Factor Analysis and the Search for the Structure of Personality PDF

Factor Analysis and the Search for the Structure of Personality 143 research for decades. One particularly interesting study often lost in the long history of this research was conducted by Donald Fiske (1949). In the summer of 1947, Fiske and his colleagues conducted extensive personality assessm...

Factor Analysis and the Search for the Structure of Personality 143 research for decades. One particularly interesting study often lost in the long history of this research was conducted by Donald Fiske (1949). In the summer of 1947, Fiske and his colleagues conducted extensive personality assessment of 128 men who had been admitted into the Veteran Administration’s clinical psychology training program. In addition to many standard trait measures, the assessment included projective tests, biographical data, interviews, and ratings from peers. When the researchers factor analyzed all these data, they identified five basic personality factors. They described these five factors as Social Adaptability (talkative, makes good company), Emotional Control (easily upset, has sustained anxieties), Conformity (ready to cooperate, conscientious), The Inquiring Intellect (intellectual curiosity, an exploring mind), and Confident Self-Expression (cheerful, not selfish). Although the data were not quite as clean as these labels suggest, as we will see, the findings foreshadowed the direction personality research would take decades later. Cattell, Fiske, and the other pioneers in this area faced some practical limitations on their work. They had to calculate the extensive mathematical computations required of factor analysis by hand and, of course, double check each decimal point and each carried number. Personality researchers today have the benefit of larger and more varied sets of data, sophisticated statistical tests, and—most important—computers that conduct extensive calculations quickly and accurately. These developments led to a burgeoning amount of research with a surprisingly consistent pattern in the findings. Although there may never be complete agreement, different teams of investigators using many different kinds of data repeatedly found evidence for five basic dimensions of personality (Digman, 1990; Goldberg, 1992; John, Nauman, & Soto, 2008; McCrae & Costa, 1997, 2008). The five basic factors uncovered in this research look like the ones listed in Table 7.1. Table 7.1 The Big Five Personality Factors Factor Characteristics Neuroticism Worried versus calm; Insecure versus secure; Self-pitying versus self-satisfied Extraversion Sociable versus retiring; Fun-loving versus sober; Affectionate versus reserved Openness Imaginative versus down-to-earth; Preference for variety versus preference for routine; Independent versus conforming Agreeableness Softhearted versus ruthless; Trusting versus suspicious; Helpful versus uncooperative Conscientiousness Well organized versus disorganized; Careful versus careless; Selfdisciplined versus weak willed Source: Copyright © 1986 by the American Psychological Association. Reproduced with permission. McCrae, R. R., & Costa, P. T. (1986). Clinical assessment can benefit from recent advances in personality psychology. American Psychologist, 41, 1001–1003. doi: 10.1037/0003-066X.41.9.1001. No further reproduction or distribution is permitted without written permission from the American Psychological Association. Copyright 2019 Cengage Learning. All Rights Reserved. May not be copied, scanned, or duplicated, in whole or in part. Due to electronic rights, some third party content may be suppressed from the eBook and/or eChapter(s). Editorial review has deemed that any suppressed content does not materially affect the overall learning experience. Cengage Learning reserves the right to remove additional content at any time if subsequent rights restrictions require it. 144 Chapter 7 / The Trait Approach The five factors described in the table have shown up in so many studies using a variety of methods that researchers now refer to them as the Big Five. Remember, these investigators did not begin with a theory about how many factors they would find or what these basic dimensions of personality would look like. Rather, they let the data do the talking. Once researchers saw which traits clustered with one another, they had to come up with descriptive terms for the five dimensions. Although different researchers sometimes use different names, the most commonly used terms are Neuroticism, Extraversion, Openness, Agreeableness, and Conscientiousness. Alert students have recognized that the beginning letters of the five labels cover the OCEAN of human personality. The Neuroticism dimension places people along a continuum according to their emotional stability and personal adjustment. People who frequently experience emotional distress and wide swings in emotions will score high on measures of Neuroticism. People high