Chapter 4 The Trait Perspective

Chapter 4 The Trait Perspective Learning Objectives 4.1 Distinguish between traits and types 4.2 Identify two different pathways to deciding which traits are important 4.3 Identify the five trait dimensions in the five-factor model 4.4 Describe behavioral manifestations of extraversion, agreeab...

Chapter 4 The Trait Perspective Learning Objectives 4.1 Distinguish between traits and types 4.2 Identify two different pathways to deciding which traits are important 4.3 Identify the five trait dimensions in the five-factor model 4.4 Describe behavioral manifestations of extraversion, agreeableness, conscientiousness, openness, and neuroticism 4.5 Contrast the five-factor model to Eysenck's model 4.6 Relate other models of traits to the five-factor model 4.7 Contrast the approach to behavior termed situationism and the one termed interactionism 4.8 Identify the concept of behavioral signature 4.9 Examine assessment and comparisons of profiles as inherent to the trait approach to personality 4.10 Identify implications of the five-factor model for understanding disorder 4.11 Identify two main criticisms of the trait approach to personality "I want you to meet a friend of mine from high school. He's really outgoing. He's friendly, but he doesn't go along with the crowd all the time. You might say he's sociable, but also independent." "My psychology professor is smart, but he's totally hopeless. He must spend all his time in his office. I can't imagine him doing anything interesting or fun. He can't help it, I guess. It's just who he is." The trait approach to personality exemplifies two points made in Chapter 1 about the concept of personality. One is that people are consistent in their actions, thoughts, and feelings over time and situations. Indeed, the concept of trait provides a way of saying that people remain the same people, even as time passes and they move from one situation to another. Traits are qualities that people carry around with them, that belong to them, that are part of them. A second point is that people differ from each other in many ways. The field of personality psychology is guided, in part, by an emphasis on such differences. This emphasis is particularly central to the trait perspective. From this view, a personality consists, in part, of a pattern of trait qualities. The composition of the pattern differs from one person to another. The intersection among traits in any given person defines his or her personality. 4.1: Types and Traits 4.1 Distinguish between traits and types The idea that people differ in fundamental ways goes back at least to Hippocrates (about 400 BC), whose ideas were later embellished by Galen (about AD 150). Back then, the idea was more specifically that people can be divided into four types, or categories: choleric (irritable), melancholic (depressed), sanguine (optimistic), and phlegmatic (calm). Each type was thought to reflect an excess of one of four basic bodily fluids. More recently, Jung (1933) argued that people are either introverts or extraverts. An introvert tends to prefer solitary activities. When facing stress, introverts tend to withdraw into themselves. An extravert prefers to spend time with others. When facing stress, extraverts tend to seek out other people. In a true typology, the types are seen as distinct and discontinuous categories (Figure 4.1, A). Typologies have faded over the years (though there remain some supporters of the idea: for discussions, see Meehl, 1992; Robins, John, Caspi, Moffitt, & Stouthamer-Loeber, 1996; Strube, 1989; York & John, 1992). Figure 4.1 \(A) Early type theories assumed a discontinuity between or among categories of people. (B) Trait theories assume that traits are ­continuous dimensions of variability on some characteristic and that the degree of presence versus absence of the characteristic is ­distributed across a population. Figure 4.1 Full Alternative Text In contrast to typologies, trait theories assume that people occupy different points on continuously varying dimensions (Figure 4.1, B). For that reason, this is sometimes called a dimensional approach. In trait theories, differences among people are seen as quantitative, rather than qualitative. People are seen as differing in how much various characteristics are incorporated in their personalities. 4.1.1: Nomothetic and Idiographic Views of Traits Thus far, we've implied that traits pertain to every person and that people just vary in how much of each quality they have. The belief that traits exist in the same way in every person is called a nomothetic view (Allport, 1961). The term nomothetic derives from the Greek word ­meaning "law." This view holds that everyone stands somewhere on each trait that exists. This allows comparisons among people. In contrast is the idiographic view (Allport, 1961), which emphasizes each person's uniqueness. In Chapter 2, we used this term to refer to an approach to research that focuses on how one person's experience varies across situations. Here, the term implies that traits are individualized. A given trait may exist for only one person in the world. Even if the same term applies to two people, its connotations differ from one to the other (Dunning & McElwee, 1995). Even if the connotations are the same, the trait may differ in importance, so the people can't be compared meaningfully (Britt & Shepperd, 1999). Some people like the idiographic view, because they think the nomothetic view provides no place for uniqueness. In reply, those who favor the nomothetic view say that uniqueness arises from unique combinations of levels on many trait dimensions, but the dimensions themselves are the same for everyone. As Eysenck put it, "the unique individual is simply the point of intersection of a number of quantitative variables" (1952, p. 18). Psychologists who emphasize the idiographic view believe that nomothetic views are always oversimplifications (even though they sometimes use them). Allport (1961) believed we should never lose sight of the fact that even traits that many people have always have a special flavor (maybe differences in how the traits are expressed) that varies from person to person. 4.2: What Traits Matter? 4.2 Identify two different pathways to deciding which traits are important Thinking of personality in terms of traits quickly leads to this question: What traits make up personality? This is a hard question to answer with complete certainty. In fact, there have been serious disagreements about where to start in answering it. Before we describe this disagreement, let's back up and consider a problem that all trait theorists share, along with a tool that helps deal with it. 4.2.1: Factor Analysis Personality is reflected in many ways---for example, in descriptive words. If each word that describes personality meant a different trait, a psychologist would go crazy trying to organize things. That, in a nutshell, is a problem trait psychologists face: bringing order to such diversity. Maybe, though, the many words reflect a small number of ­underlying traits. If so, how do you figure out what the trait dimensions are? A tool used for this is called factor analysis. The basic idea is simple: If two qualities correlate when assessed across many people, they may reflect a trait that contributes to both of them. Patterns of correlation, then, may reveal trait dimensions that lie beneath the measured qualities. Factor analysis is essentially a more complex version of correlation. Instead of looking at one correlation between two variables, a factor analysis looks at correlations among many variables. The process starts by collecting measurements on many variables (typically self-reports or observer ratings) from large numbers of people. Once the data have been collected, correlations are computed between every pair of variables (see Box 4.1). The set of correlations is then put through a procedure called factor extraction. This distills the correlations to a smaller set of factors. Each factor represents shared variations (underlying commonalities) among several measures (rather than just two at a time). Box 4.1 A Closer Look at Factor Analysis The process of factor analysis is complex, but its logic is fairly simple. It's an attempt to find patterns of association in a set of variables. The first step is collecting data. This is more complicated than it might seem. First you have to decide what behaviors you want to measure. Do you want self-reports? Observer ratings? As you can see, the first step---collecting data---entails many decisions. Let's use an example. Imagine you're interested in how people cope with stress. You've decided to use self-reports: people's ratings of how much they did certain things during their most stressful event of the past year. To collect data, get 300 or so of your friends to recall a stressful event and respond to each of 28 items (listing things people sometimes do under stress). Here are some of the items. Took action quickly, before things could get out of hand Refused to believe that it was real Did something concrete to make the situation better Tried to convince myself that it wasn't happening Went on thinking things were just like they were Changed or grew as a person in a new way Tried to look on the bright side of things The second step is to compute the correlation of every item with every other item (panel A, top). Each correlation reflects the degree to which the 300 people tended to answer one item the same as the other item. There are strong correlations between items 1 and 3, between 6 and 7, and between 2 and both 4 and 5 (which also strongly relate to each other). The others are quite weak. Because you had people rate 28 items (instead of just these 7), your correlation matrix is huge. Interpreting the pattern of correlations would be a real chore. The chore is lessened by complicated procedures that reduce your matrix to a smaller number of underlying dimensions (e.g., the links among items 2, 4, and 5 would contribute to one dimension). These dimensions are called factors. Factors are hazy entities you can imagine but can't see. The next step is to compute factor loadings for each item on each factor. Loadings tell you the relations between the items and the factors (panel B, right). Each loading indicates how much the item reflects the underlying dimension. A large number (a high loading) means the item is closely linked to that dimension; a small number means it's not. As shown, items 1 and 3 load on factor A, items 6 and 7 load on factor B, and items 2, 4, and 5 load on factor C. Similar loadings emerge for all your 28 items, letting you know which items go together. Once it's clear which items form factors, you're at the final step: naming the factors. You want to convey the essence of the underlying quality, but your only guide is which items load on it. Often the items are ambiguous. In our example, a couple of factors are