Perhaps the most obvious source of information about a person is self-report data (S-data)—the information a person reveals. Clearly , individuals may not always provide accurate information about themselves for a variety of reasons, such as the desire to present themselves in a positive light. Nevertheless, the journals that publish the latest research in personality reveal that self-report is the most common method for measuring personality.
Self-Report Data (S-Data)
Self-report data can be obtained through a variety of means, including interviews that pose questions to a person, periodic reports by a person to record the events as they happen, and questionnaires. The questionnaire method, in which individuals respond to a series of items that request information about themselves, is by far the most commonly used self-report assessment procedure.
There are good reasons for using self-report. The most obvious reason is that individuals have access to a wealth of information about themselves that is inaccessible to anyone else. Individuals can report about their feelings, emotions, desires, beliefs, and private experiences. They can report about their self-esteem, as well as their perceptions of the esteem in which others hold them. They can report about their innermost fears and fantasies. They can report about how they relate to others and how others relate to them. And they can report about immediate and long-term goals. Because of this potential wealth of information, self-report is an indispensable source of personality data.
Self-report can take a variety of forms, ranging from open-ended "fill in th blanks" to forced-choice true-or -false questions. Sometimes these are referred to as unstructured (open-ended, such as "T ell me about the parties you like the most") and structured ("I like loud and crowded parties"—answer "true" or "false") personality tests. A prime example of the open-ended form of self-report is called the Twenty Statements Test (see A Closer Look on the next page for more information). In this test, a participant receives a sheet of paper that is essentially blank, except for the words "I am" repeated 20 times. There is a space after each of these partial statements, and participants
A Closer Look
The Twenty Statements Test (TST) was published by a pair of sociologists. Manford Kuhn and Thomas McPartland were interested in attitudes people had toward themselves. In 1954, they published the "Who am I?" test. This test asked the participant to simply answer this question by completing the phrase
McPartland developed a way of scoring the test that involved analyzing the content of the person's responses. In addition, the order of each response was thought to be significant (e.g., something mentioned earlier might be more important to the self-definition than something mentioned later).
Psychologists quickly learned of this test, even though it was published in the American Sociological Review, a journal psychologists typically don't read, and began using it in their research. Because the test involved having the participants come up with 20 statements about themselves, it quickly became known in the psychological literature as the Twenty Statements Test.
In the first decade of use by psychologists, the TST was applied mainly to clinical and personality research questions. For example, one study used the TST to see if the self-concepts of persons in "unadjusted" marriages differed from the self-concepts of persons in "well-adjusted" marriages (Buerkle, 1960). Results showed that the persons in adjusted marriages tended to mention their partner, their marriage, and their family more often in their self-definitions than the persons in unadjusted marriages. This finding implies that part of a successful marriage is incorporating the marriage role into one's definition of oneself, so that self-concept includes one's spouse, marital relationship, and family.
In the 1970s, researchers turned a more critical eye on the TST. It is an open-ended questionnaire, so people with low verbal ability do not complete it as quickly or as thoroughly as persons with high verbal ability, leading the test scores to be biased by intelligence differences in participants (Nudelman, 1973). However, if people are given enough time to complete the 20 questions—at least 15 minutes—then it appears that the intelligence bias is eliminated. All in all, the TST survived this decade of questioning and emerged as a measure that the field deemed useful for assessing how people defined themselves.
In the 1980s, the TST was used in the study of timely personality topics, such as the influence of gender and other social roles in people's self-definitions. For example, one study compared married and single women (Gigy, 1980). Married women tended to respond to the "Who am I?" question by mentioning relationships (I am a mother, I am a wife), acquired roles in family life (I am the one who feeds the children), and household activities (I am the one who buys groceries). Clearly, marriage can mean a large change in self-concept, and studies such as this one document the link between social roles and the ways in which individuals see themselves.