in Neuroticism tend to become more upset over daily stressors than those low on this dimension and are more vulnerable to bouts of anxiety and depression (Chow & Roberts, 2014; Kotov, Gamez, Schmidt, & Watson, 2010; Lahey, 2009). Although there are many different kinds of negative emotions—sadness, anger, anxiety, guilt—that may have different causes and require different treatments, research consistently shows that people prone to one kind of negative emotional state often experience others (Costa & McCrae, 1992). Individuals low in Neuroticism tend to be calm, well adjusted, and not prone to extreme emotional reactions. The second personality dimension, Extraversion, places extreme extraverts at one end and extreme introverts at the other. Extraverts are very sociable people who also tend to be energetic, optimistic, friendly, and assertive. Introverts do not typically express these characteristics, but it would be incorrect to say that they are asocial or without energy. As one team of researchers explained, “Introverts are reserved rather than unfriendly, independent rather than followers, even-paced rather than sluggish” (Costa & McCrae, 1992, p. 15). As you might imagine, studies find that extraverts have more friends and spend more time in social situations than introverts (Feiler & Kleinbaum, 2015; Selfhout et al., 2010; van der Linden, Scholte, Cillessen, te Nijenhuis, & Segers, 2010). The Openness dimension refers to openness to experience rather than openness in an interpersonal sense. The characteristics that make up this dimension include an active imagination, a willingness to consider new ideas, divergent thinking, and intellectual curiosity. People high in Openness are unconventional and independent thinkers. Those low in Openness tend to prefer the familiar rather than seeking out something new. Given this description, it is not surprising that innovative scientists and creative artists tend to be high in Openness (Kaufman et al., 2014; Rubinstein & Strul, 2007). Some researchers refer to this dimension as Intellect, although it is certainly not the same as intelligence. People who are high on the Agreeableness dimension are helpful, trusting, and sympathetic. Those on the other end tend to be antagonistic and skeptical. Agreeable people prefer cooperation over competition. In contrast, people low in Agreeableness like to fight for their interests and beliefs. Researchers find that people high in Agreeableness have more pleasant social interactions and fewer quarrelsome exchanges than those low on this dimension (Donnellan, Conger, & Bryant, 2004; Copyright 2019 Cengage Learning. All Rights Reserved. May not be copied, scanned, or duplicated, in whole or in part. Due to electronic rights, some third party content may be suppressed from the eBook and/or eChapter(s). Editorial review has deemed that any suppressed content does not materially affect the overall learning experience. Cengage Learning reserves the right to remove additional content at any time if subsequent rights restrictions require it. Cattell Estate Raymond B. Cattell 1905–1998 Raymond Cattell spent most of his childhood by the sea in the resort town of Torquay in the south of England. There he developed a lifelong love for the ocean and sailing. Unfortunately, this happy childhood was interrupted when England entered World War I. Cattell suddenly found himself treating wounded and maimed soldiers in a makeshift wartime hospital. He did not realize until many years later how these experiences would one day affect his choice of careers. Cattell won a scholarship to the University of London, the only member of his family to attend college. A few months before he graduated with honors in chemistry, images of the wounded soldiers returned to him. Suddenly his plans for a career in the physical sciences were no longer appealing. At about the same time, Cattell was impressed with a lecture he attended by the famous psychologist Cyril Burt, who argued that the science of psychology offered the best hope for solving many of ­society’s problems (Horn, 2001). “My laboratory bench began to seem small and the world’s problems vast,” Cattell wrote. “Gradually I concluded that to get beyond human irrationalities one had to study the workings of the mind itself” (1974, p. 64). His decision to study psychology, which “was then regarded, not without grounds, as a subject for cranks,” led him to graduate work at London University. There ­Cattell—and psychology—­ stumbled into a fortunate association. Cattell was hired as a research assistant for the famous psychologist and mathematician Charles Spearman, who was studying the relationship between measures of intelligence. Spearman found evidence for a single general concept of intelligence, as compared to models arguing for many unrelated aptitudes. In the course of this research, Spearman developed the statistical procedure known as factor analysis. Cattell would later use factor analysis