A More Complete Picture of Personality

Three Kinds of Profiles

Personality profiles reflect patterns of trait levels within persons. For example, Tim may be somewhat “lazy,” quite “self-assured,” and also quite “friendly.” On scales for each respective variable running from 0 to 9, Tim may therefore score 5 (somewhat lazy), 7 (quite self-assured), and 7 (quite friendly). The blue line in Figure 1 represents this “raw” personality profile.

OPEN IN VIEWER

Fig. 1.

Raw, normative, and distinctive personality profiles for a hypothetical subject, Tim. (See text for details.)

In analyzing personality profiles, it needs to be considered that sample means differ between traits. For example, most people—not only Tim—may be more friendly than lazy; that is, they show behavior that is associated with the term “friendly” more often than behavior that is associated with the term “lazy.” One may represent such differences between traits by averaging the profiles of many persons, resulting in a so-called normative personality profile (Furr, 2008) that reflects the trait levels of the average person. Let us assume that the normative profile for the three traits comprises the values of 3 for “lazy,” 6 for “self-assured,” and 8 for “friendly,” as illustrated by the green line in Figure 1.

Notably, most individual profiles resemble the normative profile at least somewhat, because the mean of each trait represents the central tendency of individual persons’ scores (Wood & Furr, 2016). Consequently, the raw personality profiles of any two persons will also tend to resemble each other to some degree. This resemblance may pose problems in terms of interpretation, as it will be unclear to what extent the similarity of the profiles is due to shared normativeness: Are they similar because both persons are average, or are they similar because both persons differ from the average person in the same ways? Moreover, normativeness is confounded with social desirability (see the “Findings Concerning Normative Profiles” section below).

A way of controlling for this source of profile similarity is to subtract the normative profile scores from the corresponding raw profile scores, resulting in so-called distinctive profiles (Furr, 2008). A distinctive profile describes the pattern of how a specific person differs from the average person. In Figure 1, Tim’s distinctive profile is represented by the red line, showing that he is lazier (+2) and more self-assured (+1) but less friendly (−1) than the average person. Thus, (a) the raw profile shows that Tim is more friendly and self-assured than lazy; (b) the normative profile shows that the differences between these traits are even stronger for the average person; (c) the distinctive profile shows that Tim is lazier and more self-assured but less friendly than the average person; and (d) the distinctive profile also shows that, in terms of deviations from the trait means, Tim’s laziness exceeds his self-assuredness, and both his laziness and his self-assuredness exceed his friendliness.

Currently, raw, normative, and distinctive profiles are the focus of numerous studies in the field of person perception. We will briefly recount a number of crucial findings from this line of research before addressing some unresolved issues and suggesting directions for future research.

Average Profile Agreement

Interpersonal perception research has investigated (among other issues) whether people see others as they see themselves (assumed similarity), whether self-reports agree with descriptions by others (self-other agreement), whether targets tend to be seen similarly by others (consensus), how aware people are of how others perceive them (meta-accuracy), and whether people’s judgments of one another are accurate (Funder & West, 1993; Kenny, 1994). Many of the relevant studies have been variable-centered, but profile research has also investigated these phenomena and found evidence in favor of all of them (e.g., Borkenau & Zaltauskas, 2009; Gallrein, Carlson, Holstein, & Leising, 2013).

Findings Concerning Normative Profiles

In interpersonal perception research, individual targets’ raw personality profiles have been compared with another raw profile, a normative profile, and a distinctive profile that comprise the same traits. Similarities among raw profiles tend to be higher but more difficult to interpret than similarities among distinctive profiles (Wood & Furr, 2016). Moreover, raw profiles tend to resemble normative profiles more than they resemble distinctive profiles (Biesanz, 2010; Biesanz, West, & Millevoi, 2007; Borkenau, Mosch, Tandler, & Wolf, 2016; Human & Biesanz, 2011b). This implies that most of the agreement between any two given raw profiles is likely to reflect differences between trait means. Further, normative profiles representing different perspectives (e.g., self-ratings vs. peer ratings) tend to be extremely similar: Borkenau and Zaltauskas (2009) found a correlation of .95 between a normative self-report profile and the corresponding normative peer-report profile, implying that averaged judgments of trait levels are almost independent of who provides the judgments.

