The Meeker-Than-Average Effect

Does self-enhancement have a prayer?

Posted Apr 14, 2018

After the wars and the killings had ended, when usually there survived only some boys, some women, and children, these survivors were distributed among the Christians to be slaves. ~ Bartolomé de las Casas

God loves you! ~ Fra Pasquale di Bergamo

On the mount, the son of god declares that the meek shall inherit the earth (Matthew 5:5). From whom? And for how long? Remember the rapture and other eschatological puzzles (Krueger, 2018). Oracular as this statement is, it establishes the association between the Christian religion and the value of humility. Interestingly, a religion that was, according to Nietzsche (1886/1998), designed for slaves turned to aggressive proselytizing, using not only promises of riches (“thou shalt inherit”) but also threats of harm (the rack) or death (the stake). In time, the Christian system became the handmaiden of colonialism. A slave morality used for the enslavement of others. Nonetheless, the image, and especially the self-image, of humility runs deep, creating opportunities for self-enhancement researchers to look for better-than-average inferences.

Arguably, the case has already been made. The phenomenon of self-enhancement in the form of the better-than-average (BTA) effect is strong in the West, and particularly in the comparatively Christian environment of the United States of America. The most robust evidence emerges when people rate themselves and the average person on personality traits in the moral domain. Ratings of the self tend to be higher (i.e., better) than ratings of the other. Humility is a desirable trait in the Christian world, and it would be interesting to see direct evidence for the BTA effect on this very trait. To assert that you are humbler than others is to nullify this very claim at the same time. It’s a paradox. The claim of being humbler than average is credible only if it is denied.

Christians, like most of us, self-enhance. This is business as usual in the world of social perception, but it leaves an aftertaste of hypocrisy because Christians are not supposed to self-enhance. So why do another study? Gebauer, Sedikides, and Schrade (2017) did several because they saw an opportunity for strong inference. They reasoned that if they can demonstrate self-enhancement among Christians on those traits, beliefs, or practices that are central to the Christian self-image, then the evidence for the centrality principle is particularly strong. The centrality principle says that people self-enhance the most where it matters the most, i.e., on traits that are highly important, traits that are central to the self-image. Usually, centrality (vs. marginality) is treated as a moderator variable. Moderation is shown by sorting. We may compute self-enhancement indices separately for central and for marginal traits, and find that the effect is larger for the former. Whether this is persuasive is not clear. In most studies, rated importance is correlated with both rated descriptiveness of the self and rated desirability. Since those traits that are most important also tend to be the most desirable, it is to be expected that here the self-other differential is the greatest. Gebauer et al. set out to make a fresh start. In addition to the BTAE, they studied the so-called ‘overclaiming’ aspect of self-enhancement (a form of overconfidence in one’s own knowledgeability) and various facets of narcissism. I will limit my comments to the first of their BTAE studies.

The goal of this study was to test the hypothesis that Christians self-enhance on self-central issues. The self-central issues selected for study were adherence to commandments of faith (e.g., “Thou shalt keep the Sabbath holy.”). The evidence for this hypothesis, the authors felt, would be strong if it were also the case that there was less self-enhancement in a Non-Christian control condition on these same issues. Furthermore, the authors proposed that the evidence for the “self-centrality breeds self-enhancement” hypothesis would be stronger still if there were no differences in self-enhancement in a test involving issues not uniqulely central to the Christian self-image. The issues selected for this test were commandments of communion (e.g., “Thou shalt not covet . . .”), which tap a broader – not exclusively Christian – morality.

While the difference between the two sets of commandments is clear, the difference between conditions is not. The authors write that “there was only one, albeit crucial, difference between the Christian and the control conditions” (p. 791), namely whether the participants, when rating the average person, referred to Christians or to people in general. Yet, the demographics they report show that all participants in the Christian condition identified themselves as Christians, whereas only half did so in the control condition. The composition of the group was confounded with the instruction as to which average other to rate, although the authors submit that they "randomly assigned participants to one of the better-than-average conditions (between-subjects design): Christian or control" (p. 791). 

