No Evidence That Exposure to Materialistic Advertisements Influence Appearance Overvaluation and Financial Success Overvaluation in the Self-concept

All materials and data from the present research are publicly available via the Open Science Framework (OSF): https://osf.io/39nuj/. The hypothesis, power analysis, and data analytic plan were pre-registered on OSF in March 2021 after the pre-measures were collected, but before participants were contacted again to complete the experimental task and the post-measures: https://osf.io/pu53r.

In September 2020, 939 first and second year undergraduate students completed a 4-item version of the Beliefs About Appearance Scale (BAAS; Spangler & Stice, 2001) to measure appearance overvaluation (e.g., “How I feel about myself is largely based on my appearance”) and the 4-item Financially Focused Self-Concept Scale (FFS; Tabri et al., 2017) to measure financial success overvaluation (e.g. “How I feel about myself is largely based on the amount of money I have”). Participants responded to each item using a 5-point response scale with endpoints not at all (0) and extremely (4). The BAAS was selected to measure appearance overvaluation because it has been shown to have good psychometric properties and construct validity (Spangler & Stice, 2001) compared to the measure of appearance overvaluation used by Ashikali and Dittmar (2011; see Cash et al., 2004). Similarly, the FFS was selected to measure financial success overvaluation because it has good psychometric properties and construct validity, and was developed based on the BAAS (Tabri et al., 2017; Tabri, Philander, et al., 2021).

In March 2021, approximately six months later, all participants who completed the measures of appearance overvaluation and financial success overvaluation were contacted via email and invited to complete a study online about marketing and advertising strategies. Having a period of time between the pre- and post-measures helps to reduce the influence of demand characteristics on the results and the 6-month period was selected for practical reasons. Participants were told that we were interested in understanding their preferences for different consumer products and how their preferences are related to their attitudes, values, and opinions about consumer society. They were also told that the study would involve viewing different advertisements and completing a few questionnaires online, and that they would receive course credit for participating.

The study involved completing a consumer decision making task that was adapted from Ashikali and Dittmar’s (2011) Study 3. Participants were randomly assigned to a control condition wherein they viewed and compared several neutral advertisements or to an experimental condition wherein they viewed and compared materialistic advertisements. In the control condition, participants viewed six sets of two advertisements for the same type of product (e.g., two different advertisements for coffee) side-by-side. Likewise, in the experimental condition, participants viewed six sets of two advertisements for the same type of product (e.g., two different advertisements for champagne) side-by-side. In both conditions, the presentation order of the six sets of advertisements was the same for all participants. For each set, participants were asked to rate how attractive and how interesting each advertisement was to them in a given set and how much they desired the product. Then they were asked to select which of the two advertisements in a given set they preferred and to provide their reasoning. This task helped mask the true purpose of the experiment, which was to expose participants to either materialistic or non-materialistic advertisements.

After completing the consumer decision making task, participants completed the same measures of appearance overvaluation and financial success overvaluation as well as additional questionnaires to further mask the true purpose of the study. The additional questionnaires were a 4-item measure of body dissatisfaction (Tabri & Palmer, 2020; e.g., “When I think about my body compared to others, I feel dissatisfied”) and the 4-item Personal Relative Deprivation Scale (Callan et al., 2008; e.g., “When I think about what I have compared to others, I feel deprived”), and the Ten-Item Personality Inventory (TIPI; Gosling et al., 2003). Participants responded to each item on these measures using a 7-point Likert scale with endpoints Strongly disagree (1) and Strongly agree (7). They also completed the MacArthur Scale of Subjective Social Status (Singh-Manoux et al., 2003) with instructions to indicate where they believe they stood on the ladder relative to other Canadians. At the top of the ladder, participants believed they were Better off than most Canadians and at the bottom of the ladder participants believed that they were Worse off than most Canadians. Participants selected one of the rungs on the ladder with scores ranging from 0 to 10. All questionnaires completed after the consumer decision making task were presented in a randomized order. At the end of the study, participants responded to two questions about whether they provided high quality data. The first question related to accuracy and asked “Did you provide good, high quality responses?” with response options yes or no. The second question related to honesty and asked “Did you provide honest responses to all items?” with response options yes or no. All participants were then debriefed and compensated with course credit.