easy. The items on factor A show a tendency to try to solve the problem. This might be called problem-focused coping. Given the content of items 2, 4, and 5, factor C might be denial. Factor B seems to be positive reinterpretation or posttraumatic growth or looking on the bright side, but it's hard to be sure which is best. It's important to be careful, though, because the name you use will guide your future thinking. What we've described here is more technically called exploratory factor analysis. It's used when you don't have a good idea of what will go with what. There are also confirmatory factor analyses, which specify ahead of time what items should go on what factors. Those procedures are generally used to refine measures when a good deal is already known about them. Once the factors are extracted, each can be described by a set of factor loadings. Think of these as correlations between the factor and each item (rating) that contributes to its existence. Items that correlate strongly with the factor (usually higher than 0.40 or so) are said to "load on" that factor. Items that don't correlate strongly with the factor are said not to load on it. The items that load on the factor tell you what the factor is "about." The final step in the analysis is labeling the factors. Remember that a factor is defined by which items load on it. Thus, you choose a label to express as well as possible the essence of those items, particularly those with the highest loadings. The factor is viewed as the statistical reflection of a trait. When you name the factor, you are naming the trait. Factor naming is very subjective. Several names might seem equally good, but which name is chosen can have important consequences. People often forget that the label is an inference from the correlations, and they rely on the label to tell them what the trait is. If the label you choose is misleading, it can create huge problems later. Factor analysis as a tool in trait psychology does three things. First, it reduces the multiple reflections of personality to a smaller set of traits. Second, it provides a basis for arguing that some traits matter more than others. That is, if a factor accounts for a lot of variability in the ratings, it reflects an important trait; if it accounts for less, it's less important. Third, it helps in developing assessment devices. You keep items (or ratings) that load strongly (greatly reflect the trait) and discard items that don't. Through repeated item creation and testing, items that don't do a good job of measuring a trait are replaced by better ones. Factor analysis is a very useful tool. It's only a tool, though. What we've told you has a big hole in it. We haven't said anything about what measures to collect in the first place. A factor analysis can tell you only about what you put into it. Thus, the decision about what to measure has a huge impact on what emerges as traits. How do you decide what measures to collect? As noted earlier, different people have started off differently. Let's now return to that question. 4.2.2: Let Reality Reveal Itself The answer some give to that question is that researchers should determine empirically what traits make up personality. If you start with preconceptions, you'll lead yourself astray. This was the argument of Raymond Cattell, an early contributor to trait psychology and one of the first to use factor analysis (Cattell, 1947, 1978; Cattell & Kline, 1977). One empirical approach focused on language as a source of information (see Goldberg, 1982). A language that's evolved over thousands of years has words to describe many qualities. Presumably, any trait that matters has words to describe it. In fact, the more words for a quality of personality, the more it probably matters (for evidence of this, see Leising, Scharloth, Lohse, & Wood, 2014). This is called the lexical criterion of importance. Following this idea, Cattell (1947) took a set of trait terms, collected ratings on them, and factor analyzed the ratings. The emerging factors were the traits he believed mattered. To Cattell, personality is captured in a set of 16 dimensions. The dimensions reemerged in analyses across various types of data, and he saw them as the primary traits in personality. These factors provided a name for his personality inventory: the 16 Personality Factor inventory, or 16PF (Cattell, Eber, & Tatsuoka, 1977). 4.2.3: Start from a Theory Not everyone agreed that an empirical starting point is best. Another major contributor to trait psychology, Hans Eysenck (1967, 1975, 1986; Eysenck & Eysenck, 1985), argued that we should begin instead with well-developed ideas about what we want to measure. Then we should set about measuring those qualities well. Eysenck began with the typology of Hippocrates and Galen and observations made by Jung and Wundt (Eysenck, 1967). He set out to study whether the types identified by Hippocrates and Galen (and re-identified by others) could be created by combining high and low levels of two supertraits. The two supertraits Eysenck posed as the key dimensions of personality are extraversion (vs. introversion) and neuroticism (also called emotional stability). The extraversion dimension concerns tendencies toward sociability, liveliness, activeness, and dominance (all of which characterize extraverts). The neuroticism dimension concerns the ease and frequency with which the person becomes upset and distressed. Whereas extraverts prefer exciting activities involving other people, introverts prefer solitary activities and being alone. These two dimensions can create more diversity than you might guess. Table 4.1 portrays four sets of people with combinations of highs and lows on these dimensions. The ancient type label for each group is printed in color. In looking at these people, keep two things in mind: First, although the form of Table 4.1 suggests discontinuity, both dimensions are continuous. Second, the descriptions are of fairly extreme and clear-cut cases. Most people are closer to the middle on both dimensions and thus have less extreme characteristics. Table 4.1 Traits that are common among four categories of people deriving from the two major personality dimensions proposed by Eysenck. Each category results from combining moderately extreme levels of introversion or extraversion with either a high or a low level of neuroticism. (The green labels are the names given to personality types by Galen in the second century AD.) As Table 4.1 indicates, people who are introverted and also emotionally stable (low in neuroticism) tend to be careful, controlled, calm, and thoughtful. The combination of introversion and high neuroticism, on the other hand, creates a more pessimistic and anxious quality. Thus, introverts can differ substantially, depending on their levels of neuroticism. So can extraverts. When extraversion combines with low neuroticism, the result is an easygoing, carefree sociability. High neuroticism in an extravert infuses an excitable aggressive quality. Thus, the impact of one dimension differs as a function of the person's location on the other trait dimension. In the terms used in Chapter 2, the traits interact. Eysenck assessed these dimensions using self-report measures (Eysenck & Eysenck, 1975). He also used factor analysis to help create these measures, but he did so with a different goal in mind than Cattell had. Cattell used factor analysis to find out what dimensions exist. Eysenck used factor analysis to refine his scales, by selecting items that loaded well, and to confirm that the scales measure two factors, as he intended. Although Eysenck and Cattell started out very differently, the trait structures they produced have distinct ­similarities. The two dimensions Eysenck saw as supertraits resemble two of the first three factors of Cattell's 16PF. The similarities are even stronger in second-order factors from the 16PF. A second-order analysis tells whether the factors themselves form factors (correlate in clusters). One second-order factor from the 16PF is virtually identical to extraversion (Cattell & Kline, 1977); another is similar to neuroticism. Another reflection of the convergence can be seen in Eysenck's view that extraversion is at the top of an unfolding hierarchy of qualities (see Figure 4.2), as is neuroticism. Each supertrait is made of component traits (which resemble Cattell's primary traits). Component traits, in turn, reflect habits, which in turn derive from specific responses. Eysenck believed all levels are involved in behavior, but he saw supertraits as the most important. Figure 4.2 Eysenck's hierarchical view of personality as applied to extraversion. The top level of the model (supertraits) subsumes the elements represented at the next-lower level (traits). These elements, in turn, are made up of yet lower-order qualities (habits), which are made up of associations between stimulus and response. Figure 4.2 Full Alternative Text Two more points about Eysenck's view: First, he believed that extraversion and neuroticism link to aspects of nervous system functioning. (This aspect of his theory comes up in Chapter 7.) Second, there's a third dimension in Eysenck's view, called psychoticism, which has received less attention than the others (Eysenck & Eysenck, 1976). It involves, in part, a tendency toward psychological detachment from, and lack of concern with, other people. People high in this trait tend to be hostile, manipulative, and impulsive (Eysenck, 1992). 4.2.4: The Interpersonal Circle as Another Theoretical Starting Point Another theoretical starting point emphasized interpersonal aspects of personality. Jerry Wiggins and his colleagues (Wiggins, 1979; Wiggins, Phillips, & Trapnell, 1989) argued that the core human traits concern interpersonal life. Wiggins proposed a set of eight patterns, which he called the interpersonal circle, arrayed around two dimensions underlying human relations (Figure 4.3). The core dimensions are dominance (or status) and love. Figure 4.3 The interpersonal circle, a set of personality patterns portrayed in terms of their relative prevalence of two traits: love (the horizontal dimension) and dominance (the vertical dimension). The midpoint of each trait is the point where the lines cross. Source: Based on Wiggins, Phillips, & Trapnell, 1989. Figure 4.3 Full Alternative Text Wiggins argued (as did Eysenck) that diverse personalities arise from combinations of values on the two core dimensions. A person who's high in dominance and toward the cold-hearted end of love will seem arrogant and calculating. Put the same degree of dominance with warmth on the love dimension, though, and you get a person who's gregarious and extraverted. Introversion and extraversion do appear on this interpersonal circle (in the lower-left and upper-right corners of the figure). Here, though, they don't represent a fundamental dimension. Instead, they are seen as resulting from the intersection of two other qualities. 