There has been a trend toward using culture and ethnicity in self-definitions (Bochner, 1994). This coin cides with the sharp increase in interest in cross-cultural research. One example of a cross-cultural study using the TST is a study that compares people from Kenya with people from the United States. Several groups were compared on the percentage of responses that included references to social group categories (e.g., I am a member of the local school board or I am a player on the local softball team). U.S. college students mentioned social groups in their self-definitions 12 percent of the time. In Kenya, university students mentioned social groups 17 percent of the time. However, for traditional rural Kenyan citizens, results were quite different. Massai tribespersons in Kenya mentioned social groups 80 percent of the time in their responses, and Samburu tribesper-sons mentioned social groups 84 percent of the time in their TST responses (Ma & Schoeneman, 1997). Results such as these show how the culture in which we are raised may have a strong influence on how we view ourselves and what we consider to be important in defining our identity and in answering the question "Who am I?" The Twenty Statements Test is a useful way to measure how people define themselves and to learn what is important to a person's self-understanding. The TST has proven especially effective at identifying the most important components of a person's identity—the ingredients that provide a person with a sense of self-esteem, meaning in life, and sense of belonging in the world of other people (Vignoles, Regalia, Manzi, Golledge, & Scabini, 2006).
are asked to complete them. For example, a person might say , in this order: I am a woman; I am 19 years old; I am shy; I am intelligent; I am someone who likes quiet nights at home; I am introverted; and so on. Personality instruments that use open-ended formats require coding schemes for classifying the responses they obtain. In other words, psychologists must devise a way to score or interpret the participant' s open-ended responses. For example, to get an idea of how outgoing the woman in our example is, the psychologist might count how many statements refer to social characteristics.
More common than open-ended questionnaires are structured personality questionnaires, in which the response options are provided. The simplest form of the structured self-report questionnaire involves a series of trait-descriptive adjectives, such as active, ambitious, anxious, arr ogant, artistic, gener ous, gr egarious, gr eedy, good-natured, xenophobic, and zany. Individuals are asked to indicate whether or not each adjective describes them. The simplest format for presenting these terms is a checklist, such as the Adjective Check List (ACL) (Gough, 1980). In completing the ACL, the individuals merely place a check beside adjectives that they feel accurately describe them and leave blank items that don' t describe them. A more complex method involves requesting participants to indicate in numerical form the degree to which each trait term characterizes them, say on a 7-point rating scale of 1 (least characteristic) to 7 (most characteristic). This is called a Likert rating scale (after the person who invented it), and it is simply a way for someone to express with numbers the degree to which a particular trait describes him or her . A typical Likert rating scale looks like this:
Least characteristic Most characteristic
Most commonly, a personality scale consists of summing the scores on a series of individual rating scales. A personality scale for activity level, for example, might consist of summing up scores from rating scales on energetic, active, and vigorous.
DIRECTIONS: This list contains a series of adjectives. Please read them quickly and put an X in the box beside each one you consider to be self-descriptive. Try to be honest and accurate.
absent-minded active adaptable adventurous affected affectionate soft-hearted cheerful civilized clear-thinking clever coarse cold touchy dependent despondent determined dignified discreet disorderly zany
More common than adjective checklists, however , are self-report questionnaires in the form of statements. Examples of widely used self-report inventories are the NEO Personality Inventory (Costa & McCrae, 1989) and the California Psychological
Inventory (CPI) (Gough, 1957/1987). Sample items from the CPI are I enjoy social gatherings just to be with people; I looked up to my father as an ideal man; a person needs to "show off' a little now and then; I have a very str ong desire to be a success in the world; I am very slow in making up my mind . Participants read each statement and then indicate on an answer sheet whether they agree with the statement and feel that it is true of them or disagree with the statement and feel that it is false about them. Sample items from the NEO Personality Inventory are I like most people I meet; I laugh easily; I often get disgusted with people I have to deal with . Participants indicate the degree to which they agree the item describes them using a 1 to 5 Likert scale, with 1 anchored with the phrase strongly disagree and 5 anchored with strongly agree.
Pick a personality characteristic you would like to measure. Start by writing down a clear definition of that characteristic. For example, you might choose such characteristics as friendly, conscientious, anxious, or narcissistic. Then write a short questionnaire, about five items long, to measure this characteristic. Your items can be statements or adjectives, and they can be open-ended, true-false, or on a Likert response scale. Then give your questionnaire to other people. How easy was it to write items? Do you think your measure accurately assesses the trait?
Self-report measures, like all methods, have limitations and weaknesses. For the self-report method to be effective, respondents must be both willing and able to answer the questions put to them. Yet people are not always honest, especially when asked about unconventional experiences, such as unusual desires, unconventional sex practices, and undesirable traits. Some people may lack accurate self-knowledge. Because of these limitations, personality psychologists often use sources of data that do not rely on the honesty or insight of the participant. One of those sources is observers.