A total of 193 participants completed the consumer decision making task questionnaires. Thus, the response rate was 20.87%. However, the data of seven participants were excluded because they did not complete the measures of appearance and financial success overvaluation following the consumer decision making task. Also, the data of one participant was excluded because they indicated that they did not provide either accurate or honest data. Therefore, 185 participants (79.8% women, 19.1% men, 1.1% non-binary) were included in the analyses. Participants were 17 to 58 years of age (M = 20.27, SD = 4.91). All multi-item measures had good to high internal consistency (ɑs ranged from .71 to .89). Based on an a priori power analysis (see OSF), a minimum of 150 participants would be needed to detect a small effect (R² = .04) at an alpha of .05 with 80% power. Thus, the current research was adequately powered.

The present research was reviewed and cleared by our university’s Research Ethics Board.

Two linear regression analyses were conducted. In the first regression analysis, the dependent variable was a difference score indexing the change in appearance overvaluation from before to after the consumer decision making task. The independent variables were condition (exposure to neutral advertisement vs. materialistic advertisements; coded 0 and 1, respectively) and the pre-measure of appearance overvaluation. The same regression analysis was conducted for financial success overvaluation. Prior to conducting the analyses, we used the Breusch-Pagan and Koneker tests to test for the presence of heteroscedasticity. When there was no evidence of heteroscedasticity, normal-theory SEs were used. When there was evidence of heteroscedasticity, Huber-White robust SEs were used. Because there was no missing data, all analyses were conducted in SPSS version 26.

Descriptive statistics for appearance overvaluation and financial success overvaluation before and after the consumer decision making task are in Table 1. For appearance overvaluation, the Breusch-Pagan test was statistically significant (Lagrange multiplier [LM] = 20.91, p < .001), but the Koenker test was not statistically significant (LM = 1.737, p = .42). For financial success overvaluation, the Breusch-Pagan (LM =.71, p = .70) and Koenker (LM = .06, p = .97) tests were both not statistically significant. As such, Huber-White robust SEs were used in the regression analysis examining change in appearance overvaluation whereas normal-theory SEs were used in the regression analysis examining change in financial success overvaluation.

Results of the regression analyses are reported in Table 2. For change in appearance overvaluation, the main effect for the pre-exposure measure of appearance overvaluation on change in appearance overvaluation was statistically significant, b = -.39, SE = .06, t = -6.30, p < .001, and 95% CI [-.51, -.27]. That is, participants who reported more (relative to less) appearance overvaluation prior to viewing the advertisements reported a greater decline in appearance overvaluation from before to after viewing the advertisements. Unexpectedly, however, the main effect of exposure to materialistic advertisements compared to non-materialistic advertisements on change in appearance overvaluation was not statistically significant, b = .01, SE = .11, t = .08, p = .94, and 95% CI [-.21, .23]. The regression model explained 18% of the variance.

Likewise, for change in financial success overvaluation, the main effect of pre-exposure financial success overvaluation on change in financial success was statistically significant, b = -.44, SE = .06, t = -7.47, p < .001, and 95% CI [-.55, -.32]. That is, participants who reported more (relative to less) financial success overvaluation prior to viewing the advertisements reported a greater decline in financial success overvaluation from before to after viewing the advertisements. Unexpectedly, however, the main effect of exposure to materialistic advertisements compared to non-materialistic advertisements on change in financial success overvaluation was not statistically significant, b = -.03, SE = .10, t = -.30, p = .76, and 95% CI [-.22, .16]. The regression model explained 24% of the variance.