to understand the structure of personality. After five years working at various clinics in England, Cattell was tempted to come to the United States by an offer to work with the learning theorist E. L. ­Thorndike at Columbia. He also worked at Clark University until Gordon Allport invited him to join the faculty at Harvard in 1941. It was at Harvard, while working alongside Allport and Henry Murray, that Cattell developed the idea that factor analysis could be a useful tool for personality researchers (Raymond B. Cattell (1997)). He put this insight into practice after joining the faculty at the University of Illinois in 1945, where he spent most of his career. Cattell was always a hard worker, sometimes going into his office on Christmas day. The result of this diligence was 56 books and more than 500 research articles. His decision to study personality clearly was psychology’s gain and physical science’s loss. Jensen-Campbell & Graziano, 2001; Malouff, Thorsteinsson, Schutte, Bhullar, & Rooke, 2010). They are also more willing to help those in need (Habashi, Graziano, & Hoover, 2016). The Conscientiousness dimension refers to how controlled and self-disciplined we are. People on the high end of this dimension are organized, plan oriented, and determined. Those on the low end are apt to be careless, easily distracted from tasks, and undependable. The characteristics that define Conscientiousness often show up in achievement or work situations, which is why some researchers refer to this dimension as Will to Achieve or simply Work. But these characteristics surface in other areas of our lives as well. For example, because they are more likely to develop good 145 Copyright 2019 Cengage Learning. All Rights Reserved. May not be copied, scanned, or duplicated, in whole or in part. Due to electronic rights, some third party content may be suppressed from the eBook and/or eChapter(s). Editorial review has deemed that any suppressed content does not materially affect the overall learning experience. Cengage Learning reserves the right to remove additional content at any time if subsequent rights restrictions require it. 146 Chapter 7 / The Trait Approach health habits and take steps to keep themselves safe, people high in Conscientiousness tend to be healthier and live longer than those low on this dimension (Hampson, Edmonds, Goldberg, Dubanoski, & Hillier, 2015; Sutin et al., 2016; Turiano, Chapman, Gruenewald, & Mroczek, 2015; Weston, Hill, & Jackson, 2015). Of course, the five factors show up when researchers look at responses to selfreport trait inventories. But researchers also find evidence for five basic factors when examining other indicators of personality, such as the terms people use to describe their friends and acquaintances (Watson, Hubbard, & Wiese, 2000) and the way teachers describe their students (Digman & Inouye, 1986; Goldberg, 2001). The five factors emerge in studies with elementary school children (Markey, Markey, Tinsley, & Ericksen, 2002; Measelle, John, Ablow, Cowan, & Cowan, 2005) and appear to be fairly stable over time (Terracciano, Costa, & McCrae, 2006; Vaidya, Gray, Haig, Assessing Your Own Personality Conscientiousness Indicate the extent to which each of the following terms describes you. Use a 9-point scale to indicate your response, with 1 = Extremely Inaccurate and 9 = Extremely Accurate. Careful Careless* Conscientious Disorganized* Efficient Haphazard* Inconsistent* Inefficient* Impractical* Neat Negligent* Organized Practical Prompt Sloppy* Steady Systematic Thorough Undependable* Unsystematic* This scale was developed by Goldberg (1992) to measure Conscientiousness, one of the Big Five personality dimensions. Although different scoring procedures are possible, the most straightforward procedure is as follows (­Arthur & Graziano, 1996): Reverse the answer values for the 10 items with an asterisk (i.e., for these items only, 1 = 9, 2 = 8, 3 = 7, 4 = 6, 5 = 5, 6 = 4, 7 = 3, 8 = 2, 9 = 1). Then add all 20 answer values. Arthur and Graziano (1996) report a mean score of 123.11 for a sample of college students, with a standard deviation of 23.99. *Scale: Big Five Factor Markers for Conscientiousness Source: Copyright 1998 by the American Psychological Association. Reproduced with permission. Goldberg, L. R. (1992). The development of markers for the Big-Five factor structure. Psychological Assessment, 4, 26–42. doi: 10.1037/1040–3590.4.1.26. No further reproduction or distribution is permitted without written permission from the American Psychological Association. Copyright 2019 Cengage Learning. All Rights Reserved. May not be copied, scanned, or duplicated, in whole or in part. Due to electronic rights, some third party content may be suppressed from the eBook and/or eChapter(s). Editorial review has deemed that any suppressed content does not materially affect the overall learning experience. Cengage Learning reserves the right to remove additional content at any