Experience sampling—a new wrinkle in self-report. A relatively new source of data in personality research is called experience sampling (e.g., Hormuth, 1986; Larsen, 1989). In this method, people answer some questions, perhaps about their moods or physical symptoms, every day for several weeks or longer. People are usually contacted electronically (paged) one or more times a day at random intervals to complete the measures. In one study, 74 college students reported on their moods every day for 84 consecutive days (Larsen & Kasimatis, 1990). The investigators were interested in discovering the links between the day of the week and mood. Not surprisingly, they found a strong weekly cycle in the moods of the college students, with positive moods peaking on Friday and Saturday and negative moods peaking on Tuesday and Wednesday (Monday was not the worst day of the week). The introverts turned out to have a much more regular weekly mood cycle than extraverts. That is, the moods of the introverts were more predictable from this 7-day rhythm than the moods of the extraverts. This difference was probably due to the fact that extraverts are less likely to wait for the
Application (Continued )
weekend to do things that put them in a good mood—partying, socializing, or going out for a special meal with friends. Extraverts typically avoid routine in their daily lives, and introverts typically lead more predictable lives.
Although experience sampling uses self-report as the data source, it differs from more traditional self-report methods in being able to detect patterns of behavior over time. Thus, experience sampling provides information not readily available using questionnaires taken at just one point in time. It's an excellent method, for example, for obtaining information about how a person's self-esteem may go up and down over time, or how a person reacts to the stress of life day after day.
Observer-Report Data (O-Data)
In everyday life, we form impressions and make evaluations of others with whom we come into contact. For each individual, there are typically dozens of observers who form impressions. Our friends, families, teachers, and casual acquaintances are all potential sources of information about our personalities. Observer-report data (O-data) capitalize on these sources and provide tools for gathering information about a person's personality.
Observer reports of fer both advantages and disadvantages as sources of personality data. One advantage is that observers may have access to information not attainable through other sources. For example, observers can report about the impressions a person makes on others, his or her social reputation, whether interactions with others are smooth or full of strife, and the person' s relative status within the group hierarchy .
A second advantage of observer -reports is that multiple observers can be used to assess each individual, whereas in self-report only one person provides information. The use of multiple observers allows investigators to evaluate the degree of agreement among observers—also known as inter-rater reliability. Furthermore, statistical procedures, such as averaging the assessments of multiple observers, have the advantage of reducing the idiosyncratic features and biases of single observers. Typically, a more valid and reliable assessment of personality can be achieved when multiple observers are used.
A key decision point that researchers face when using observers is how to select them. Personality researchers have developed two strategies. One strategy is to use professional personality assessors who do not know the participant in advance. The other strategy is to use individuals who actually know the tar get participants. We will discuss each strategy in turn.
One setting in which professional observers are used is the Institute for Personality and Social Research (IPSR) at the University of California at Berkeley. Participants go to the institute for periods of time ranging from one to five days, so that a wide vari ety of in-depth personality assessments can take place. Participants are invited to go to the IPSR as part of
specific studies. For example, one study contacted a set of architects who were judge by their peers to be highly creative, as part of a study to determine the personality predictors of creativity . Another study looked at novelists judged to be creative. A third assessed graduate students in an MBA program to determine the personality predictors of success in business. During studies at the IPSR, trained personality assessors observe the participants in a variety of contexts. Subsequently , each observer provides an independent personality description of the participants.
A second strategy for obtaining observational data is to use individuals who actually know the tar get participants. For example, close friends, spouses, mothers, and roommates have all been used to provide personality data on participants (e.g., Buss, 1984; Ozer & Buss, 1991). The use of observers who have existing relationships with the participant has advantages and disadvantages when compared with professional assessors. One advantage is that such observers are in a better position to observe the tar get's natural behavior . In the relatively public context of an IPSR assessment, in contrast, professional observers cannot witness the more private actions of a person and must settle for observing his or her public persona. A spouse or close friend has access to privileged information often inaccessible through other sources.
A second advantage of using intimate observers is that multiple social personalities can be assessed (Craik, 1986). Each one of us displays dif ferent sides of ourselves to different people—we may be kind to our friends, ruthless to our enemies, loving toward a spouse, and conflicted toward our parents. Our manifest personali ties, in other words, vary from one social setting to another , depending on the nature of relationships we have with other individuals. The use of multiple observers provides a method for assessing the many aspects of an individual' s personality.