Because the results were unexpected, exploratory Bayesian regression analyses were conducted to examine support for the null hypothesis using JASP software (JASP Team, 2020). That is, we examined whether a null regression model fits better with the observed data compared to a regression model with only exposure to materialistic relative to non-materialistic advertisements using Bayes Factors (BF). The noninformative Jeffreys-Zellner-Siow (JZS) prior was used in the Bayesian exploratory analyses.¹ The noninformative prior was used because the posterior distribution would reflect relations in the data that are unaffected by information external to the data. Doing so provides a connection between the confirmatory (frequentist) results with the exploratory (Bayesian) analyses. The Bayesian analyses helped to determine whether there is evidence of no effect of exposure to materialistic relative to non-materialistic advertisements on change in overvaluation.

Model comparison statistics are reported in Table 3. After observing the data, the odds in favor of the model containing only exposure as a predictor of change in appearance overvaluation increased by a factor of $7.04e−8.$ This suggests that exposure contributes little towards explaining change in appearance overvaluation. The point estimate of the effect was $1.09e−3$ and the 95% Credible Interval included zero [-.07, .12]. In terms of relative predictive adequacy, the observed data were 5.88 times more likely under the null model compared to a model containing only exposure as a predictor of change in appearance overvaluation.

Likewise, after observing the data, the odds in favor of the model containing only exposure as a predictor of change in financial success overvaluation increased by a factor of $1.83e−10.$ This suggests that exposure contributes little towards explaining change in financial success overvaluation. The point estimate of the effect was $−3.63e−3$ and the 95% Credible Interval included zero [-12, .03]. In terms of relative predictive adequacy, the observed data were 5.88 times more likely under the null model compared to the model containing only exposure as a predictor of change in financial success overvaluation.

In sum, there is no evidence that exposure to materialistic relative to non-materialistic advertisements increases appearance overvaluation and financial success overvaluation.

Based on the Consumer Culture Impact Model, we tested the hypothesis that people who are exposed to materialistic advertisements (relative to non-materialistic advertisements) will report an increase in appearance and financial success overvaluation. The hypothesis, data analytic plan, and a priori power analysis used to determine sample size were pre-registered (see OSF). Unexpectedly, confirmatory (frequentist) analyses did not support a relation between change in appearance and financial success overvaluation as a function of exposure to materialistic relative to non-materialistic advertisements. Instead, there was a decrease in appearance and financial success overvaluation from before to after exposure to materialistic relative to non-materialistic advertisements. The decrease may be due, in part, to regression to the mean. Extending the results, exploratory (Bayesian) analyses indicated no evidence that exposure to materialistic relative to non-materialistic advertisements influenced change in appearance and financial success overvaluation. Taken together, the findings of the current research do not support the Consumer Culture Impact Model.

The findings of the current research challenge those of prior research in which people exposed to materialistic advertisements reported more appearance overvaluation compared to people exposed to non-materialistic advertisements (Ashikali & Dittmar, 2011; Study 3). Indeed, we observed no evidence that appearance overvaluation increases from before to after exposure to materialistic advertisements. Also, although not included in our pre-registered data analytic plan, we did not theoretically replicate Ashikali and Dittmar’s (2011) Study 3 results. More specifically, the post-exposure scores in Table 1 for appearance overvaluation show little or no difference between those exposed to materialistic advertisements and those exposed to non-materialistic advertisements. There are at least three reasons for the discrepant findings. First, Ashikali and Dittmar (2011) included women only in their sample whereas we included both women and men.² Second, appearance overvaluation in Ashikali and Dittmar’s (2011) study was assessed with a measure that has been shown to have poor construct validity (Cash et al., 2004). Third, it is unclear how Ashikali and Dittmar (2011) determined sample size. Their reported effect size was small-to-moderate and results from a post-hoc power analysis that we conducted indicated that the effect was 58% powered, which is below the conventional 80% power. The low power in Ashikali and Dittmar’s (2011) Study 3 suggests that their finding may be a false positive.