time if subsequent rights restrictions require it. Factor Analysis and the Search for the Structure of Personality 147 & Watson, 2002). One team of researchers used college student interviews and questionnaire data from 1939 to 1944 to determine Big Five scores (Soldz & Vaillant, 1999). These scores correlated highly with scores on Big Five measures participants completed when they were 45 years older. Ongoing Questions Related to the Big Five Model Although research on the five-factor model has produced impressively consistent findings and an unusually high level of agreement among personality researchers, several questions about the model remain. First, there is some debate about what the five factors mean. For example, these factors may simply represent five dimensions built into our language. That is, although personality may in reality have a very different structure, our ability to describe personality traits is limited to the adjectives available to us, which may fall into five primary categories. It may also be the case that our cognitive ability to organize information about ourselves and others is limited to these five dimensions. Thus, although people may describe personality as if all traits can be subsumed under five factors, this model may not accurately capture the complexities and subtleties of human personality. In response to these concerns, many researchers have looked at the structure of personality among people who speak languages other than English (McCrae et al., 2004; McCrae & Terrraciano, 2005; McCrae, Terrraciano & Personality Profiles of Cultures Project, 2005). Although a few exceptions to the rule are found, the results from numerous studies indicate that the five-factor model does not merely reflect the structure of the English language but appears to be a universal pattern for describing personality. Second, there remains some disagreement about the structure of the five-factor model (Ashton, Lee, Goldberg, & de Vries, 2009). Some factor-analytic studies find patterns that do not fit well within the five-factor structure. In recent years, researchers have on occasion found evidence for seven (Simms, 2007), six (Ashton & Lee, 2007), three (De Raad et al., 2010), two (Simsek, Koydemir, & Schutz, 2012), and even one (Loehlin, 2012) basic factor(s). Some of the confusion about the number of personality dimensions goes back to the question of what kind of data to include in the factor analysis (McCrae & Costa, 1995). For example, most studies finding five factors do not include traits that are evaluative, such as special or immoral. When these terms are included, researchers sometimes find additional personality factors (Almagor, Tellegen, & Waller, 1995; Benet-Martinez & Waller, 1997). Beyond this, a few personality descriptors simply do not fit well within the five-factor model. These maverick traits include religiousness, youthfulness, frugality, humor, and cunning (MacDonald, 2000; Paunonen & Jackson, 2000; Piedmont, 1999; Saucier & Goldberg, 1998). Different research outcomes may also reflect differences in how broadly or narrowly investigators conceive of personality structure. That is, if we think of personality structure in very broad terms, it may be possible to combine some of the Big Five factors to create a smaller number of dimensions. Similarly, if we want finer distinctions, we can probably divide some of the factors into smaller parts and thereby create a larger number of dimensions. Thus, it is not the case that one investigator’s findings contradict another investigator’s results. On the contrary, Copyright 2019 Cengage Learning. All Rights Reserved. May not be copied, scanned, or duplicated, in whole or in part. Due to electronic rights, some third party content may be suppressed from the eBook and/or eChapter(s). Editorial review has deemed that any suppressed content does not materially affect the overall learning experience. Cengage Learning reserves the right to remove additional content at any time if subsequent rights restrictions require it. 148 Chapter 7 / The Trait Approach the similarities between the factor patterns uncovered using different methods and different populations are really quite remarkable (McCrae, 2001). Indeed, one would be hard-pressed to find many examples of such consistent findings in other areas of personality research. Third, many researchers have looked into the stability of the five factors over time. That is, do our personalities change as we age? The answer appears to be “yes and no.” Let’s start with the no. Researchers who follow individuals over long periods of time generally find that our personalities become fairly stable during our 20s and show little sign of change after the age of 30 (Costa & McCrae, 2006; Ferguson, 2010; Terracciano, McCrae, & Costa, 2010). That is, how you score on measures of the Big Five personality dimensions during your early adulthood is likely to be quite similar to how you will score on