Although there are advantages in using intimate observers in personality assessment, there are also drawbacks. Because intimate observers have relationships with the tar get person, they may be biased in certain ways. A participant's mother, for example, may overlook the negative and emphasize the positive features of her child.
In addition to deciding what type of observers to use, personality researchers must determine whether the observation occurs in a natural or an artificial setting. In naturalistic observation, observers witness and record events that occur in the normal course of the lives of their participants. For example, a child might be followed throughout an entire day, or an observer may sit in a participant' s home. In contrast, observation can take place in contrived or artificial settings, such as occur at the IPSR. Experimenters ca instruct participants to perform a task, such as participation in a group discussion, and then observe how individuals behave in these constructed settings. For example, psychologists John Gottman and Robert Levenson have had married couples go to their laboratory and discuss a topic on which they disagree. The psychologists then observe the couple have a small ar gument. The way in which a couple conducts an ar gument can predict the likelihood that the couple will remain together or get divorced (Gottman, 1994). Even the facial expressions displayed during these laboratory conflicts predic subsequent marital outcomes (Gottman, Levenson, & Woodin, 2001).
Naturalistic observation offers researchers the advantage of being able to secure information in the realistic context of a person' s everyday life, but at the cost of not being able to control the events and behavioral samples witnessed. Observation in experimenter-generated situations has the advantage of controlling conditions and eliciting the relevant behavior . But this advantage comes at a cost—sacrificing the real ism of everyday life.
In summary, there are many dimensions along which O-data dif fer, and personality researchers must take these into account. Decisions about whether to use (1) professional assessors or intimate observers and (2) a naturalistic or an artificia setting for observation must be made on the basis of the specific purposes of the per sonality study. The strengths and weaknesses of the options must be evaluated with the goals of the investigation in mind. No single method is ideally suited for all assessment purposes.
Test Data (T-Data)
Beyond self-report and observer -report data sources, a third common source of personality-relevant information comes from standardized tests— test data (T -data). In these measures, participants are placed in a standardized testing situation. The idea is to see if dif ferent people react dif ferently to an identical situation. The situation is designed to elicit behaviors that serve as indicators of personality variables (Block, 1977). An interesting example is the bridge-building test found in Henry Murray' s (1948) classic book The Assessment of Men. In this test, the person being assessed is given two assistants and a collection of wood, rope, and tools, and he or she has the task of building a bridge over a small creek. The person being assessed cannot do the work him- or herself but must instruct the two assistants on how to build the bridge. Unbeknownst to the person being assessed, the two assistants are role-playing: one is acting dim-witted and has trouble understanding instructions; the other is a "know-it-all," who has his or her own ideas about how the bridge should be built and often contradicts the person being assessed. These two "helpers" actually are there to frustrate the person being assessed. While the person being assessed thinks he or she is being observed on leadership skills, the person is actually being evaluated on tolerance of frustration and performance under adversity .
One fascinating example of the use of T-data is Edwin Megargee's (1969) study on manifestations of dominance. Megargee wanted to devise a laboratory test situation in which he could examine the ef fect of dominance on leadership. Toward this end, he first administered the California Psychological Inventory Dominance scale t a large group of men and women who might serve as potential research participants. He then selected only those men and women who scored either very high or very low on dominance. On completion of this selection procedure, Megar gee took pairs of individuals into the laboratory , in each case pairing a high-dominant participant with a low-dominant participant. He created four conditions: (1) a high-dominant man with a low-dominant man; (2) a high-dominant woman with a low-dominant woman; (3) a high-dominant man with a low-dominant woman; and (4) a high-dominant woman with a low-dominant man.
Megargee then presented each pair with a lar ge box containing many red, yellow, and green nuts, bolts, and levers. Participants were told that the purpose of the study was to explore the relationship between personality and leadership under stress. Each pair of participants was to work as a team of troubleshooters to repair the box as fast as possible—by removing nuts and bolts with certain colors and replacing them with other colors. The participants were told that one person from the team had to be the leader, a position which entailed giving instructions to his or her partner . The second person was to be the follower, who had to go inside the box and carry out the menial tasks requested by the leader . The experimenter then told the participants that it was up to them to decide who would be the leader and who would be the follower.