4.3: The Five-Factor Model 4.3 Identify the five trait dimensions in the five-factor model Despite the different starting points taken by various people, a substantial consensus has begun to emerge about what traits are basic. The emerging consensus has overtones of several ideas already presented, but it extends beyond them. The emerging consensus is that the structure of personality may incorporate five superordinate factors. These are often referred to as the five-factor model or the big five (Goldberg, 1981; McCrae & Costa, 2003; Wiggins, 1996). Evidence for a five-factor view of personality structure accumulated slowly for a long time (Digman, 1990). During the 1980s and 1990s, there was an explosion of work on this topic. Diverse samples have been studied, including teachers' ratings of children (Digman & Inouye, 1986), peer ratings (McCrae & Costa, 1987), and nonverbal assessments (Paunonen, Jackson, Trzebinski, & Forsterling, 1992). The model was also tested against measures developed from entirely different lines of thought (Costa & McCrae, 1988a; McCrae & Costa, 1989). Peabody and Goldberg (1989; Peabody, 1984) used scales that were chosen to be sure there were enough common trait words, instead of words that mean more to psychologists than to other people. Haas (2002) even explored the idea that proverbs capture the five factors. Data have now been collected from many cultures and languages. The findings, as a group, suggest that the five factors transcend many boundaries of language and culture (e.g., Benet-Martínez & John, 1998; Church, 2001; Katigbak, Church, Guanzon-Lapeña, Carlota, & del Pilar, 2002; McCrae & Costa, 1997; McCrae, Zonderman, Costa, Bond, & Paunonen, 1996; Paunonen et al., 1992; Saucier & Ostendorf, 1999; Somer & Goldberg, 1999; Stumpf, 1993). The cultures examined in this work are as diverse as those of Turkey (Somer & Goldberg, 1999) and the Philippines (Katigbak et al., 2002). One study of observer ratings collected data in 50 cultures (McCrae, Terracciano, et al., 2005). It has even been argued (Gosling, 2001) that the factors (or at least some of them) apply to lower animals! There have been some failures to find the pattern and some imperfections in the findings (e.g., Benet & Waller, 1995; Church & Burke, 1994; Di Blas & Forzi, 1999; ­Lanning, 1994). The pattern is clearest in Western languages and hard to find in some other languages (Saucier & Simonds, 2006), and it's a bit fuzzier among people with less education (Rammstedt, Goldberg, & Borg, 2010). Yet the body of work, as a whole, is impressive in its fit to the five-factor model (McCrae & Costa, 1997, 2003; Ozer & Reise, 1994). 4.3.1: What Are the Five Factors? Given what we've said so far, what comes next may surprise you. There's still a certain amount of disagreement as to exactly what the five dimensions are (Johnson & Ostendorf, 1993; Pytlik Zillig, Hemenover, & Dienstbier, 2002; Saucier, 1992). The disagreement has at least two sources. First, recall that naming factors can be hard. You do it by looking at the items that load on the factor and trying to extract the underlying thread that connects them. But most words have several connotations, and trait words often portray blends of factors rather than only one factor per word (Hofstee, de Raad, & Goldberg, 1992). Naturally, then, there are disagreements in interpretation. Second, exactly what a factor looks like depends on what items are used. If a particular quality is left out, its importance to a trait will be missed (Peabody & Goldberg, 1989). Thus, studies with slightly different measures can lead to different conclusions about what defines the factors, even when it's clear that more or less the same factors emerged. Table 4.2 displays the five traits, using a variety of names for each. Peabody and Goldberg (1989) suggested that the five factors are the metaphorical equivalent of a piece of music in which there's a theme and a series of variations on it. That's pretty much what you see in Table 4.2. The labels listed under each factor all share a theme, but there are also variations. Some of the basis for the variation is displayed in Table 4.3, which lists examples of the descriptive terms that loaded on the five factors in one study or another. Table 4.2 Labels used by various authors to refer to the "Big Five" factors in personality. Labels in the rows are from (in order) Fiske (1949), Norman (1963), Borgatta (1964), Digman (1990), and Costa and McCrae (1985). The final row provides a characterization by Peabody and Goldberg (1989) of the life domain to which the trait pertains. The first factor is usually called extraversion, but there's a good deal of variation in what it includes. This helps account for the different labels. Sometimes it seems based in assertiveness, sometimes in spontaneity and energy. Sometimes it's based in dominance and confidence, sometimes in a tendency toward happiness. It often conveys a sense of sociability (Watson, Clark, McIntyre, & Hamaker, 1992), but some argue that that's a by-product of other features of extraversion (Lucas, Diener, Grob, Suh, & Shao, 2000). Extraverts do, however, interact more with others in day-to-day life (Srivastava, Angelo, & Vallereaux, 2008). There's a greater deal of agreement (though still not unanimity) about the meaning of the second factor. Neuroticism, or emotional stability, is regarded by most people as being what Eysenck referred to with those labels. Although there are other overtones, what's at the heart of this factor is the subjective experience of anxiety and general distress. The third factor in Table 4.3 is most commonly called agreeableness. This trait is often characterized as reflecting a concern with maintaining relationships. It can also imply nurturance and emotional supportiveness, which require inhibition of negative affect (Graziano & Eisenberg, 1999). Indeed, such inhibition seems to occur automatically among persons high in agreeableness (Haas, Omura, Constable, & Canli, 2007). The opposite pole of this dimension has an oppositional or antagonistic quality verging toward hostility (Digman, 1990). Fitting this, people low in agreeableness use power displays as a way to resolve social conflict more than people higher in agreeableness (Graziano, Jensen-Campbell, & Hair, 1996). There's also evidence that they actually experience more conflicts (Asendorpf & Wilpers, 1998). The essence of the fourth factor is also a little hard to capture. The most commonly used label is conscientiousness. However, this label doesn't fully reflect the qualities of planning, persistence, and purposeful striving toward goals (Digman & Inouye, 1986). Indeed, because the word conscientious itself has two shades of meaning, that word loads both on this factor and on agreeableness. That hints that conscientiousness may not be a perfect name for this factor. Digman (1990) suggested that it be thought of as the will to achieve or simply will. Other suggested names include constraint and responsibility. Roberts, Walton, and Bogg (2005) recently examined the qualities that various theorists consider part of conscientiousness and concluded that no single measure of the trait includes all of them. The largest disagreement may concern the last factor. The disagreement stems at least partly from differences in measures. Early on, Cattell measured aspects of intelligence. Then he stopped doing so and started using the term culture to refer to the qualities that remained. The label stuck. Peabody and Goldberg (1989) pointed out, though, that if intelligence-related measures are reintroduced, they join with culture. They suggest the factor should more properly be labeled intellect. Costa and McCrae (1985) favored yet another label: openness to experience. Peabody and Goldberg (1989) argued that Costa and McCrae's measure of this factor taps one aspect of intellect (the imaginative side) but misses the other side (the logical side). They said that when both sides are measured, they merge (implying that this factor is really intellect). On the other hand, there's evidence that qualities of intellect and openness rely on different aspects of the brain (DeYoung, Shamosh, Green, Braver, & Gray, 2009), suggesting that they differ. 4.4: Reflections of the Five Factors in Behavior 4.4 Describe behavioral manifestations of extraversion, agreeableness, conscientiousness, openness, and neuroticism For some time, most work on the five-factor model was aimed at the factors themselves, showing that they exist in diverse cultures and emerge from many ways of assessment. Subsequently, however, researchers have turned more to looking at how the five traits are reflected, or expressed, on the broader canvas of people's lives. 4.4.1: Extraversion and Agreeableness Let's start with the traits that are most social in nature: extraversion and agreeableness. Several projects have suggested that extraversion and agreeableness are both tied to social situations, but in different ways. Extraversion seems to relate to having social impact; agreeableness seems to relate to maintaining positive relations with others (Jensen-Campbell & Graziano, 2001). Fitting this, extraversion predicts being prominent in fraternities and sororities (Anderson, John, Keltner, & Kring, 2001), but agreeableness does not. In a study of adolescents (Jensen-Campbell et al., 2002), extraversion and agreeableness both related to peer acceptance, but agreeableness also protected against being victimized by peers. Adults high in agreeableness also report greater social support from family members (Branje, van Lieshout, & van Aken, 2004), and they're liked more by peers (Wortman & Wood, 2011). All of this makes sense, if agreeableness is largely about maintaining good relations. People high in agreeableness care about maintaining positive relations with others. A variety of other findings fit this idea, as well. Agreeableness predicts endorsement of conflict resolution tactics among children (Jensen-Campbell, Gleason, Adams, & Malcolm, 2003). Agreeable adults get less angry over bad outcomes caused by other people than do less agreeable adults (Meier & Robinson, 2004), thus short-circuiting aggressive responses (Meier, Robinson, & Wilkowski, 2006). But agreeable people are also especially sensitive to violations of communal goals (Kammrath & Scholer, 2011). Agreeable people are less likely to join the military, but the military then seems to make people less agreeable (Jackson, Thoemmes, Jonkmann, Lüdtke, & Trautwein, 2012). Agreeableness has been related to greater responsiveness in parenting (Clark, Kochanska, & Ready, 2000), less negativity in marital interactions (Donnellan, Conger, & Bryant, 2004), and less seeking of revenge after being harmed (McCullough & Hoyt, 2002). Agreeableness also predicts less poaching of romantic partners, less responsiveness to poaching attempts by others (Schmitt & Buss, 2001), and greater cooperation in resolving social dilemmas over resources (Koole, Jager, van den Berg, Vlek, & Hofstee, 2001). Agreeableness has also been linked to less antisocial behavior (Miller, Lynam, & Leukefeld, 2003) and better job performance (Sackett & Walmsley, 2014). Extraversion can also be helpful socially, in ways that differ from effects of agreeableness. Extraverted men interact better with women they don't know than introverts do (Berry & Miller, 2001), and extraverts have the firm handshake that conveys confidence (Chaplin, Phillips, Brown, Clanton, & Stein, 2000). When extraverts and introverts tell catch-up stories to their friends, extraverts construct the stories along with their friends, whereas introverts construct the plots by themselves (Thorne, Korobov, & Morgan, 2007). On the other hand, extraverts are less cooperative than introverts when facing social resource dilemmas (Koole et al., 2001). These two trait dimensions also relate in consistent ways to personal values and life goals. Extraversion relates to valuing achievement and stimulation; agreeableness relates to valuing benevolence and tradition (Roccas, Sagiv, Schwartz, & Knafo, 2002). Extraversion relates to desires for a high-status career, political influence, an exciting lifestyle, and children; agreeableness relates to desires for group welfare and harmonious family relations and actually relates inversely to desires for wealth, political influence, and an exciting lifestyle (Roberts & Robins, 2000). That may be good, because agreeable men earn less than do less agreeable men (Judge, Livingston, & Hurst, 2012). 