those same measures 20, 30, or 40 years from now. On the other hand, researchers sometimes find general trends in Big Five scores over the lifespan (Kandler, Kornadt, Hagemeyer, & Neyer, 2015; Roberts, Walton, & Viechtbauer, 2006; Soto, John, Gosling, & Potter, 2011; Specht, Egloff, & Schmukle, 2011; Stephan, Sutin, & Terracciano, 2014). That is, even though a strong extravert is unlikely to ever become an introvert, people may experience small shifts along some of the dimensions as they age. Although to date research findings are not as consistent as we would like, we can identify a few general patterns (Anusic, Lucas, & Donnellan, 2012; Soto et al., 2011). Older adults tend to be higher than younger adults in Conscientiousness and Agreeableness. People also tend to become lower in Neuroticism as they move through adulthood. Fourth, there are questions about when to use scores from Big Five measures versus scores from specific trait scales. That is, would psychologists be better off relying on only five main traits instead of the hundreds of smaller traits they now use? In most cases, the answer is “No.” Examining a specific trait is usually better for predicting relevant behaviors than measuring a global personality dimension (Mershon & Gorsuch, 1988; Paunonen, 1998; Paunonen & Ashton, 2013). For example, being sociable and being adventurous may be part of the larger personality concept of Extraversion. However, if researchers want to understand how people act in social situations, it is probably more useful to examine their sociability scores than to measure the more general dimension of Extraversion. This is exactly what researchers found when they looked at cooperative and competitive behavior (Wolfe & Kasmer, 1988). Although Extraversion scores predicted who would act cooperatively and who would act competitively, researchers obtained even better predictions when they looked at scores for sociability. Another example makes the point even clearer. Scales designed to measure the Big Five personality dimensions usually combine items measuring anxiety with items measuring depression as part of the global dimension Neuroticism. Although it makes sense that both anxiety and depression contribute to this larger dimension, surely psychotherapists and researchers would want to know whether their clients and participants suffer more from one of these emotional difficulties than the other. This is not to say that understanding where an individual falls on the basic five dimensions is not useful. On the contrary, research suggests that, among other uses, the Big Five model can be valuable for diagnosing clinical disorders and working with therapy patients (Lyman & Miller, 2015; Widiger & Costa, 2013; Widiger & Presnall, Copyright 2019 Cengage Learning. All Rights Reserved. May not be copied, scanned, or duplicated, in whole or in part. Due to electronic rights, some third party content may be suppressed from the eBook and/or eChapter(s). Editorial review has deemed that any suppressed content does not materially affect the overall learning experience. Cengage Learning reserves the right to remove additional content at any time if subsequent rights restrictions require it. The Situation Versus Trait Controversy 149 2013) and for identifying problem health behaviors (Atherton, Robins, Rentfrow, & Lamb, 2014; Israel et al., 2014). In addition, as you will see later in this chapter, how people score on measures of the Big Five dimensions is often related to how they perform on the job. The Situation Versus Trait Controversy T he trait concept has come a long way since Allport’s early battles to gain acceptance for his theory. Trait measures have been embraced by psychologists from nearly every perspective and used by professionals working in a wide variety of ­settings. Patients admitted to mental health facilities often spend hours taking tests that yield scores on a variety of traitlike measures. Educators commonly use achievement and aptitude measures to classify children and identify problem cases. Anyone who has gone through the American education system in recent years can recall many such tests, often beginning in the first grade. And for several decades now, academic personality researchers have been busy developing trait measures and correlating scores with a number of behaviors. Criticism of the Trait Approach “Can ­personality psychologists predict behavior? Yes, of course we can—­ sometimes.” Walter Mischel Unfortunately, along with the widespread use of personality measurement comes the possibility of abuse. Several decades ago, one psychologist in particular criticized the way many psychologists were using and interpreting test scores. Walter Mischel (1968) pointed out that too many psychologists relied on one or two scores to make important decisions, such as psychiatric diagnoses or whether an individual should be imprisoned. Although critics accused Mischel of denying the existence of personality