The key variable of interest for Megargee was who would become the leader and who would become the follower, so he simply recorded the percentage of highdominant participants within each condition who became leaders. He found that 75 percent of the high-dominant men and 70 percent of the high-dominant women took the leadership role in the same-sex pairs. When highdominant men were paired with low-dominant women, however, 90 percent of the men became leaders. But the most startling result occurred when the woman was high in dominance and the man was low in dominance. In this condition, only 20 percent of the high-dominant women assumed the leadership role.
From these laboratory findings alone, one migh conclude that the dominant women in this condition were suppressing their dominance, or that the men in this condition, despite being low in dominance, felt compelled to assume a traditional sex role by taking char ge. It turns out, however, that neither of these conclusions was supported. Megargee happened to have tape-recorded the conversations within each pair of participants while they were deciding who would be the leader . When he analyzed these tapes, he made a startling finding: the high-dominant women wer appointing their low-dominant partners to the leadership position. In fact, the highdominant women actually made the final decision about the roles 91 percent of th time. This finding suggests that women are expressing their dominance in a dif ferent manner than the men in the mixed-sex condition.
Megargee's study highlights several key points about laboratory studies. First, it shows that it is possible to set up conditions to reveal key indicators of personality. Second, it suggests that laboratory experimenters should be sensitive to manifestations of personality that occur in incidental parts of the experiment, such as the discussions between the participants. And, third, there are often interesting links between S-data obtained through questionnaires and T-data obtained through controlled testing conditions. Such links enhance the validity of both the questionnaire and the laboratory test of dominance.
Like all data sources, T-data have limitations. First, some participants might try to guess what trait is being measured and then alter their responses to create a specific impression of themselves. A second challenge is the dif ficulty in verifying tha the research participants define the testing situation in the same manner as the exper imenter. An experiment designed to test for "obedience to authority" might be misinterpreted as a test for "intelligence," perhaps raising anxiety in ways that distort subsequent responses. Failure to confirm the correspondence between the conception of experimenters and those of participants may introduce error .
A third caution in the use of T-data is that these situations are inherently interpersonal, and a researcher may inadvertently influence how the participant behave. A researcher with an outgoing and friendly personality , for example, may elicit more cooperation from participants than a cold or aloof experimenter (see Kintz, Delprato, Mettee, Parsons, & Schappe, 1965). The choice of who runs the experiment, in short, including the personality and demeanor of the experimenter , may inadvertently introduce effects that skew the obtained results.
Despite these limitations, T-data remain a valuable and irreplaceable source of personality information. Procedures used to obtain T-data can be designed to elicit
behavior that would be dif ficult to observe in everyday life. They allow investigators to control the context and to eliminate extraneous sources of influence. And they enable experimenters to test specific hypothese by exerting control over the variables that are presumed to have causal influence. For these reasons, T-data procedures remain an indispensable set of tools for the personality researcher .
Personality psychologists have been enterprising in adapting technological innovations for the study of personality . An example of researcher ingenuity is the use of the "actometer" to assess personality dif ferences in activity or ener gy level. The actome-ter is essentially a modified self-winding watch, which can be strapped to the arm or legs of participants (typically , children). Movement activates the winding mechanism, registering the person's activity on the hands of the dial. Of course, day-to-day and even hour -to-hour fluctuations in mood, physiolog , and setting limit the usefulness of any single sample of activity level. However , several samples of activity level can be recorded on dif ferent days to generate composite scores, reflecting, for eac person, whether he or she is hyperactive, normally active, or sedentary (Buss, Block, & Block, 1980).
In one study, preschool children ages 3 and 4 wore actometers on the wrist of the nonfavored hand for approximately two hours (Buss et al., 1980). The dial of each actometer was covered with tape, so that the children would not be distracted. Indeed, in pretesting, the children who could observe the dial became preoccupied with it—sitting in one spot, shaking the device back and forth—a practice that interfered with the usefulness of the measure. The experimenters had to be careful to eliminate data if a child removed the watch during the session or if illness or rainy weather limited the range within which a child' s activity level could be expressed. Several separate recording sessions were held, and the actometer readings were aggregated, in order to obtain a more reliable index of each child' s activity level.
The experimenters then sought answers to three questions: (1) Does activity level measured with the actometer yield the same results as activity level measured through observation? (2) To what extent is activity level stable over time? (3) Do activity level measurements using this mechanical recording device relate to observer-based judgments of personality functioning? To answer these questions, the children's teachers provided observer evaluations using the children' s version of the California Q-Sort—an instrument designed to produce a wide-ranging description of children's personality characteristics (Block & Block, 1980). Examples of items on the Q-Sort are is a talkative individual; behaves in a giving way toward others; is basically submissive; is guileful and deceitful, manipulative, opportunistic; has a high ener gy level. These observations were made when the children were 3, 4, and 7 years old, whereas the actometer measures were recorded at ages 3 and 4.