Edwards (1953) noted that the rated social desirability of traits strongly predicts the respective trait means: People, on average, tend to attribute positive characteristics to themselves and to their acquaintances (Asendorpf & Ostendorf, 1998). This normativeness-desirability confound poses a problem for the interpretation of profile similarity indices, as two profiles may be similar just because they are both positive (Wood & Furr, 2016). Furthermore, the extent to which a raw profile is desirable/normative varies systematically with the informant’s attitude toward the person being described, including self-descriptions: Borkenau and Zaltauskas (2009) found that the self-report profiles of participants with higher self-esteem were more normative: The more fondly the participants thought of themselves, the more they described their personality like targets were described on average—that is, favorably. Likewise, Leising, Erbs, and Fritz (2010) showed that informants who liked another person more described that person’s personality more normatively (cf. Human & Biesanz, 2011b). Thus, the profiles of two target persons are likely to be more similar the more both targets are liked by the respective perceivers.

Notably, this association between profile normativeness and liking seems to be asymmetric: Personality profiles could, in principle, be stereotypically negative if perceivers who dislike their targets simply attributed everything that is negative, but nothing that is positive, to them. In a study by Leising, Ostrovski, and Zimmermann (2013), however, a different pattern emerged: Whereas profiles of targets whom the perceivers liked were very favorable and similar to each other, profiles of targets whom the perceivers disliked tended to be more evaluatively neutral and less similar. Perceivers seem to differentiate more between people they do not like than between people they like (Leising, Ostrovski, & Borkenau, 2012). These findings bear considerable resemblance to those from a current line of research in experimental psychology that highlights the greater perceived similarity of positively evaluated stimuli in general (Unkelbach, Fiedler, Bayer, Stegmüller, & Danner, 2008).

In most studies, the close connection between profile normativeness and profile desirability has been acknowledged but not directly addressed in the data analyses. Recently, however, Rogers and Biesanz (2015) included both profile characteristics simultaneously in the same analysis and showed that there is a reliable component in profile normativeness that is independent of social desirability: People seem to differ systematically in their knowledge of the distributions of traits in the population (irrespective of how those traits are evaluated) and in how aptly they use that knowledge to improve the accuracy of their own judgments.

Findings Concerning Distinctive Profiles

Several moderators of distinctive profile agreement have been identified. First, distinctive agreement between self-reported and other-reported personality profiles seems to increase with the level of acquaintance between target and informant (Biesanz et al., 2007). Second, such agreement is higher for traits that are more easily observed (e.g., higher for “talkative” than “ruminative”; Human & Biesanz, 2011a). These findings are in line with those of previous studies that used an item-wise approach to study interjudge agreement (e.g., Borkenau, Mauer, Riemann, Spinath, & Angleitner, 2004; Funder & Dobroth, 1987; John & Robins, 1993).

Third, informants who are expressly instructed to be accurate achieve higher distinctive self-other agreement: Biesanz and Human (2010) presented videos of a set of targets to a sample of informants that was subdivided into two groups: One group received no further instruction, whereas the other group was instructed “to form the most accurate impression possible.” The latter instruction resulted in lower profile normativeness and higher distinctive self-other profile agreement. This pattern of results could not have been established using an item-wise approach. Fourth, there is evidence that targets who report better psychological adjustment show higher distinctive self-other agreement (Human & Biesanz, 2011a; Tandler, Mosch, Wolf, & Borkenau, 2015). A possible explanation for this effect is that targets who are better psychologically adjusted provide others with more valid cues to their own more private experiences (Human, Biesanz, Finseth, Pierce, & Le, 2014).

Finally, several studies have addressed the question of what distinguishes distinctively more accurate from distinctively less accurate judges of personality. Profile analysis is the ideal method of approaching this question because it enables comparisons among individual perceivers in terms of the average accuracy of their judgments across a broad range of traits. Surprisingly, the search for attributes of good judges of personality has failed to yield replicable results: Whereas informants tend to be generally aware of targets’ distinctive personality features, they seem not to differ very much from each other in this respect (Biesanz, 2010; Tandler et al., 2015).