4.4.2: Conscientiousness, Openness, and Neuroticism Conscientiousness has also received a good deal of attention in recent years. Greater conscientiousness predicts less unsafe sex (Trobst, Herbst, Masters, & Costa, 2002) and other risky behaviors (Markey, Markey, & Tinsley, 2003). Conscientious people are less likely to try to steal someone else's romantic partner and are less responsive to being lured away (Schmitt & Buss, 2001). Conscientiousness has been linked to more responsive parenting of young children (Clark et al., 2000) and to use of negotiation as a conflict-resolution strategy (Jensen-Campbell & Graziano, 2001). Conscientiousness has also been shown to be important in the development of relationships in adolescence (Jensen-Campbell & Malcolm, 2007). Conscientiousness relates to the desire for a career but not necessarily a high standard of living (Roberts & Robins, 2000). Conscientiousness in adolescence predicts higher academic achievement (Chamorro-Premuzic & Furnham, 2003; Noftle & Robins, 2007; Poropat, 2009; Wagerman & Funder, 2007), more interest in science (Korpershoek, Kuyper, van der Werf, & Bosker, 2010), and higher religiousness (McCullough, Tsang, & Brion, 2003). Conscientiousness also predicts peer ratings of greater social influence in organizational settings (Harms, Roberts, & Wood, 2007) and better job success (Sackett & Walmsley, 2014). On the other hand, conscientious people are more affected by losing a job (Boyce, Wood, & Brown, 2010). Conscientiousness also has health implications. In a study of cancer risk factors, conscientiousness led to more restrictive household bans on smoking (Hampson, Andrews, Barckley, Lichtenstein, & Lee, 2000). People high in conscientiousness live longer (Kern & Friedman, 2008; Martin, Friedman, & Schwartz, 2007), presumably because they take better care of themselves (Christensen et al., 2002). Consistent with this, conscientiousness relates to various kinds of health-linked behaviors (Bogg & Roberts, 2004; Roberts et al., 2005) and better following of doctors' orders (Hill & Roberts, 2011). In fact, conscientiousness in childhood has been related to health behaviors 40 years later (Hampson, Goldberg, Vogt, & Dubanoski, 2006). Conscientiousness has also been related to less substance abuse (Chassin, Flora, & King, 2004; Lynam, Leukefeld, & Clayton, 2003; Roberts & Bogg, 2004; Walton & Roberts, 2004) and to less antisocial behavior more generally (Miller, Pedersen, Earleywine, & Pollock, 2003). Openness to experience has also been linked to a range of social experience (McCrae, 1996). Openness to experience predicts greater engagement with the existential challenges of life (Keyes, Shmotkin, & Ryff, 2002), to more favorable inter-racial attitudes (Flynn, 2005), and less likelihood of stigmatizing others (McCrae et al., 2007). Openness relates to greater sexual satisfaction in marriage (Donnellan et al., 2004). People high in openness say they desire artistic expression and devalue the possibility of an easy, lazy life (Roberts & Robins, 2000). They also react less intensely to stress (Williams, Rau, Cribbet, & Gunn, 2009). Neuroticism has been studied for many decades. A high level of neuroticism relates to distress in a wide variety of difficult circumstances. For instance, it relates to more difficult interactions among married partners (­Donnellan et al., 2004) and less satisfaction in the relationship. People who are highly neurotic are also more likely to distance themselves from their partners after a negative event (Bolger & Zuckerman, 1995). Neuroticism impairs academic performance (Chamorro-Premuzic & Furnham, 2003), and it predicts a negative emotional tone when writing stories about oneself (McAdams et al., 2004). Neuroticism also predicts earlier death (Hampson & Friedman, 2008), partly (but not exclusively) because people higher in neuroticism smoke more (Mroczek, Spiro, & Turiano, 2009). Death comes even sooner if one develops an even higher level of neuroticism over time (Mroczek & Spiro, 2007). 4.5: Relations to Earlier Trait Models 4.5 Contrast the five-factor model to Eysenck's model Today, when people think of trait psychology, they generally think first of the five-factor model. However, recall from earlier in the chapter that several other trait models preceded this one. How does the five-factor model relate to them? The easiest comparison is to Eysenck's theory. It's obvious from Table 4.2 that two of the "big five" are virtually the same as Eysenck's supertraits: extraversion and emotional stability. It's been suggested that Eysenck's third dimension, psychoticism, is a blend of agreeableness and conscientiousness (Goldberg, 1993b; Zuckerman, Kuhlman, Joireman, Teta, & Kraft, 1993). A second similarity to Eysenck is that the five factors are superordinate traits, incorporating narrower traits. For example, in Paul Costa and Robert McCrae's (1985, 1992) NEO Personality Inventory (NEO-PI-R; NEO stands for neuroticism, extraversion, and openness; agreeableness and conscientiousness were added after the name was coined; the R stands for revised), there are measures of six narrow traits for each domain of the five-factor model. The six narrow traits combine into a score for that supertrait. Thus, many people who use the five-factor model share with Eysenck the idea that the core traits are supertraits, which are, in turn, composed of more specific facet traits. Another useful comparison is with the interpersonal circle. The basic dimensions of the circle are dominance and love. Love may be equivalent to agreeableness. If dominance were seen as roughly equivalent to extraversion, the interpersonal circle would comprise two factors of the five-factor model (McCrae & Costa, 1989; Peabody & Goldberg, 1989). Trapnell and Wiggins (1990) expanded a measure of the interpersonal circle to have additional scales and an even better fit to the five-factor model (see also Saucier, 1992). This comparison with the interpersonal circle also raises an issue, however. As noted earlier (see Figure 4.3), Wiggins saw extraversion as a combination of two qualities in the circle, not as a basic dimension. Doesn't this conflict with the five-factor model? It depends on how you define extraversion. Remember, there are diverse opinions on how to view that factor. If it's really about dominance and assertiveness, it would fit with the interpersonal circle. To summarize some of the points made thus far, the five-factor model of personality structure has emerged as a candidate for integrating a variety of earlier models. The data make this set of broad traits look very much as though they represent universal aspects of personality (McCrae & Costa, 1997). Remember, though, that what comes out of a factor analysis depends on what goes into it. It can be dangerous to draw conclusions too fast. Nonetheless, at present the five-factor model seems to offer the best promise of a consensus about the dimensions of personality that trait psychology has ever seen. 4.6: Other Variations 4.6 Relate other models of traits to the five-factor model Consensus is not unanimity, however. People have disagreed with this view for a variety of reasons (e.g., Block, 1995, 2001; Eysenck, 1992, 1993; Zuckerman, 1992). Several other trait models also exist that differ from the five-factor model in various ways. One is Tellegen's (1985) model. It greatly resembles Eysenck's (1975, 1986) in having three supertraits, though with somewhat different origins and overtones. Tellegen (1985) recast neuroticism slightly as a tendency to experience negative emotions, and he recast extraversion as a tendency to experience positive emotions. Positive emotionality (like extraversion) has been tied to social success, and negative emotionality (like neuroticism) has been tied to indices of poor adjustment (Shiner, Masten, & Tellegen, 2002). Tellegen's third factor, constraint, resembles psychoticism in Eysenck's model but viewed from the opposite direction. It also predicts similar outcomes: Low constraint is linked to criminal and antisocial behavior (Krueger, 2002; Shiner et al., 2002; Shoal & Giancola, 2003). The idea of five factors was carried in a somewhat different direction by Zuckerman and his colleagues (1993), who proposed an "alternative 5." Once again, remember that what comes out of a factor analysis depends partly on what goes into it. These people put slightly different things in. This model has a sociability factor that resembles extraversion (if you view extraversion as mostly social). Neuroticism--anxiety is most of neuroticism but without the hostility that others include there. Hostility moves to aggression--hostility, which otherwise looks like agreeableness (reversed). Impulsive sensation seeking looks like conscientiousness (reversed). The last factor in this model is Activity. What may be the most important difference between this and the other five-factor model is that Zuckerman et al. located hostility outside neuroticism. Interestingly, there are reasons why that may actually be a better location for it (Carver, 2004; Jang, Livesley, Angleitner, Riemann, & Vernon, 2002; Peabody & DeRaad, 2002; Saucier & Goldberg, 2001). 