It turns out that there was a strong correspondence between actometer measures of activity level and the observer-based measures. Activity level also turns out to be moderately stable over time. For example, actometer measures at age 3 showed a moderate correspondence with actometer measures at age 4. Is there any relationship between actometer measurements of activity level and observer -based judgments of personality? The highly active children, as assessed with the actome-ter, were judged by their teachers to be vital, ener getic, and active. In addition, the highly active children were judged to be restless and fidgety—all attributes that ar more or less indicative of hyperactivity . Of particular interest is that the active children were also seen by teachers as uninhibited, assertive, competitive, aggressive physically and verbally , attention-getting, and manipulative of others. Thus, actometer-based activity scores are linked to other personality characteristics, traits that have important consequences for social interaction.
In sum, some aspects of personality can be assessed through mechanical recording devices, such as the actometer . These forms of T-data have several advantages and disadvantages. Their main advantage is that they provide a mechanical means of assessing personality, one that is not hampered by the biases that might be introduced when a human observer is involved. A second advantage is that they can be obtained in relatively naturalistic settings—such as a children's playground. Their primary disadvantage is that relatively few personality dispositions lend themselves readily to being assessed by mechanical devices. There are no mechanical devices, for example, to directly measure introversion or conscientiousness. Nonetheless, mechanical devices can serve as powerful sources of personality data in the domains in which they can be used. Perhaps future technological advances will expand the range of personality traits amenable to mechanical assessment.
A critical source of personality data enjoying a resur gence of interest is physiological measurement. Physiological measures can provide information about a person' s level of arousal, a person' s reactivity to various stimuli, and the speed at which a person takes in new information—all potential indicators of personality . Sensors can be placed on dif ferent parts of a person' s body, for example, to measure sympathetic nervous system activity , blood pressure, heart rate, and muscle contraction. Brain waves, such as reactivity to stimuli, also can be assessed. And even physiological changes associated with sexual arousal can be measured via instruments such as a penile strain gauge (Geer & Head, 1990) or a vaginal bloodflow mete .
In Chapter 7 we go into some detail on physiological measures. For our purposes here—in examining alternative ways of measuring personality—we will look at only one example of using physiological data as a source of personality information. Psychologist Christopher Patrick (1994, 2005) has been studying psychopaths, particularly men in prison who have committed serious crimes against other people, particularly violent crimes. One theory about psychopaths is that they do not have the normal fear or anxiety response that most people have. Things that might make most people anxious may not make the psychopath anxious. To test this idea, Dr . Patrick used a technique called the "eyeblink startle reflex," which had previously been use in studies of fear .
When we are startled, as when a loud noise occurs, we exhibit the startle reflex, which consists of blinking our eyes, lowering our chin toward the chest
and inhaling suddenly. If we are already anxious for some reason, we will exhibit the startle reflex faster than when we are feeling normal. It makes adaptive sens that we will be prepared to have a faster defensive startle if we are already in a fearful or anxious state. You can demonstrate this by showing persons pictures of frightening or unpleasant scenes, such as a snake, a vicious dog, or spiders, which most people find make them a little anxious. If they are startled while looking a these scenes, they will exhibit a faster eyeblink startle response than when they are looking at nonfeared objects, such as a house, a tree, or a table. Interestingly , Patrick found that psychopaths, who were in prison for violent crimes, did not exhibit the faster eyeblink response while viewing the anxiety-producing photographs, suggesting that they were not feeling the same level of fearfulness or anxiety as normal participants viewing these objects. Perhaps psychopaths commit their crimes because they don't have the normal level of anxiety or guilt that prevents most of us from doing anything wrong. This is a good example of how physiological measures can be used to examine and understand various personality characteristics.
A more recent physiological data source comes from functional magnetic resonance imaging (fMRI), a technique used to identify the areas of the brain that "light up" when performing certain tasks such as verbal problems or spatial navigation problems. It works by gauging the amount of oxygen that is brought to particular places in the brain. When a certain part of the brain is highly activated, it draws lar ge amounts of blood. The oxygen carried by the blood accumulates in that region of the brain. The fMRI is able to detect concentrations of iron carried by the oxygen contained in the red blood cells and thus determine the part of the brain that is used in performing certain tasks. The colorful images that emer ge from fMRI brain scans are often quite dramatic.