Relations Between Profile Normativeness and Distinctive Profile Agreement

Normative and distinctive profiles stemming from the same set of raw profiles are independent statistically. Nevertheless, profile normativeness and distinctive profile agreement may be related to other variables in similar or opposite ways. Opposite relations were found in the study by Biesanz and Human (2010) on the effects of instructions to be accurate in forming impressions of targets (described above) and in the study by Biesanz et al. (2007), who found that profile normativeness decreased whereas distinctive profile agreement increased with the level of acquaintance between target and perceiver. An example of concordant effects of profile normativeness and distinctive profile agreement was found in a study by Human, Sandstrom, Biesanz, and Dunn (2013), who reported that both higher profile normativeness and higher distinctive self-other agreement predicted more liking of, and more future interactions with, the target person.

Summary and Outlook

Raw, normative, and distinctive profiles are differentially associated with other variables, and ignoring these differences may lead to erroneous conclusions, such as mistaking effects of the shared positivity of different targets’ raw profiles for effects of their similarity in personality (Wood & Furr, 2016). Therefore, a general recommendation is to decompose raw profiles into their normative and distinctive components and then run analyses for all three types of profiles. Whereas raw profiles tend to show stronger relations to other variables than distinctive profiles do, the implications of normative and distinctive profiles are more straightforward.

A possible limitation of the profile approach may lie in the difficulty of obtaining objective criterion variables on many traits at once. Thus, if personality judgments are supposed to be validated against “hard” data (e.g., observational measures), the variable-centered approach will frequently be more convenient. The same is true when properties of specific traits are under scrutiny. So the choice between item-wise and profile analysis primarily depends on the respective research question. The profile approach seems more appropriate when aiming to generalize across traits and to incorporate predictors of judgment similarity at the level of individual perceivers, targets, or dyads.

We conclude this article with an overview of issues that should be addressed by future research. First, there are many systematic differences between personality descriptors besides their desirability. For example, personality-descriptive terms also differ in how much they refer to temporally stable (vs. unstable) characteristics or broad (vs. narrow) ranges of behaviors (Leising, Scharloth, Lohse, & Wood, 2014). Perceivers who know their targets well could be more inclined to attribute stable characteristics to them than perceivers who know their targets less well. Profile analysis is ideally suited for addressing such questions because generalizable conclusions in this regard may be obtained only when studying large and representative sets of traits.

Second, there is evidence that knowing targets better tends to improve distinctive profile agreement, whereas liking targets better is associated with greater profile normativeness. Liking and knowing, however, are positively correlated (Leising et al., 2010): We tend to like the persons we are well acquainted with. So do knowing and liking operate independently? Or—and under what conditions—do their effects counteract each other? And if they do, which effect is stronger? For example, extreme liking may be associated with loss of distinctive accuracy: When profiles become more positive/normative/similar, they may also become less characteristic of the individual target. Profile analysis permits addressing this question in an elegant fashion because it enables the simultaneous inclusion of several predictors of agreement (e.g., knowing and liking) at the level of individual perceiver-target dyads.

Third, one of the most puzzling unresolved issues at present seems to be the unclear origin of the strong association between profile normativeness and profile desirability: People tend to be judged as behaving in accordance with normative expectations as to how one should behave. Is this because people know those expectations and behave accordingly? Or do perceivers adjust their normative expectations to the ways people actually behave? Or are both true? As of now, the answer is unknown.

Finally, we do not yet fully understand the asymmetric nature of the association between normativeness and desirability very well: Why do favorable personality descriptions resemble each other more than unfavorable personality descriptions? Currently, models employing set points appear to bear the greatest promise in answering that question (e.g., Alves, Koch, & Unkelbach, 2016): Usually, there are few ways of being ideal but many possible ways of not being ideal. Consequently, the more different targets approach the ideal, the more similar they become. Future theory and research needs to consider such mechanisms as well.