People Are Like Me

1977 was Lee Ross’s annus mirabilis. He published three now-classic papers. In “The ‘false consensus effect,’” with Greene and House, Ross breathed new life into the well-known phenomenon of social projection. In “Social roles, social control,” with Amabile and Steinmetz, he reinvented the correspondence bias as the fundamental attribution error. In “The intuitive scientist” (i.e., you), he presented a general account of the limits of human rationality. Reflecting on his unique career, Ross (2018) sees the social mind rather as he did back then. The intuitive scientist still falls short. Yet, much has happened in the last half century, and our understanding of human (ir)rationality has become more nuanced. Understandably, though, Ross’s own provocative demonstrations of human error loom large in his retrospective. Yet, these demonstrations come with their own limitations.

Let’s consider one of these limitations with the example of ‘false consensus.’ In the iconic demonstration (Ross et al., 1977, studies 3 & 4), students were asked whether they might help out with a study on communication by walking around campus wearing a sandwich board that read either “Eat at Joe’s” or the more ominous “Repent!” Students were then asked to estimate “What percentage of your peers do you estimate would agree to carry the sandwich board around campus? What percentage would refuse to carry it? (Total should be 100%)” (p. 290).

In study 3, the wearers estimated on average that 61.4 percent of their peers would wear the sign and that 38.6 percent would not. The non-wearers estimated on average that 30.4 percent of their peers would wear the sign and that 60.4 percent would not (Table 4, p. 292). A statistical comparison of the ‘wear-sign’ estimate reveals that the wearers themselves provided a higher estimate (60) than the non-wearers (30). This is the false consensus effect. A comparison of the ‘not-wear-sign’ estimate is redundant; it yields the same result because the estimates of non-wearing are 100 minus the estimates of wearing.

The basic effect is robust. It had been known since the 1940s (Wallen, 1943), and it has been replicated since 1977. Ross (2018) refers to a meta-analysis summarizing the evidence of the following 10 years (Marks & Miller, 1987), but has nothing to say about alternative theories and analyses of social projection (see Krueger, 1998, for a review). There is one interesting complication in the study of perceived consensus that appears to have escaped attention. Here’s a brief sketch of what needs to be done.

Notice that the conventional assessment and analysis is very limited in that each respondent provides one, and only one, judgment. The wearers (the ‘yays’ hereafter) estimate the percentage of yays, thereby implying their estimates of nays as 100 – yays. The non-wearers (the ‘nays’ hereafter) do the same. Yet, the observed result can arise from four different underlying processes, or some combination thereof. First, there is something about the way the yays make judgments. Second, there is something about the way the nays make judgments. Third, there is something about the way yays are judged. Fourth, there is something about the way the nays are judged. From the data available in the typical consensus study, it is impossible to tell where the action is. Many social psychologists will not care. They will be content to note that the effect is significant and conclude that ‘people’ are flawed intuitive scientists. With this self-imposed restraint, we will not know if the bias lies with some people and not others, or if the bias attaches to some issues and not others.

Interestingly, other, seemingly related, social-perceptual phenomena have attracted more in-depth analysis. Consider self-enhancement. Self-enhancement is, on the face of it, the antithesis of social projection (or ‘false consensus’). When people self-enhance, they set the self positively apart, that is, they accentuate a positive difference. In contrast, when they project, they assimilate others to the self, that is, they attenuate the difference. In research on self-enhancement, various efforts have been made to separate perceiver and target effects. Self-enhancement is robust if it is the case that a person’s self-judgment is more favorable than this person’s judgment of others (perceiver effect) and if this person’s self-judgment is more positive than the mean judgment of this person made by others (target effect) (see Krueger & Wright, 2011, for a review).

The same kind of effect separation can be attained in studies of consensus. With a suitable design consisting of multiple groups of judges and targets, it will be possible to ask if it is the case that certain people project by believing that their own group is larger or their own type is more common than are other groups or types (perceiver effect), and to ask if it is the case that certain people project by believing that their own group or type is more common than it is believed to be by others who are not in this group or of this type. This sort of design and analysis will be more complex than the traditional two-group approach, but it promises to take us beyond blanket inferences about ‘people and their shortcomings.’