The attitudes that we believe a good person should or shouldn’t hold are tremendously diverse & reliably connected to a sense of self; also, we are very good at making peace with our own "bad" attitudes

A Good Person Shouldn’t Feel This Way: Moralized Attitudes, Identity, and Self-Esteem. Pierce Ekstrom, Calvin Lai. Collabra: Psychology (2022) 8 (1): 36344. Jun 16 2022. https://doi.org/10.1525/collabra.36344

Abstract: Moralized attitudes are the attitudes that people construe as matters of right and wrong. In this study, we examine how moralized attitudes relate to how people evaluate themselves using the Attitudes, Identities, and Individual Differences (AIID) dataset—a survey of over 200,000 individuals asked to report their attitudes in one of 95 domains. In pre-registered analyses that were based on exploratory analyses of a subset of the data, we found that the specific attitudes that people moralize differ greatly from individual to individual and that moralized attitudes are more central to one’s identity than non-moralized attitudes. We also examined whether mental conflict between identity-central attitudes and gut feelings about the corresponding attitude objects would be related to lower self-esteem, finding mixed and weak evidence supporting that claim. Together, our findings indicate that the attitudes that people moralize are tremendously diverse and are reliably connected to a sense of self. At the same time, peoples’ self-esteem may be resilient to specific instances in which their gut feelings fall short of the attitudes that are central to their identity.

Keywords:Morality, Attitudes, Self-Esteem, The Self

Discussion

Summary of Results

Our analyses yielded three key results. First, our descriptive analyses suggest that the attitudes that people perceive to be matters of right and wrong are extraordinarily diverse. Some participants saw room for debate concerning attitudes and behavior that most people would consider to be unequivocally moral. For example, a handful of participants “agreed” or “strongly agreed” that “Because of my personal values, I believe that making negative judgments about giving is acceptable” (6.3% of those who saw the question). At the same time, some participants passed stark judgment on some attitudes and behavior that most people would perceive as matters of taste. For example, some participants “strongly agreed” that “Because of my personal values, I believe that making negative judgments about Harry Potter is unacceptable” (8.5% of participants who saw the question). The descriptive statistics presented in Figure 4 illustrate this diversity. The end result is that there are probably very few (if any) attitudes or behaviors that all people would agree to be moral, immoral, or morally neutral. This diversity is in its own right an interesting characteristic of moral psychology, but it also suggests that researchers interested in how people think and feel about morally charged stimuli cannot always safely assume that their participants will construe a given stimulus in moral (or non-moral) terms.

Second, our results yield robust support for the Morality-Identity hypothesis. We found that when participants perceived their attitudes to be connected to their personal values, they were more likely to identify with those attitudes. They were also more likely to identify with the targets of those attitudes implicitly and explicitly (e.g., to see Harry Potter or Christians as part of their self-concept rather than to merely identify as someone who likes Harry Potter or Christians). This evidence is consistent with our prediction that people would perceive their idiosyncratic beliefs about right and wrong as a defining feature of who they are.

Finally, our results yield weak and inconsistent support for the Identity Rubric hypothesis. We find that our participants did not consistently evaluate themselves on the basis of whether their gut feelings were consistent with the “actual” attitudes that they cherish as defining features of their identity. Our first analysis found some evidence suggestive of this phenomenon; when participants reported negative gut feelings about targets with which they identified, they also reported slightly lower self-esteem. However, this may be because negative gut feelings about identity-central targets are, in and of themselves, negative gut feelings about the self. For example, negative gut feelings about Christians, Jews, European Americans, and African Americans could easily translate to negative self-evaluations among people who are themselves Christian, Jewish, European American, or African American—regardless of whether they think it is morally “wrong” or “acceptable” to judge people from these groups positively. Other models were more directly inconsistent with our predictions. When participants reported negative gut feelings about targets that they thought they ought to like, their reported self-esteem was almost identical to that of participants whose gut feelings and explicit attitudes were perfectly consistent. We observed a similar null effect for implicit identification with targets and for every other indicator of attitude strength that we analyzed. Participants generally did not have lower self-esteem when their gut feelings were inconsistent with moralized, important, certain, or extreme attitudes.

Based on these findings, our initial theory requires significant revision. We hypothesized that moralized attitudes inform self-evaluation because they structure individuals’ self-concepts. Although we did find that moralized attitudes were relatively central to participants’ identities (consistent with the Morality-Identity hypothesis), we found little if any connection between participants’ self-esteem and the extent to which their gut feelings were consistent with those attitudes. Participants did not report feeling meaningfully worse about themselves when they were attracted to things they believed they shouldn’t like or repulsed by things they believed they ought to accept.

Limitations

That said, our study has several limitations that complicate this test of our theory.