4.6.1: Expanding and Condensing the Five-Factor Model The idea that what comes out of a factor analysis depends on what goes in is also reflected in a viewpoint that builds on the five-factor model by adding another factor. Ashton and his colleagues believe the five-factor model is incomplete. In tests involving seven languages, they found a sixth supertrait that they call honesty--humility (Ashton, Lee, Perugini, et al., 2004). Subsequent work established that this factor can also be found in analyses of English words (Ashton, Lee, & Goldberg, 2004). Ashton and his colleagues suggest that this trait tends to be absorbed by agreeableness in some measures but is a distinct quality that stands out on its own, if it's allowed to do so. They developed a measure (and model) that they call the HEXACO framework (Ashton & Lee, 2007), and they have shown that this additional factor adds predictive validity above and beyond the five-factor framework (Ashton & Lee, 2008). Some have made the opposite argument: that the five-factor model can be condensed. That is, putting the five traits into a higher-order analysis yields two factors. The first is defined by (low) neuroticism, agreeableness, and conscientiousness. Digman (1997) called it socialization, because these qualities all influence whether people get along in social relations. The second is defined by extraversion and openness. Digman characterized it as reflecting personal growth, because these qualities influence whether people expose themselves to new things, thereby fostering growth. DeYoung (2006) found the same two higher-order factors and called them stability and plasticity. He argued that they reflect, respectively, an organismic need to maintain a stable organization of psychological functioning and also a need to explore and grow. 4.6.2: Are Superordinate Traits the Best Level to Use? There remains at least one more question to raise, even for people who accept the five-factor model. As we said, this is a model of supertraits. Supertraits have facets. As noted earlier, Costa and McCrae's NEO-PI-R measures six facets of each factor. Those who use the five-factor model sometimes point to the utility of examining patterns of traits within each factor (Costa & McCrae, 1995; Goldberg, 1993a). Is anything lost when lower-level traits are combined to form the supertraits? This is essentially what Cattell and Eysenck argued about when they disagreed about second-order factors (see also Briggs, 1989; Cattell, 1993; Funder, 1991). The evidence suggests that, yes, something is lost when facet traits are merged. Paunonen and Ashton (2001a) compared the "big five" factors to specific facet scales as predictors of 40 behaviors, which were measured by self-reports and peer ratings. The behaviors were chosen because they had some social importance (altruistic behavior, smoking, alcohol consumption, religiosity, and so on). For a substantial number of these behaviors, facet scales added significantly to prediction after the five factors had been entered as predictors. Thus, something is lost if only the "big five" are used. Conceptually similar findings have come from a number of other studies (Mershon & Gorsuch, 1988; Paunonen, 1998; Paunonen & Ashton, 2001b; Schimmack, Oishi, Furr, & Funder, 2004; Wolfe & Kasmer, 1988). Better prediction from specific, narrow traits comes at a cost, though. The cost is that to understand the findings, you have to hold a larger number of traits in mind at once. In general terms, that's the trade-off: Using supertraits creates a picture that's more intuitive and easier to hold in mind, whereas using narrower traits may often give greater accuracy. An in-between position has also been suggested (DeYoung, Quilty, & Peterson, 2007). This position derives from evidence that a broad set of facets within a given supertrait can be reduced to two aspects per trait. DeYoung et al. suggest that this intermediate position provides many of the benefits of the facet approach but keeps the number of variables manageable. 4.7: Traits, Situations, and Interactionism 4.7 Contrast the approach to behavior termed situationism and the one termed interactionism We turn now to a very different issue pertaining to the trait perspective on personality. It derives from an important controversy that occurred over a period from about 1970 to about 1990. 4.7.1: Is Behavior Actually Traitlike? A question shook the foundations of trait psychology in the early 1970s: whether behavior actually shows traitlike consistency. As we said at the start of this chapter, traits are assumed to be stable aspects of personality that influence behavior in a wide range of settings. The reason for assuming traits in the first place was to account for consistency in thoughts and actions across time and circumstances (see also Box 4.2). Differences on a trait should predict differences in trait-related behaviors. Box 4.2 How Stable Is Personality over Long Periods? Discussions of consistency and stability in personality often focus on fairly short time periods. However, the trait concept also implies stability over much longer periods. Do people's personalities stay the same, even years later? Research on this question is hard (it can require following people for decades). Despite that, several projects have been done (for reviews, see Caspi, Roberts, & Shiner, 2005; Roberts, Walton, & Viechtbauer, 2006). As a whole, the evidence is impressive in showing both continuity and change. This may sound contradictory, but it isn't. When investigators look at a given trait across a large number of people across a period of time, they usually find a high degree of stability in how people rank on the trait dimensions (thus, a strong positive correlation over time). Indeed, a review of 152 longitudinal studies found that correlations of traits grow increasingly stronger from college, through middle adulthood, to later adulthood (Roberts & Del Vecchio, 2000; see also Costa & McCrae, 1988b, 1989; McCrae, 1993). Thus, if Rachel is more agreeable than most other people in the sample when she's a senior in high school, she's very likely to be more agreeable than most of the same people when they're all 4 years out of college and when they're all 50. It's also possible to ask a second question about stability, however: Is there an absolute change in a person's standing on a trait dimension over time? That is, if Rachel is a 6 on a scale of 10 on agreeableness at age 18, will she still be about a 6 when she's 28? The answer to this question is that there tend to be systematic overall changes over time. These changes occur both in adolescence and throughout adulthood. Interestingly, although we tend to think of adolescence as a time of great change, Roberts et al. (2006) found that the largest changes occur in young adulthood (from 20 to 40 years old). As a group, adolescents tend to get higher on agreeableness and lower in neuroticism from about age 12 to about age 17 (Klimstra, Hale, Raaijmakers, Branje, & Meeus, 2009). These changes are consistent with developing greater maturity during this period. Across adulthood, people tend to become even higher in agreeableness up to about age 60 and even lower in neuroticism up to about age 40---then the curves flatten out (Lucas & Donnellan, 2011; Roberts et al., 2006). People also become more conscientious as they age, even up to age 70. Openness to experience tends to be stable across adulthood until about age 50, then drifts downward. Extraversion is perhaps the most puzzling case. Results from several studies conflict. Roberts et al. (2006) may have solved the puzzle by splitting extraversion into subcomponents of social vitality (sociability and positive emotion) and social dominance (assurance and agency). Social dominance goes up in adolescence and early adulthood, then stabilizes. Social vitality goes up in adolescence, falls until about age 25, and then falls again starting at about age 55 (Roberts et al., 2006). Thus, even as rank orders stay very stable, overall levels show considerable change. It was somewhat surprising, then, that trait measures and behavior often didn't correlate well (Mischel, 1968; Vernon, 1964). Mischel (1968) pointed out that correlations between trait self-reports and actual behavior typically were modest---around 0.30. This means that the trait accounts for about 9% of the variation in the behavior, with the remaining 91% unaccounted for. What, then, were we to think about traits? If traits don't predict people's actions, then why should the trait concept be considered useful? 4.7.2: Situationism Indeed, some people went so far as to ask why the concept of personality should be considered useful. The extreme form of this view was called situationism: the idea that situational forces determine behavior, not personality. This view was promoted by some social psychologists, who traditionally emphasize the role of the environment, rather than personality, as causes of people's actions. This view argued that correlations between traits and behavior were low because situational variables overwhelm the effect of personality. This turned out to be quite wrong. Funder and Ozer (1983) pointed out that effects of situations and traits are reported with different statistics, making them hard to compare. They returned to several famous studies of the impact of situations on behavior and converted the original statistics to correlations. To the astonishment of many, these correlations were about the same size as the personality coefficients that had been criticized so sharply. 