Gebauer et al. (2017)
Source: Gebauer et al. (2017)

Participants rated their own and the average other’s adherence to the two types of commandments on a percentage scale. The first two columns in the figure to the left show that in the Christian condition self-ratings for commandments of faith were higher than other-ratings. The next two columns show that there was no such difference in the control condition. The next set of four columns shows that for commandments of communion, there was a large self-enhancement effect in both conditions. One might stop here and note that the hypothesis that Christians self-enhance less than non-Christians do has been refuted. Had it been otherwise, Christians would have self-enhanced less than others on shared commandments (communion).

One might expect to see a test comparing self-enhancement on self-central issues with self-enhancement on non-central issues within the Christian sample. But there is no such test. It is not even clear whether commandments of faith are more central than commandments of communion to Christians. If it were so, Christian self-enhancement should be greater on commandments of faith than on commandments of communion. But it is smaller. Giving the benefit of the doubt to the centrality hypothesis, we assume that commandments of communion are highly central to Christians and others. What remains is the lack of a self-other difference on commandments of faith in the control condition. And why should there be a difference? To the atheists and the Buddhists in the sample, keeping the Sabbath holy is irrelevant. In short, there is evidence for self-enhancement where the basic condition of relevance is met, but not much more.

The authors then use a mediation analysis to argue that there is indeed more. The research is about the self-centrality principle after all. The authors test the hypothesis that for commandments of faith the independent variable (Christian vs. control) produces a difference in self-enhancement (present vs. absent) via the third variable of self-centrality. To obtain a measure of the latter, they asked all participants to rate how important it was to them to live up to the commandments. Statistically, the mediation analysis was successful. When ratings of self-centrality were controlled, there no longer was a difference in self-enhancement between the Christian and the control condition.

Let us remember what a mediation analysis does. A mediation analysis tests a model laying out a sequence of two causal paths (Fiedler, Meiser, & Schott, 2011). In an experiment, a treatment or intervention may affect a psychological state, which, when measured, can be shown to affect a person’s behavior. As we have seen, however, a participant in the control condition was there because he or she was not a Christian; Christians were found in both conditions. Although participants in the two conditions made ratings about a different kind of average person (Christian or generic), this “crucial difference between conditions” hardly qualifies as an experimental intervention. There is little justification for the claim that being in the Christian condition made participants regard commandments of faith as particularly self-central. Being placed in the Christian (vs. the control) condition and viewing commandments of faith as self-central (vs. not) are rather two sides of the same coin. Indeed, the authors report a correlation between the two of .85 (which they then, with 967 degrees of freedom [sic; it should be 965] test of statistical significance from zero [an empty statistical ritual, but that’s another story].

What have we learned? It is interesting to see that individual Christians believe themselves to be better keepers of the commandments than the average Christian is. It is not a meek mindset. The self-centrality issue remains open, however. Finding that atheists don’t self-enhance on Christian commandments is not very illuminating. One might look at the correlation between self-enhancement and self-centrality (importance) within the Christian sample. This correlation is likely positive.

What does all this have to do with Fra Pasquale you may ask. I met the Fra in Bergamo, his hometown (Krueger, 2012). He was radiant and joyful. He told me "God loves you!" The conviction in his voice was overwhelming. For 15 minutes my mind was afloat. I felt-enhanced. Then again, the Fra assured everyone. With the average being high, there was no room for comparative advantage.

Fiedler, K., Meiser, T., & Schott, M. (2011). What mediation analysis can (not) do. Journal of Experimental Social Psychology, 47, 1231-1236.

Gebauer, J. E., Sedikides, C., & Schrade, A. (2017). Christian self-enhancement. Journal of Personality and Social Psychology, 113, 786–809. 

Krueger, J. I. (2012). Fuori servizio. Psychology Today Online.

Krueger, J. I. (2018). Self-enhancement and the apocalypse. Psychology Today Online.

Nietzsche, F. (1998/1886). On the genealogy of morality. Indianapolis, IN: Hackett