and Feingold1994) meta-analysis on the effects of defendants' physical attractiveness on sentencing recommendations by mock jurors as affording more evidence for the ubiquitous benefits of physical attractiveness, although averring that the attractiveness advantage was greater for perpetrators of some crimes than for others. Although the meta-analysis did indeed find a positive effect of target attractiveness on punishment recommendations, the mean effect size (d) of 0.12 was trivial, and the Defendant Attractiveness × Crime interaction was misinterpreted as ordinal. Although physical attractiveness was beneficial to defendants guilty of theft, rape, or cheating, no effect of attractiveness was found for swindling, and attractive defendants were treated more punitively than unattractive defendants for negligent homicide. (The strong crossover interaction explained the weak main effect of attractiveness captured by the mean effect size from all studies.) Hence, the right interpretation of the meta-analytic results is that physical attractiveness may or may not affect sentencing recommendations for defendants depending on the nature of crime, and its effects may be positive or negative.
The main contribution of the target article is in providing a conceptual model that affords a foundation for a subsequent meta-analysis of findings of attractiveness effects discussed that were not synthesized in earlier quantitative reviews, much as Berscheid and Walster (Reference Berscheid, Walster and Berkowilz1974) paved the way for past meta-analyses of attractiveness effects from older work. Such a research synthesis of new domains should address moderators of attractiveness effects besides gender, and their associated effect sizes and confidence intervals.
The identification of the physical attractiveness stereotype – initially encapsulated as “what is beautiful is good” (Dion et al. Reference Dion, Berscheid and Walster1972) – and a seminal qualitative review of the then emerging physical attractiveness literature (Berscheid & Walster Reference Berscheid, Walster and Berkowilz1974) engendered a tsunami of multidisciplinary research on physical appearance over the past four decades. Unsurprisingly, meta-analysis has been commonly used to review findings from the thousands of studies of physical attractiveness. Meta-analyses have examined the attractiveness stereotype (Eagly et al. Reference Eagly, Ashmore, Makhijani and Longo1991; Feingold Reference Feingold1992a), correlations between attractiveness and other characteristics (Feingold Reference Feingold1992a; Jackson et al. Reference Jackson, Hunter and Hodge1995; Langlois et al. Reference Langlois, Kalakanis, Rubenstein, Larson, Hallam and Smoot2000), and the effects of attractiveness on decision making in mate selection (Feingold Reference Feingold1988; Reference Feingold1990), employment (Hosoda et al. Reference Hosoda, Stone-Romero and Coats2003), and the courtroom (Mazzella & Feingold Reference Mazzella and Feingold1994).
In these meta-analyses, many of which were cited in the target article, investigators began by specifying methods for study retrieval (e.g., databases searched) and study inclusion criteria. Although these steps should be taken for narrative (qualitative) and for meta-analytic reviews, they were rarely used for the former until their importance was highlighted by pioneering meta-analysts critical of traditional reviews. In addition, meta-analyses focus on effect sizes (including moderators of effect sizes) and their confidence intervals for findings, rather than on null hypothesis significance tests. The former approach is now widely recognized as preferable and is supplanting the latter, even in empirical research (Cumming Reference Cumming2013; Feingold Reference Feingold2015).
The target article is an old-school qualitative review in the mold of Berscheid and Walster (Reference Berscheid, Walster and Berkowilz1974). The authors do not provide the criteria used to locate or include studies for their review, although one evident criterion for inclusion was that a study had to have found a positive effect of physical attractiveness on social treatment. However, when a meta-analysis found an overall effect of attractiveness to be nonsignificant, there were often empirical studies included in that review that had found significant effects. Hence, with a narrative review approach that allows for cherry picking of studies based on statistical significance (and capitalizing on type I errors) to be cited, physical attractiveness effects may appear to be more pervasive and robust than they actually are. Consequently, researchers who conduct qualitative reviews of the attractiveness literature may draw tendentious conclusions that are inconsistent with findings from past or future meta-analyses conducted with a comprehensive and objectively selected set of studies and addressing issues with publication bias, outliers, and among-study variations in effects.
The discussion in the target article (and its abstract) adheres to the classical view of the physical attractiveness stereotype, that is, that all desirable attributes are more likely to be ascribed to the attractive than to the unattractive, and ignores moderation of attractiveness effects by type of attribute. The review also concluded that the stereotype is strongest for perceptions of women and of the opposite sex. But are these broad conclusions supported by meta-analyses of the attractiveness stereotyping literature?
Two independent meta-analyses that have synthesized findings from experiments on the physical attractiveness stereotype (Eagly et al. Reference Eagly, Ashmore, Makhijani and Longo1991; Feingold Reference Feingold1992a) suggest otherwise. Feingold (Reference Feingold1992a) found observers were most likely to attribute social skills and extraversion (sociability and dominance) to physically attractive targets. The attractive were seen as only moderately more mentally healthy and intelligent. Most important, given the frequent use of the term “good” to describe perceptions of the attractive, this meta-analysis found that attributions of character (e.g., honesty and trustworthiness) were unaffected by target attractiveness, and the attractive were viewed as less modest (e.g., vainer). The moderation of effect sizes by target gender was also examined, and no significant differences on attributions of social competence, intelligence, and personality traits were found. Most important, Target Gender × Target Attractiveness × Participant Gender interactions were all nonsignificant.
Eagly et al. (Reference Eagly, Ashmore, Makhijani and Longo1991) also examined the moderation of effects of attractiveness by attribute. The abstract of their meta-analysis noted, “The differences in subjects' perception of attractive and unattractive targets were largest for indexes of social competence; intermediate for potency, adjustment, and intellectual competence; and near zero for integrity and concern for others” (p. 109), which is consistent with the findings from Feingold (Reference Feingold1992a). The Eagly et al. meta-analysis also found no significant differences in effect sizes between male and female targets for the different attributes examined, which was also in accord with Feingold (Reference Feingold1992a). However, the findings from the meta-analyses are inconsistent with the conclusion of the target article that the physical attractiveness stereotype is more potent in perceptions of women than of men, and for opposite-sex rather than same-sex evaluations.
The target article also cited Mazzella and Feingold's (Reference Mazzella and Feingold1994) meta-analysis on the effects of defendants' physical attractiveness on sentencing recommendations by mock jurors as affording more evidence for the ubiquitous benefits of physical attractiveness, although averring that the attractiveness advantage was greater for perpetrators of some crimes than for others. Although the meta-analysis did indeed find a positive effect of target attractiveness on punishment recommendations, the mean effect size (d) of 0.12 was trivial, and the Defendant Attractiveness × Crime interaction was misinterpreted as ordinal. Although physical attractiveness was beneficial to defendants guilty of theft, rape, or cheating, no effect of attractiveness was found for swindling, and attractive defendants were treated more punitively than unattractive defendants for negligent homicide. (The strong crossover interaction explained the weak main effect of attractiveness captured by the mean effect size from all studies.) Hence, the right interpretation of the meta-analytic results is that physical attractiveness may or may not affect sentencing recommendations for defendants depending on the nature of crime, and its effects may be positive or negative.
The main contribution of the target article is in providing a conceptual model that affords a foundation for a subsequent meta-analysis of findings of attractiveness effects discussed that were not synthesized in earlier quantitative reviews, much as Berscheid and Walster (Reference Berscheid, Walster and Berkowilz1974) paved the way for past meta-analyses of attractiveness effects from older work. Such a research synthesis of new domains should address moderators of attractiveness effects besides gender, and their associated effect sizes and confidence intervals.