traits, he argues that this was never his point (Mischel, 1973, 1990, 2009). Mischel maintains his complaint was with the overinterpretation of personality test scores. As a result of the discussion and debate that ensued, most psychologists today are aware of the dangers of overreliance on test scores. Psychologists who might have once used a single test score now consider information from a number of relevant sources before making diagnoses or recommending a certain type of education program. Although the trait approach is as popular today as it has ever been (Swann & Seyle, 2005), Mischel and other critics raised important questions about some of its key assumptions. It is useful to look at two of these criticisms in particular, as well as the responses trait researchers gave in their defense. First, critics argued that trait measures, as well as other types of test scores, do not predict behavior as well as many psychologists claim. Second, critics maintained that there is little evidence for the consistency of behavior across situations. Trait Measures Do Not Predict Behavior Well At the heart of this argument is the issue of whether personality or the situation determines our behavior. Do you act the way you do because of the situation you are in or because of the kind of person you are? Advocates on one extreme argue that our behavior is almost entirely determined by the situation. Although these psychologists don’t assert that everyone acts the same in a given setting, they often refer to individual Copyright 2019 Cengage Learning. All Rights Reserved. May not be copied, scanned, or duplicated, in whole or in part. Due to electronic rights, some third party content may be suppressed from the eBook and/or eChapter(s). Editorial review has deemed that any suppressed content does not materially affect the overall learning experience. Cengage Learning reserves the right to remove additional content at any time if subsequent rights restrictions require it. 150 Chapter 7 / The Trait Approach differences in behavior as merely “error variance.” Advocates on the other extreme claim that stable individual differences are the primary determinants of how we act. Early in this debate, some psychologists sought an answer to the personversus-situation question by measuring how well personality scores and situations predict people’s behavior. Typically, this research found that both the person and the situation were related to behavior and that knowing about personality and the situation was better than having information about only one (Endler & Hunt, 1966, 1968). Unfortunately, this approach has a major weakness. The results of any such investigation are limited by the type of situation and the kind of personality variable examined. For example, we can think of situations in which nearly all people react the same. It would be absurd to try to predict whether high- or low-self-esteem people will run outside when a building catches fire. Although the situation would account for nearly all the variance in this case, it also would be incorrect to conclude that differences in self-esteem are therefore not related to behavior. If we look at other behaviors in other situations, such as how people react to criticism, we will probably find large differences between high- and low-self-esteem individuals. Today, most psychologists agree that the person and the situation interact to determine behavior (Funder, 2009). Knowing only that a person is high in aggressiveness or that a particular situation is frustrating is not as useful for predicting behavior as knowing both of these facts. In this example, researchers would expect the highest amount of aggression when an aggressive person is placed in a frustrating situation. This way of looking at the relationship among traits, situations, and behaviors is called the person-by-situation approach. Nonetheless, it is appropriate to ask how well traits by themselves predict behavior. Mischel’s criticism was that personality trait scores rarely correlate with measures of behavior above the .30 or .40 correlation coefficient level. This “personality coefficient,” as it is derogatorily called, statistically accounts for only about 10% of the variance in behavior, which leaves 90% unexplained by traits. There Is Little Evidence for Cross-Situational Consistency In one of the earliest studies on personality traits, a research team spent several years looking at honesty in more than 8,000 elementary school children (Hartshorne & May, 1928). They measured honesty in 23 different ways (lying, cheating, stealing, and so on) and found an average intercorrelation among these measures of only .23. Because personality traits are assumed to show some consistency across situations, this finding was widely cited as a challenge to the trait approach. Knowing that a child is honest in one situation, such as telling the truth to a parent, may reveal little about whether the child will cheat