It is possible, however, that exposure to materialistic (relative to non-materialistic) advertisements may increase appearance and financial success overvaluation only among people who hold materialistic values. Materialistic values involve linking the acquisition of possessions in one’s life to happiness and success (Richins & Dawson, 1992). Critically, people who report higher (relative to lower) appearance overvaluation (Felix & Garza, 2012) or financial success overvaluation (Tabri et al., 2017) also report higher endorsement of materialistic values. Of note, Ashikali and Dittmar (2011; see Study 3) measured the extent to which participants hold materialistic values and found that it did not moderate the link between exposure to materialistic (relative to non-materialistic) advertisements and appearance overvaluation. The lack of a moderation effect may have been due to low statistical power. As well, only between-person differences in appearance overvaluation—not within-person differences in appearance overvaluation—were examined by Ashikali and Dittmar (2011). As such, future research should examine whether exposure to materialistic (relative to non-materialistic) advertisements increases overvaluation from before to after exposure for those who score higher (relative to lower) on materialistic values.

A possible limitation of the current research is that we did not include a manipulation check and so it is unclear whether the advertisements in the materialistic condition portrayed materialism to participants. We did not include a manipulation check for two reasons. First, including a manipulation check may increase demand characteristics. Second, manipulation check items may interfere with the experiment in unintended ways (see Hauser et al., 2018). That said, future research that attempts to replicate our results should consider examining the validity of the manipulation.

Other potential limitations have to do with the sample. In particular, the low response rate indicates that the results may be affected by self-selection bias. Indeed, participants who opted to complete the decision-making task six months later may share characteristics that make them different from those who opted not to complete the decision-making task. Although it is impossible to know for sure, it is plausible that participants who completed the task may be more conscientious than those who did not thereby making their responses to the pre- and post-measures of overvaluation more consistent. As well, the nature of the sample may be a limitation. Specifically, Ashikali and Dittmar (2011) recruited a community sample of women whereas we recruited a young adult university student sample that included both men and women. University student samples are not representative of the general population in that most are from Western, Educated, Industrialized, Rich, and Democratic (WEIRD) societies (Henrich et al., 2010). That said, examining factors that contribute to overvaluation of a domain in the self-concept among the young adult university student population is important because the developmental stage of adolescents and young adulthood is marked by self-concept exploration and formation (for a review see Klimstra & van Doeselaar, 2017). Moreover, the university student population is vulnerable to eating disorders (for a review, see Fitzsimmons-Craft et al., 2019), which are characterized by an overvaluation of appearance in the self-concept (Fairburn et al., 2003).

The purpose of the present research was to identify possible antecedents of appearance overvaluation and financial success overvaluation, two factors linked to the development and maintenance of disordered eating and disordered gambling, respectively. Based on the Consumer Culture Impact Model, we tested the idea that exposure to materialistic advertisements increases appearance and financial success overvaluation. Unexpectedly, there was no evidence that exposure to materialistic advertisements influences appearance and financial success overvaluation and so the results do not support the Consumer Culture Impact Model.

Contributed to conception and design: IRLB, NT

Contributed to acquisition of data: IRLB, NT

Contributed to analysis and interpretation of data: IRLB, NT

Drafted and/or revised the article: IRLB, NT

Approved the submitted version for publication: IRLB, NT

The research was supported by a Social Sciences and Humanities Research Council (SSHRC) of Canada Insight Development research grant to Dr. Nassim Tabri [430-2019-00941]. Isabella Bossom was supported by a SSHRC Canada Graduate Scholarship - Master’s (CGS-M). SSHRC did not have a role in determining the aims and outcomes of the current research or in the decision to submit the article for publication.

The authors declare no competing interests.

All materials and data from the current research are publicly available via the Open Science Framework: https://osf.io/39nuj/