TRAITS, SELF-REPORT, AND I/O

TRAITS, SELF-REPORT, AND I/O

That our field focuses almost exclusively on a trait approach combined with self-report measurement seems to us self-evident. Nevertheless, we offer two bits of evidence. First, in reviewing the edited volume Personality Psychology in the Workplace (Hogan & Roberts, 2001), we note that the only mention made of non-trait concepts (e.g., moral character, ego) is made in chapters written by non-I/O psychologists. Second, in reviewing the first three issues of the Journal of Applied Psychology for 2003, we counted 16 articles in which the personality (as opposed to attitudinal variables such as organizational commitment) of real (as opposed to paper) people was assessed. Of these, all but one used self-report measures (the other used a projective measure), and only four involved motives, goals, beliefs, etc., as opposed to traits such as conscientiousness and extraversion.

Let us then consider what is missing. By focusing exclusively on traits, we

P ERSONALITY IN I NDUSTRIAL /O RGANIZATIONAL P SYCHOLOGY 121 fail to consider (a) motives, schemas, goals, beliefs, and other less behavioral

aspects of personality (Fiske, 1994; Pervin, 1994), (b) situational factors that might influence the translation of human hardwiring into behavior (Mischel & Shoda, 1995), and (c) the explanations for the observed patterns of behavior (John & Robins, 1994; Snyder, 1993). At best, traits tell us what people tend to do (at worst, they tell us what people want us to believe that they tend to do). Traits tell us little about the motives, the cognitions, and the biochemistry that lead to behavior (Ozer & Reise, 1994). Neither do they tell us anything about the aspects of situations that might channel the influence of motives, cognitions, and biochemistry into behavior.

By using ‘objective’ self-report, we restrict ourselves to self-concept at best. The problems with self-report measures of personality are well known and needn’t be repeated here. It suffices to say that we have not been satisfied with self-report measures of other variables of interest (e.g., cognitive ability, job performance), and that we have responded to calls (e.g., Dunnette, 1963; Guion, 1961) for more construct-oriented approaches to the measurement of these other variables. So it seems odd that we would accept self-report of personality as adequate, knowing what we do about the ability and willingness of people to fake as well as the degree to which people like to think well of themselves.

And what is the price that we have paid? It is impossible to know for certain. For now, we suggest that one index of the costs associated with our narrow approach to personality is the degree to which our criterion- related validities are not commensurate with what we believe to be the case regarding the overlap between personality and performance domains. Next, we discuss the possible advantages associated with broadening our concep- tual and empirical perspectives.

Barrick and Mount (1991), without doubt, deserve much of the credit for re-awakening our interest in personality as a predictor of workplace out- comes. The paper was one of the most widely cited of the decade, and with good reason. However, it is interesting to note that the predictor criterion correlations presented in Barrick and Mount (1991) are more or less the same correlations that led Guion and Gottier (1965) and others to suggest a moratorium on research exploring the predictive value of personality. Increasing acceptance of statistical corrections for artifacts has led to larger and larger corrected values (e.g., Mount & Barrick, 1995), but the uncor- rected values remain disappointing. Moreover, if we go beyond bivariate correlations and consider incremental validity (e.g., Cortina, Goldstein, Payne, Davison, & Gilliland, 2000), we see that typical measures of person- ality contribute very little to the prediction of performance. As Mischel and Shoda (1994) aptly stated, ‘The basicdata of our field remain reassuringly stable: only the interpretations seem to shift’ (p. 158).

If one agrees with the notion that employers place at least as much im- portance on personality as they do on other characteristics, then something

122 I NTERNATIONAL R EVIEW OF I NDUSTRIAL AND O RGANIZATIONAL P SYCHOLOGY 2005 must be wrong. As Hogan and Roberts (2001) put it, ‘Individuals whose

income depends on the performance of others understand that personality matters’ (p. 5). No matter what the job, people make many choices every day, and these choices influence a person’s level of output. On the other hand, it is rare to use all of one’s abilities. In other words, one’s level of ability rarely sets one’s performance apart from the performance of others if for no other reason than because one rarely makes use of all of one’s ability. Dispositions, on the other hand, create distinctions among people all day, every day. And yet, most of the data suggest otherwise. We contend that the reasons for these disappointing results lie in our nearly exclusive focus on traits and our reliance on self-report measurement.

This is not to say that work on traits has been fruitless. The development and acceptance of the Big Five has been critical to the development of the field of personality psychology. As Eysenck put it, ‘ . . . the taxonomy precedes causal analysis; we must analyze and classify the entities in our field of study before we can frame meaningful theories concerning their behavior’ (1991, p. 779). In the words of Fiske, the Big Five gives us some ‘fairly solid ground on which to stand as we explore the wetlands of personality’ (1994, p. 123). Allport (1968), Digman (1990), Goldberg (1993), John and Robins (1994), and many others have echoed these sentiments.

Prior to the acceptance of the Big Five, trait psychology was a morass of narrow descriptors, many of which were clearly similar to one another. The field grew very slowly because the study that examined the predictive value of, say, dependability was not necessarily seen as building upon previous work on, say, conscientiousness. The Big Five gave us a common lexical thread with which we could tie our work together. It also gave us a structure from which we could explore causes.

But we didn’t. Fortunately, others have. In the section that follows, alternatives to the

trait-based approach and their relevance to understanding behavior in the workplace are discussed. This is followed by a discussion of operational issues and viable alternatives to self-report measurement.