4.7.3: Interactionism Another approach to weak links between traits and actions is interactionism (e.g., Ekehammer, 1974; Endler & Magnusson, 1976; Magnusson & Endler, 1977; Ozer, 1986; Pervin, 1985). Interactionism is the idea that traits and situations interact to influence behavior. Neither the setting alone nor the person alone provides a complete account. The term interactionism is tied in part to a statistical understanding of how two variables (or in this case, two classes of variables) influence an outcome. Recall from Chapter 2 how experimental personality research often combines two variables as factors in a single study. We now restate that point in terms of persons and situations. When a situation and a trait are examined in the same study, there are three sources of influence on behavior. Sometimes, variations in the situation have an effect on everyone; for example, stressful situations may cause everyone to seek out other people for social support. Sometimes, variations on a trait have an effect in all situations; for example, extraverts may always spend more time with other people than introverts. It's also possible, however, for the situation and trait to interact (see Figure 4.4). An interaction here means that variations in the situation affect some people in one way and others in a different way. For example, stress may cause extraverts to seek out others more, but not introverts. This interaction might occur in addition to one or both of the overall effects, or it may occur instead of them. In the latter case, it would create a picture of weak effects for both the trait and the situation. Figure 4.4 Interactionism. (A) Sometimes there's an interaction between a situation and a trait variable, such that variations in the situation affect some people but not others. (B) Sometimes the interaction is even more interesting, with some people being affected one way and other people being affected the opposite way. Figure 4.4 Full Alternative Text In this view, situations and dispositions can interact in several ways to determine behavior. Perhaps the most obvious (the case in Figure 4.4, A) is that a situation may influence one kind of person but not others. Sometimes a situational change causes one kind of behavior in one person and a different behavior in another person. For example, a stressful situation may cause extraverts to seek out others and introverts to withdraw from others (Figure 4.4, B). Some situations act to constrain behavior and hide individual differences. Other situations allow the free expression of personality. Here's another way to describe such interactions: Some situations permit easy expression of personality. Other situations force behavior into channels, thus preventing expression of personality (Monson, Hesley, & Chernick, 1982; Schutte, Kenrick, & Sadalla, 1985). The first set are called weak situations, the second set are called strong situations (Mischel, 1977). As an example, the lawn of a college campus on a Sunday afternoon is a weak situation. Individual differences can be expressed easily; in fact, the situation seems to invite it. An army boot camp is a strong situation. It dampens any expression of individual differences. People exercise choice over the settings they enter, which influences the behaviors they engage in. Some people choose to go to baseball games; other people do not. 4.7.4:  Other Aspects of Interactionism The model just outlined derives from lab research, a context in which researchers put people into identical situations. It tends to assume that people outside the lab also enter identical situations. This, of course, is wrong---a point made by a number of authors (e.g., Buss, 1984; Emmons & Diener, 1986; Emmons, Diener, & Larsen, 1986; Magnus, Diener, Fujita, & Pavot, 1993; Scarr & McCartney, 1983). In life outside the lab (and rarely, but occasionally, even in the lab), people exercise a lot of choice over which environments they enter. Some people choose to go to church, others choose not to. Some people choose to go to basketball games, some to rock concerts, some to country meadows. By choosing the settings they enter, people thereby influence the behaviors they engage in. Indeed, there's evidence that people choose their marriage partners partly by whether the partner lets them be who they are (Caspi & Herbener, 1990). The choices that people make about what situations to enter depend partly on their personalities (Brandstätter, 1983; Emmons & Diener, 1986; Emmons et al., 1986). Another way persons and situations interact is that people differ in the kinds of responses they elicit from others (Scarr & McCartney, 1983). Some people naturally bring a smile to your face, others can make you frown just by entering the room. Introverts tend to steer conversations in one direction, extraverts in another (Thorne, 1987). Indeed, people actively manipulate each other, using such tactics as charm, coercion, and silence (Buss, Gomes, Higgins, & Lauterbach, 1987). All these effects change the situation, so the same situation is actually different for one person than it is for another. This reciprocal influence is another way persons and situations interact. 4.7.5: Was the Problem Ever Really as Bad as It Seemed? Trying to understand why there were weak links from trait to behavior led to a great deal of information about how they relate. In the process, however, doubt arose about whether the problem ever was actually as bad as it seemed to be. After Mischel (1968) wrote that personality correlated with behavior around 0.30, others pointed out that the studies leading to that conclusion weren't the best of studies (Block, 1977; Hogan, DeSoto, & Solano, 1977). Later studies, more carefully designed, found far stronger relationships than that (for a review, see Fleeson & ­Gallagher, 2009). There also turn out to be statistical reasons why a correlation of 0.30 isn't so bad! Most actions are influenced by more than one trait. For example, when you get to a party where you don't know anyone, what you do will depend not only on how extraverted you are but also on how anxiety prone you are. As it happens, whenever a behavior is influenced by several traits at once, the mere fact of multiple influence puts limits on how strong a correlation can be for any single trait (Ahadi & Diener, 1989). This limit looks, in fact, very nearly the same as the much-maligned 0.30 correlation. Maybe the core problem really wasn't ever as bad as it seemed in 1968. But work addressing the problem told us a lot about how behavior emerges. Indeed, this work has led many people to hold a more elaborate view of traits than they might otherwise have developed. We consider this view next. 4.8: Interactionism as Context-Dependent Expression of Personality 4.8 Identify the concept of behavioral signature Psychologists put a lot of effort into developing the ideas known collectively as interactionism. Nonpsychologists, however, seem to naturally approach traits with that ­mentality. That is, people seem to know intuitively that whether a trait affects behavior varies from setting to setting. In reality, you shouldn't expect a given trait to operate all the time---only in situations where it's relevant. This is reflected in the fact that people often use verbal hedges in discussing personality (Wright & Mischel, 1988). A hedge (in this context) means a word or phrase that limits a trait's applicability. As examples, you might describe someone as "shy with strangers" or "aggressive when teased." The ultimate hedge is sometimes. Using a hedge implies that you think the trait-based behavior occurs only in some kinds of situations (Shoda, Mischel, & Wright, 1989). Such evidence, along with the insights of interactionism more generally, led Mischel and Shoda (1995) to a deeper analysis of how traits affect behavior (see also Cervone, 1997, 2004; Mischel, Shoda, & Mendoza-Denton, 2002). In this view, traits are not freestanding tendencies to act, but patterns of linkages between situation and action. Given situation x, action y is likely. A key point is that a given action shouldn't be expected to occur all the time, because the situation that elicits it isn't always present. Thus, a behavior may appear inconsistent across situations---especially situations that differ a lot. But in situations that seem similar to the person, the behavior is consistent (Furr & Funder, 2004). Thus, there's a lot of consistency, despite the variability. Another key point here is that the pattern of linkage between situation and behavior differs from one person to another. This is a source of individuality, of uniqueness: the pattern of situation--behavior links the person has established over time and experience. This pattern is sometimes referred to as the person's behavioral signature. Even if two people tend toward the same kind of behavior, the situations that elicit it may differ from one person to the other. If so, these two people will act differently in many situations, despite having the same trait. This, in fact, may be a way for idiographic traits to exist. Each person's unique pattern of links from situation to action creates a trait that's just a little different from that of any other person. The idea that traits represent patterns of situation--action links opens other possibilities, as well. For example, imagine a person who's mostly an introvert but occasionally acts like an extravert---for example, by becoming talkative. From the perspective of the linkage model, this would mean that there are classes of situations (perhaps infrequent) that link to those actions for this person. To this way of thinking, there would be no contradiction in the idea that a person can display qualities from one end of a trait dimension in one situation and qualities from the opposite end of the dimension in another. Fleeson (2001) has reported considerable support for this argument. He found that most people do things that reflect the entire range of a trait dimension. It's just that the things they do most often reflect a narrower portion of that dimension (see Figure 4.5). In the same way, it's been found that the positive emotions tied to extraversion vary from hour to hour, right along with the degree of extraverted behavior the person is engaging in (Fleeson, Malanos, & Achille, 2002). Figure 4.5 Traits as situation-linked frequency distributions of states. People occasionally act extraverted, even if they are essentially introverts (such as Pati); people occasionally act introverted, even if they are essentially extraverts (such as Jenifer). The person's generalized trait is reflected in the fact that particular sorts of behavioral states are most frequent. Figure 4.5 Full Alternative Text The linkage viewpoint seems to deal well with some problems people have had in thinking about traits. It doesn't distort the trait concept, but it clearly adds something to the concept that was discussed in the first part of this chapter. This theory has other elements that are considered in Chapter 12. For now, the point is that the impact of traits seems context-dependent (see also Fleeson & Leicht, 2006). This conclusion is quite consistent with the interactionist view. 4.8.1: Fitting the Pieces Together Let's put these ideas together with what we discussed earlier. If you had read only the first half of this chapter, you might have been tempted to assume that most trait theorists hold the view portrayed in Figure 4.6, A or B, in which traits have a constant influence on behavior. People who discuss the five-factor model tend not to talk much about how traits and situations interact. It can be easy to infer that panel A or B is what they assume. Figure 4.6 Three views of the effects of traits on behavior (portrayed for the trait of gregariousness). (A) A naive model, in which people are assumed to display their traits at a relatively constant level, no matter what situation they're in (what Magnusson & Endler, 1977, called absolute consistency). (B) A model in which situations influence the overall levels at which the trait is displayed, but people retain the same ordering (relative consistency). (C) An interactionist model, in which some situations (2 and 4) permit or even elicit individual differences, whereas others (1 and 3) don't do so. Figure 4.6 Full Alternative Text But traits don't work that way. The research just described makes that clear. Traits sometimes influence behavior a lot, and sometimes not at all. Whether the trait matters depends on the situation (Figure 