NTERNATIONAL R EVIEW OF I NDUSTRIAL AND O RGANIZATIONAL P SYCHOLOGY 2005 ALTERNATIVES TO ASSUMPTIONS OF STABILITY

130 I NTERNATIONAL R EVIEW OF I NDUSTRIAL AND O RGANIZATIONAL P SYCHOLOGY 2005 ALTERNATIVES TO ASSUMPTIONS OF STABILITY

One of the assumptions that underlie much of personnel selection is that the attributes that we use to select, promote, and otherwise classify people are stable over time and across situations. This assumption may work for cog- nitive ability, especially in adults, but it does not appear to work so well for personality in general or for traits in particular.

This is not to say that there is no intra-individual, cross-situational stab- ility in personality. However, the search for such stability, from Hartshorne and May (1928) to Mischel (1968) to Mischel and Peake (1982), has often proved disappointing. That said, there are two forms of instability on which we would like to focus: intra-individual instability across situations and inter- individual instability within trait levels. Both have important implications for our field.

Intraindividual Instabilityacross Situations

Research by Shoda et al. (1994) shows that individual children behave in an extraverted manner in some situations, and in an introverted manner in others. Other children respond to the same situations in the opposite fashion. Likewise, some children are aggressive in response to provocation from one kind of source (e.g., adults), but passive in response to provocation from another kind of sources (e.g., peers), whereas other children exhibit the opposite pattern. Thus, there is a great deal of intra-individual inconsistency. When situational factors are taken into account, however, a pattern of within- situation consistency often emerges. Those children who respond aggres- sively to provocation from a peer at Time 1 are very likely to respond aggressively to provocation from a peer at Time 2. Those same children might respond passively to adults at Time 1 and Time 2. This is not to say that one cannot average across situations. The problem that can arise from using an averaging strategy is that the child who responds aggressively to adults and passively to peers is then equated with the child who responds passively to adults and aggressively to peers.

Thus, a system in which a single trait value is attached to a given person by averaging across situations misses the person by situation interaction. An I/O psychologist might complain that the Mischel research was done with chil- dren and is therefore irrelevant to I/O, but similar results have been found with adults (Mischel & Peake, 1982). In any case, it should be noted that this work begins with a theory explaining why cross-situational inconsistency and intra-situational consistency should exist. Situations have ‘psychological features’ to which we respond differently. Some of us are perfectly at ease one on one but are terrified at the prospect of speaking to large groups. Others are (or become) comfortable in large, impersonal settings but seem awkward and misplaced in more intimate settings. These differences may be

P ERSONALITY IN I NDUSTRIAL /O RGANIZATIONAL P SYCHOLOGY 131 generated by biological or by early environmental differences, but, regardless

of their source, they must be important for understanding behavior in the workplace.

But the story doesn’t end there. Just as people may gravitate towards jobs that are commensurate with their level of cognitive ability (Wilk, Demarais, & Sackett, 1995), so may they gravitate towards employment situations with psychological features that allow them to behave in a manner that is both comfortable to them and consistent with their values or goals (Roberts & Robins, 2000). Thus, we would expect to find less variability in situations and their features for people at later career stages. For those who are at earlier career stages, however, or for those for whom job change is constrained, we might expect a vast array of situations, a vast array of features, and a vast array of trait levels even within an individual.

None of this should come as a surprise. Other areas of psychology have considered, for example, the evolutionary bases of personality. This research suggests that particular levels of personality variables have adaptive value and that the identification of the levels of key personality variables of others has adaptive value (Bodenhausen, Macrae, & Hugenberg, 2003; Buss, 1991; Fiske, 1993; Fiske & Taylor, 1991). Is there any reason to believe that the exhibition of a particular behavior, without regard for the features of the situation, is adaptive? Being organized may be adaptive on the whole, but reorganizing the file system while the office is on fire is, perhaps, misguided. Being assertive may have its advantages, but anyone who asserts himself while being arrested for armed robbery only succeeds in adding a hospital sentence to his jail sentence.

It should also be noted that this sort of intra-individual instability is different from the sorts of multiplicative models that we tend to focus on in I/O psychology. For example, Barrick and Mount (1993) examined auton- omy as a moderator of the relationship between conscientiousness (among other things) and performance. The idea was that the relationship between personality and performance would depend on the situation. This was a good study with important implications. However, it assumed that a given trait was constant across situations. The influence of the trait was hypothesized to vary across situations, but the trait itself was constant. It is another matter entirely to suggest that the level of the trait itself varies with the situation.

Inter-individual Instabilitywithin Trait Levels

There is also evidence of inter-individual instability within trait levels. Apart from the role played by the situation, is it the case that a given trait level is associated with a single pattern of behavior? For example, does high extra- version result in the same constellation of behaviors for every person?

Winter and his colleagues have examined the possibility that motives and traits interact to produce behavior. Motives serve as the impetus while traits

132 I NTERNATIONAL R EVIEW OF I NDUSTRIAL AND O RGANIZATIONAL P SYCHOLOGY 2005 ‘channel’ this impetus into a particular direction (Winter et al., 1998).

Although it must be said that it is unclear what Winter et al. (1998) mean by ‘traits’, the notion of interactions between one aspect of a person (i.e., motives) and another (i.e., traits) is an intriguing one. Winter et al. use a comparison of US Presidents Nixon and Reagan to illustrate. Nixon is offered as the prototype of the person who is high in need for affiliation (motive) but low in extraversion (trait) and perhaps emotional stability. Such a person is driven by a desire for friendship, but whose pursuit of this goal is constrained by other aspects of his personality. This results in

a person who is warm with those close to him or those with whom he is comfortable, but who is awkward and stilted in a more impersonal forum. By contrast, Reagan would be characterized as high in extraversion and emotional stability, but low in need for affiliation. This results in a man who can easily create the impression of warmth from afar, but who, upon closer inspection, seems cold and disengaged.

US President John F. Kennedy might also have been characterized as high in extraversion. We doubt that anyone would draw many similarities between Kennedy and Reagan. Perhaps they differed in their need for affiliation, their need for power, or even their level of cognitive ability. The point is that knowing one’s level on a trait is not enough. In order to predict their actions, we must know more about them.

Each of these sources of instability forces us to question our trait-based approach to personality. This is not to suggest that it is not useful to estimate the correlation between, say, conscientiousness and job performance. The point is that this is a very small part of the total picture. If there is consider- able within-person variability in conscientious behavior across situations, or if the effect of this trait on an outcome depends on other attributes of the person, then the emphasis on a single number to represent the standing of a given person on a dimension like personality seems misguided.