Mixed evidence for key measures’ validity. The evidence from our validity study was mixed and weaker than we had anticipated. The most widely used measure of attitude moralization (moral conviction) only sometimes predicted responses to the AIID items. For one item, the relation was what we predicted for both positive and negative attitudes, though those relations were weak. For the other three items, the relation was only present either for positive or negative attitudes. We encountered similar problems with the AIID measures of attitude identity centrality. Despite this limitation, we remain confident in our conclusions for three reasons.

First, despite evidence that some measures were more valid than others and more or less valid for positive versus negative attitudes, we find no evidence that our results depended on which AIID items we used to assess attitude moralization or attitude identity centrality. Second, although the measures we used to assess the validity of the AIID items are certainly more widely used and better established as measures of the target constructs, they are still just measures, not perfect reflections of the constructs of interest. Given that the AIID measures each make explicit reference to their respective target constructs (e.g., “personal values,” decisions about what is “wrong” or “acceptable,” whether reactions are “important to” and “inconsistent with” the “self-concept”), these measures may capture aspects of moralization and identity centrality that other measures do not. Finally, we found converging evidence for our conclusions with analyses that does not rely on the AIID measures. To do this, we used the validation study to conduct an additional un-pre-registered test of the Morality-Identity hypothesis. Our theory would predict that Skitka and colleagues’ moral conviction measure would be related to Luhtanen and Crocker’s (1992) measure of identity centrality for the 20 attitude targets in this study. It was (b = 0.32, 95% CI: [0.27, 0.37], p < 0.001), even controlling for the effect of importance. In addition, the analyses described in Table 9 suggest that alternative measures of moralization or identification would probably not yield any stronger support for the Identity Rubric hypothesis. Participants’ self-esteem was basically unmoved regardless of how severely their gut feelings contradicted certain, important, extreme, “identity-central,” or “moralized” attitudes. Even if we have failed to find a direct, precise measure of moralization or identity centrality, surely at least one of these indices of attitude strength would at least be correlated with such a measure. If the Identity Rubric hypothesis were true, then, it seems unlikely that all of these tests would be so uniformly null.

Still, our validity study offers an important caveat for our own and future work. We cannot safely assume that when people say that their attitude is moral or important to who they are, they will also say that the opposite attitude is immoral or anathema to their self-concept. Although we might say people “moralize” an attitude when they judge it to be desirable, acceptable, or wrong, these judgments probably do not lie along a single dimension.

Narrow threats to the self and a broad measure of self-esteem. In hindsight, our predictions may have presumed self-esteem to be too fragile. The Rosenberg Self-Esteem scale is intended to measure self-esteem as a global trait. Meanwhile, we analyzed participants’ gut feelings and attitudes toward only one or two targets. Contradictions so narrow and specific may be insufficient to impact global trait-level self-esteem in a meaningful way.

On the one hand, narrower measures of self-esteem might prove to be more malleable. People’s thoughts and behavior during a particular event or time period might affect how they feel about themselves during that specific time. On the other hand, more frequent, numerous, or chronically salient contradictions between people’s gut feelings and the attitudes they believe to be appropriate might have a stronger impact on self-esteem than the one or two attitudes we were able to assess.

Some scholars have conducted experiments to confront participants with moral failures and trace their impact on self-evaluations (see Wojciszke, 2005). These participants often end these experiments feeling just fine. These results are consistent with decades of social-psychological research have documented individuals’ capacity to rationalize their behaviors and what they might consider to be failures to practice what they preach, and scholars have often argued that the point of this rationalization is to protect people’s positive self-images (e.g., Aronson, 1969; Kunda, 1990).

Small wonder, then, that slight divergences between “gut feelings” and “actual” attitudes in a single attitude domain failed to leave a dent in our participants’ self-esteem. Our study leaves open the possibility that some attitude-inconsistent feelings or behavior may be more uncomfortable than others, and that measures of more specific or shorter-term self-evaluation might be more likely to change in the wake of these behaviors. Future work might find that when people behave in ways that are clearly at odds with multiple important, moralized, or identity-central attitudes, they briefly feel worse about themselves.

Correlational Design. We have tested a multi-step causal framework with a correlational dataset, which cannot permit strong causal inferences. For example, we cannot know whether people come to identify with certain attitudes because they see them in moral terms, moralize attitudes because they are central to their identity, or come to moralize and identify with their attitudes simultaneously as a part of some larger process. The reality is probably a combination of these possibilities. For example, someone may come to identify with their pro-choice attitude because they see abortion access as a moral issue and also come to moralize their attitude toward Britney Spears, 50 Cent, or Harry Potter because they identify as a fan. Regardless, our evidence suggests that these processes are connected. At the same time, the latter part of the model proposed in Figure 1 is now less plausible, as our correlational evidence was inconsistent with the Identity Rubric hypothesis.