4.6, C). This dynamic approach to the role of traits in the constantly varying social environment recognizes complexities in the creation of behavior (see also Box 4.3). Box 4.3 Theoretical Issue: What Really Is a Trait? Throughout this chapter, we've described traits as being the underlying causes of behavior. Taken literally, it's almost as though a trait were a black box inside you that sent out little blips of behavior every now and then. No one knows where the trait is, but theories often seem to treat it as an entity inside the person that generates a range of trait-related behaviors. That's certainly the way early trait theorists talked about it. More recently, though, a lot of people have questioned whether that's really right. Maybe the kind of trait we've talked about in this chapter (e.g., extraversion, or agreeableness) isn't the source of anything. Instead, maybe these traits are really summary statements about sets of action tendencies that themselves are caused by other forces (Ashton & Lee, 2005; DeYoung, in press; Wood, Gardner, & Harms, 2015). This possibility raises a lot of questions. For one, if the behavior tendencies that we think of as reflecting a broad trait don't have a common source, why do they seem to go together? Why does Sandy have most of the characteristics of an extravert (Table 4.3) instead of just one or two? One approach to this question says we should think about the specific bits of behavior as the traits, instead of thinking of traits as broad underlying sources (Wood et al., 2015). Specific behavioral bits co-occur in a person to the extent that they both help that person get some desired outcome, and to the extent that the person also finds them easy to do. Two bits that help get the same outcome will naturally co-occur more often than will one that helps get the outcome and one that has nothing to do with the outcome. In this view, traits are organized into broad clusters because of some constancy among desired outcomes. But it's the bits that are really the traits, not some underlying black box. Another question that arises is what use is the broad trait concept, if it doesn't refer to anything that has an actual existence. Some people are comfortable with the idea that traits are summary statements, but it certainly leaves something of a vacuum. It tends to render the person more a piecemeal aggregation of molecules than a coherent entity. This may turn out to be a place where the trait perspective eventually has to blend together with other perspectives, in which smaller-scale bits of behavior play a larger role, such as the learning perspective and the cognitive perspective. We will see how well that works out in years to come. This picture is certainly more compelling than the simple one. Interestingly enough, though, the core idea isn't all that new. Some trait theorists of earlier eras said much the same thing, but not in as much detail as is used today. As early as 1937, Gordon Allport wrote that "traits are often aroused in one situation and not in another" (Allport, 1937, p. 331). His conception of a trait explicitly included the assumption that the trait doesn't influence all behaviors and that it may not influence a given category of behavior at all times (Zuroff, 1986). Rather, the effect of the trait depends on whether it's evoked in that situation. Allport even believed that people have contradictory traits. The fact that the contradictory traits are aroused by different situations keeps this from being a problem (Fleeson, 2001, 2004). Allport also anticipated another contemporary theme when he noted that people choose the situations they enter and actively change the situations they're in (Zuroff, 1986). Thus, the ideas that would become known as interactionism go back a long way. 4.9: Assessment from the Trait Perspective 4.9 Examine assessment and comparisons of profiles as inherent to the trait approach to personality The trait approach focuses on assessment more than do most other viewpoints on personality. Indeed, the first part of this chapter discussed how various theorists developed measures. In this section, we consider briefly how the measures are used. 4.9.1: Comparing Individuals Using Personality Profiles The trait approach makes extensive use of self-report inventories, which ask people to describe their views of themselves by making ratings of some kind. The most common ratings involve indicating whether an adjective applies to you or not, or where on a dimension or continuum (anchored by opposing adjectives) you'd fall, or whether you agree or disagree with a statement. The ratings may be made as "yes--no" or "agree--disagree" decisions, or they may be made using multipoint scales. Recall that traits are seen as fundamental qualities of personality, reflected in diverse behaviors. For this reason, self-reports usually include ratings for several reflections of each trait being measured. A scale using adjectives would have many adjectives for each trait; a scale made up of statements would include statements implying diverse ways the trait might be expressed. Regardless of the exact form of the inventory, nomothetic trait psychology assumes that everyone can be placed somewhere along each trait dimension. Inventories measuring these traits are used to create profiles. A personality profile describes a person's place on each dimension the inventory measures (see Figure 4.7). Knowing the dimensions and the person's place on each can create a sense of what he or she is like and how he or she will act. Figure 4.7 An illustration of a personality profile, adapted from the NEO-PI-R. The NEO-PI-R provides both an overall profile of the five major factors (top portion) and a profile of the facets within each of the "big five" (lower portion). The top profile provides a quick and simple summary for the person's personality; the other provides a more detailed picture. Source: Reproduced by special permission of the Publisher, Psychological Assessment Resources Inc, 16204 North Florida Avenue, Lutz, Florida 33549, from the NEO Personality Inventory Revised by P. T. Costa, Jr., and R. R. McCrae, PhD, copyright 1978, 1985, 1989, 1992 by Psychological Assessment Resources, Inc. (PAR). Further reproduction is prohibited without ­permission of PAR. Figure 4.7 Full Alternative Text The profile in Figure 4.7 illustrates the kind of information provided by a personality inventory. At first glance, a profile can seem like nothing more than a string of beads (indeed, Allport \[1961\] said that's exactly what they are). Perhaps a better metaphor is a bar code. Nomothetic theorists believe that the profile is where uniqueness lies. You can see from Figure 4.7 that a shift on a single trait changes the balance of a person's qualities. It can thereby change how the person will act in various settings and how the person will seem to someone else. Since every person has a unique combination of trait levels, everyone is different from everyone else. Furthermore, trait theorists believe traits can interact with one another. To put it differently, how a given level of one trait influences behavior may differ from person to person, as a function of where each person is on other traits. For example, two adventuresome people may display their boldness differently as a function of how sociable they are. The highly sociable one may engage in risky interpersonal exchanges, the less sociable one may climb mountains. Thus, a given trait can be reflected in unique ways for each person because of the modifying effect of differences on other traits. (Recall the earlier discussion of extraversion and neuroticism and Table 4.2.) This is true even though any particular trait dimension is the same from one person to another. 4.10: Problems in Behavior, and Behavior Change, from the Trait Perspective 4.10 Identify implications of the five-factor model for understanding disorder The trait approach was the starting point for some of the earliest efforts to assess disorder. Those efforts were based on the idea that problems directly reflect people's traits. Differences among categories of problems occur because each trait (or group of traits) relates to a different kind of problem. The attempt to understand psychopathology from this trait-based viewpoint was largely an attempt to categorize it. Categorizing was a matter of determining the trait indicators in people's behavior that relate to a given class of problem. Some traits relate to problems because the traits themselves are problematic. As noted earlier, Eysenck's model has a dimension termed psychoticism. Psychoticism is a tendency toward certain kinds of problem behaviors, such as antisocial actions and alcohol and drug abuse (Sher, Bartholow, & Wood, 2000). Because people vary in psychoticism, they vary in the degree to which they will likely display those problems. Neuroticism is a tendency toward emotional distress. Many disorders are characterized by a high level of distress. Thus, people who are high in neuroticism are more likely to display those problems than people lower in neuroticism. 4.10.1: The Five-Factor Model and Personality Disorders The emerging influence of the five-factor model of personality has led to renewed interest in the traits related to disorders, especially personality disorders (see Clark, 2007; Costa & Widiger, 2002; Widiger & Mullins-Sweatt, 2009; Widiger & Costa, 2012). Personality disorders are stable, enduring patterns of behavior that deviate from normal cultural expectations and interfere with the person's life or the lives of others. Many theorists suspect that personality disorders are essentially extreme manifestations of several of the "big five" traits (Larstone, Jang, Livesley, Vernon, & Wolf, 2002; Markon, Krueger, & Watson, 2005; Widiger, Trull, Clarkin, Sanderson, & Costa, 2002; Wright et al., 2012). For example, O'Connor and Dyce (2001) found that all personality disorders are represented within the five-factor model. Reynolds and Clark (2001) also found that the "big five" did a good job of representing personality disorder, and that the facet scales (the narrow scales within the five domains) did an even better job. An edited volume containing diverse reviews of relevant evidence and theoretical statements on how the "big five" relates to personality disorders is now in its second edition (Costa & Widiger, 2002). One recent study even found that clinicians find the "big five" more useful clinically than the categories of the diagnostic system (Samuel & Widiger, 2006). This exploration of the five factors and disorders is not limited to personality disorders. The question is being raised more generally about abnormalities of all types (Krueger, Watson, & Barlow, 2005; Nigg et al., 2002; O'Connor, 2002). Might most of them turn out to reflect extremes of specific traits? This area of work will likely continue to be an important focus for more exploration in future years. Even a person prone to being afraid will not experience fear unless he or she encounters a fear-producing situation. 