on the playground or steal something from another child’s desk. Mischel used this and other research findings to challenge the assumption of cross-situational consistency in traits. Although most of us think we see consistency in personality across different situations, Mischel referred to this belief as “more apparent than real.” For many reasons, he argued, we tend to see consistent behavior that, on close examination, is not really there. For example, people often see what they expect to see. If you expect Karen to be unfriendly, you tend to notice when she insults someone but ignore the times when she pays a compliment. In addition, we typically Copyright 2019 Cengage Learning. All Rights Reserved. May not be copied, scanned, or duplicated, in whole or in part. Due to electronic rights, some third party content may be suppressed from the eBook and/or eChapter(s). Editorial review has deemed that any suppressed content does not materially affect the overall learning experience. Cengage Learning reserves the right to remove additional content at any time if subsequent rights restrictions require it. The Situation Versus Trait Controversy 151 see people in only one type of situation or role and fail to fully realize the extent to which the situation, not the person, is responsible for the behavior. Students are sometimes surprised to find that their stuffy, conservative professor is a fun-­loving, adventurous person outside the classroom. Finally, sometimes the way we treat people causes them to act more consistently than they otherwise might. If you assume Ron is going to be hostile, you will probably approach him in such a confrontational way that he will react with hostility. For all these reasons, we may see people acting more consistently across situations than they really are. In Defense of Personality Traits Naturally, attacks on something as central to personality theory as the use of traits have not gone unchallenged. Responses to Mischel’s criticisms center around the question of how behaviors and traits are measured and the importance of the variance these traits explain. Measuring Behavior Proponents of the trait approach argue that, on the surface, denying the existence of personality traits is absurd (Epstein, 1980, 1983). If behavior were completely inconsistent over time and across situations, how would we know whom to marry or whom to hire? Without predictable behavior patterns, we might as well marry someone at random because our spouse’s behavior will change from day to day depending on the situation. Trait psychologists argue that researchers often fail to produce strong links between personality traits and behavior because they don’t measure behavior ­correctly. The typical investigation held up by critics uses trait scores to predict only one measure of behavior. For example, investigators might measure the number of minutes spent on an activity or ask people to indicate on a 7-point scale the likelihood that they will volunteer for a charity drive. This approach violates a basic concept in psychological testing. A behavior score based on one item or one measure is so low in reliability that it is almost impossible to find a correlation with any test score higher than the .30 to .40 “personality coefficient.” To understand this principle, think about why a final examination would never consist of just one true–false question. A student who knows the material might miss one particular item for any number of reasons. But over the course of, say, 50 items, the student who knows the material is likely to get a higher score than the student who does not. In psychometric terms, the 50-item test has greater internal consistency and is thus a better indicator of the student’s knowledge. Unfortunately, at the time Mischel launched his attack, many trait studies measured behavior with what were essentially one-item tests. A personality trait may be a good predictor of behavior, but psychologists will never know if they don’t measure behavior reliably. As an alternative to one-item measurement, researchers can aggregate data. For example, if you want to measure how much time students spend studying, you’ll obtain a much better score by observing their behavior each night over the course of a few weeks than by observing just one night. Consider a study in which scores on an extraversion scale were used to predict social behavior (Epstein, 1979). Undergraduates in this study recorded the number of social contacts they initiated each day. Copyright 2019 Cengage Learning. All Rights Reserved. May not be copied, scanned, or duplicated, in whole or in part. Due to electronic rights, some third party content may be suppressed from the eBook and/or eChapter(s). Editorial review has deemed that any suppressed content does not materially affect the overall learning experience. Cengage Learning reserves the right to remove additional content at any time if subsequent rights restrictions require it.