4.10.2: Interactionism in Behavior Problems As described earlier in the chapter, evidence suggesting a poor link between traits and actions led to interactionism. The logic of interactionism is useful not just for understanding normal behavior but also for understanding problems. One tenet of interactionism is that individual differences matter in some situations but not others. As applied to problems, this idea takes on a slightly different twist. Think of a trait as a vulnerability or susceptibility to a problem. Saying a person is susceptible to a problem doesn't mean that he or she has the problem. Rather, it means the problem will emerge more easily for this person than for someone else. To put it in terms of interactionism, the susceptibility matters in some situations but not in others (recall Figure 4.4). The susceptibility usually matters in situations involving a lot of stress. Therefore, this approach to problems is called a diathesis-stress model. (Diathesis means "susceptibility.") In this model, an interaction is required between the diathesis and a stress for the problem to develop (Meehl, 1962). Diathesis-stress models have been quite common in thinking about psychological problems. 4.10.3: Behavior Change What about the process of therapeutic behavior change? The trait approach is inherently a little pessimistic about change. If traits define a person's personality, how can problems be resolved without changing the person's personality? Traits are stable. Any change that therapy produces will likely be in how the traits are displayed in behavior, not in the traits themselves. On the other hand, there is evidence that even introverts are happier when they are induced to act in an extraverted manner (Zelenski, Santoro, & Whelan, 2012). So maybe traits are more malleable than they seem. The interactionist approach also has an implication here. If problems arise through an interaction between susceptibilities and difficult situations, it should be helpful for the susceptible person to avoid entering situations in which the relevant stresses are likely to occur. Avoiding such situations should help prevent the problems from arising. This, of course, is something that people often do on their own. As we said earlier in the chapter, people exercise some control over what situations they choose to enter. Just as some people choose to go to church and some do not, some people choose to avoid situations in which their vulnerabilities place them at risk. Shy people may avoid singles bars, for example. People with short tempers may try to avoid arguments. People who routinely overspend their credit cards may cancel the cards and switch to using only cash. Avoidance isn't always possible. Yet if people learn which stressors they can and cannot handle, this knowledge should make them more effective in managing their lives. 4.11: Problems and Prospects for the Trait Perspective 4.11 Identify two main criticisms of the trait approach to personality The trait view is, in many respects, the most basic of all the approaches to personality. The very concepts of type and trait arose literally thousands of years ago to account for consistency in behavior across time and circumstances. The concepts have been elaborated and embellished over the years, but their core remains much the same. On the other hand, some people find this view unsatisfying. It's been criticized on several grounds (Block, 1995; for more opinions on both sides, see Block \[2010\] and the commentaries that follow it). One problem is that in their early years, trait theories had little to say about how personality works or how the person gets from trait to action. To put it differently, the trait approach had little to say about intrapersonal functioning. This resulted in a picture of personality that seems static and empty. McAdams (1992) called trait psychology the "psychology of the stranger," because it provides information that would be important if you knew nothing about a person but doesn't portray the dynamic aspects of personality. Labeling a person as friendly, sociable, or dominant gives a name to what you see. But it doesn't tell you much about how or why the person acts that way. This has been a major criticism of the trait concept. Several responses have been offered to this criticism. One is that trait psychology doesn't claim to present a complete picture of the person but rather one angle of view (McAdams & Walden, 2010; McCrae, 2010). Another response is that recent years have seen far more serious attempts to develop an understanding of how traits operate on behavior. One example described earlier is the work showing how expression of traits is situation specific. This work doesn't much resemble the trait approach of years past, but it may be the trait approach of the future. Another response is that the trait perspective is developing links to other perspectives that are providing more of a sense of mechanism behind the influence of traits (see Chapter 7). The idea that the trait viewpoint has had little to say about the process side of personality is often made jointly with a second criticism: that trait theories sometimes use circular explanations. As an example, imagine a woman who acts in a dominant manner---not just occasionally but often, and not just in one situation or with one set of people but in many situations, with whoever else is around. You may feel justified in concluding from this that she has a high level of the trait of dominance. But ask yourself two questions and think about your natural answers. Question 1: Why does she behave that way? (Answer: Because she's dominant.) Question 2: How do you know she's dominant? (Answer: Because she behaves that way.) The problem here is that the information about the behavior is used to infer the existence of a trait, which is being used, in turn, to explain the behavior. This is called circular reasoning, because it can go around and around in an endless circle. The circularity can be broken if the trait is used to predict something new, and sometimes, trait theorists do that. However, this view on personality is more vulnerable than most to the criticism of circularity. A final point, which favors the future of the trait approach, is this: No matter how hard various people have tried to dispense with the use of traits as explanatory mechanisms, the trait concept has retained an active place in the working vocabulary of the personality psychologist. The long history of these concepts attests to their hardiness. Somehow, it seems as though the personality psychologist needs them. The fact that they've endured the test of time seems to imply a fundamental correctness that's hard to deny. Summary: The Trait Perspective The trait approach begins with the assumption that personality consists of stable inner qualities, which are reflected in behavior. Types are discontinuous categories of personalities, with each person falling into one category or another. This concept is no longer prominent in personality psychology, however. Traits are continuous dimensions of variability, along which any person can be placed. Most trait approaches are nomothetic, emphasizing how people differ but assuming that the trait dimensions are the same for everyone. An idiographic approach emphasizes uniqueness and treats some dimensions as unique to specific persons. Factor analysis is a tool used by many trait psychologists. It tells what items (or ratings, etc.) go together. The more variability in ratings that a factor accounts for, the more important the factor. Factor analysis also reveals which observations do and don't reflect a factor well, thus helping refine scales. An important question in trait psychology is what traits are basic and important. Some researchers believe we must let reality tell us the structure of personality. Others believe we must start with a theory. Several theoretical views have been developed, including one that emphasizes traits that have a long history in ideas about personality (extraversion and neuroticism) and one that emphasizes traits that are relevant to social interaction (the interpersonal circle). Many now favor the idea that there are five major factors in personality. Evidence for this view is strong, and the five factors have a reasonable fit to aspects of preexisting models of personality structure. There is disagreement about the precise nature of the five factors, but commonly used labels for them are extraversion, agreeableness, conscientiousness, emotionality, and openness. Recent research has examined how these traits relate to behaviors and experiences in people's lives. The usefulness of the trait concept was questioned by the finding that people's behavior often wasn't well predicted from self-reports of traits. This led some to doubt whether traits actually influence behavior. Situationism---the idea that behavior is controlled primarily by situational influences---proved wrong. Interactionism holds that personality and situations interact in several ways to determine behavior. For example, some situations permit or even elicit individual differences, whereas other situations don't. People also choose which situations to enter, and then they influence the nature of the situations by their own actions. The idea that the influence of traits on behavior is dependent on situations has expanded into a broader view of personality structure, in which traits are individualized linkages between situations and actions. This view accounts for stability over time within the person, as well as for variability across situations. This view of the nature of traits provides a sense of process for trait models. Personality assessment from the viewpoint of trait psychology is a matter of developing a personality profile of the person being assessed---a description of where the person falls on all the dimensions being measured by the inventory. To these psychologists, the profile holds the key to understanding the person's uniqueness. Regarding problems in behavior, trait theorists say that some problems result from having a trait that's intrinsically problematic, such as psychoticism or neuroticism. Other kinds of problems stem from having an extreme position on some trait dimension. Interest in the relationship between personality disorder and the five-factor model is growing. The interactionist position suggests the following possibility (termed a diathesis-stress model): Certain dispositions may create a susceptibility to some kind of problem, but the problem occurs only under certain conditions, usually involving stress. Therapeutic behavior change, from the trait perspective, may mean changing how a trait is reflected in behavior, because a person's traits aren't easily altered. Alternatively, it may mean avoiding situations in which the problem behavior arises.