In principle, fMRI provides a physiological data source that can be linked with personality dispositions, intelligence, or psychopathology . In practice, however , the method has limitations on what it reveals. Since fMRI must compare the "activated" state with a "resting" state, it becomes critical to know what the resting state really is. If men's resting state turns more to sports and women' s resting state turns more to social interactions, for example, it is possible that a comparison of a task such as looking at faces to the resting state would suggest that men and women are performing the task dif ferently, when in fact the dif ference is due entirely to a sex dif ference in the resting state (Kosslyn & Rosenber g, 2004).
One of the key benefits of physiological data is that it is di ficult for partici pants to fake responses, particularly on measures of arousal or reflexive responses such as the eyeblink startle reflex. Nonetheless, physiological recording procedure share most of the same limitations as other laboratory test data. In particular, recording is typically constrained by a relatively artificial laboratory situation, and th accuracy of the recording hinges on whether the participants construe the situation in the manner that the experimenter wants them to construe it.
Another type of T-data are projective techniques, in which the person is given a standard stimulus and asked what he or she sees. The most famous projec-tive technique for assessing personality is the set of inkblots developed by Hermann Rorschach. However ,
there are others—for example, the hand technique, in which the person is given pictures of hands and is asked to make up a story about what the hands just did and what they are going to do next. The hallmark of any projective technique is that the person is presented with an ambiguous stimulus, such as an inkblot or a picture of a hand. The person is then asked to impose structure on this stimulus by describing what he or she sees—for example, what is in the inkblot or what the hand has just done. The idea behind projective techniques is that what the person sees in the stimulus is directly related to what is on his or her mind. What the person sees in the stimulus is interpreted to reveal something about his or her personality . Presumably, the person "projects" his or her concerns, conflicts, traits, and ways of seeing or deal ing with the world onto the ambiguous stimulus.
Projective techniques are considered T-data because all persons are presented with a standard testing situation, all are given the same instructions, and the test situation elicits behaviors that are thought to reveal personality .
To the psychologist interpreting a person' s responses to the inkblots, the content of those responses is important. Someone with a "dependent personality ," for example, might produce a high frequency of responses such as food, food providers, passively being fed, nurturers, oral activity , passivity, helplessness, and "baby talk" (Bornstein, 2005). In addition to content, the psychologist is interested in how the perceptions are formed. For example, one participant might focus on the lines dividing the ink from the white area, whereas another might focus only on the ink. In sum, all projective measures present the participant with ambiguous stimuli, asking him or her to provide structure by interpreting, drawing, or telling a story about the stimuli. Psychologists who advocate projective measures ar gue that they are useful for getting at wishes, desires, fantasies, and conflicts that the participants themselves may b unaware of and, so, could not report on a questionnaire. Others are critical of pro-jectives, questioning their validity and reliability as accurate measures of personality (Wood et al., 1996).
Life-Outcome Data (L-Data)
Life-outcome data (L-data) refers to information that can be gleaned from the events, activities, and outcomes in a person' s life that are available to public scrutiny . For example, marriages and divorces are a matter of public record. Personality psychologists can sometimes secure information about the clubs a person joins; how many speeding tickets a person has received; and whether he or she owns a handgun. Whether a person gets arrested for a violent or white-collar crime is a matter of public record. Success at one's job, whether one is upwardly or downwardly mobile, and the creative products one produces, such as books published and music recorded, are often important outcomes in a person's life. These can all serve as important sources of information about personality .
Personality psychologists often use S-data and O-data to predict L-data. An example that illustrates how O-data can be used to predict important life events is provided by Avshalom Caspi and his colleagues (Caspi, Elder , & Bem, 1987). Based on clinical interviews with mothers of children ages 8, 9, and 10, these researchers created two personality scales to measure ill-temperedness. One scale was based on the severity of temper tantrums; it noted physical behaviors such as biting, kicking, striking, and throwing things, and verbal expressions such as swearing, screaming, and shouting. The other scale assessed the frequency of these temper tantrums. Caspi and his colleagues summed these two scales to create a single measure of temper tantrums. This measure represents O-data, since it is based on the mothers' actual observations. Then, in adulthood, when the participants were 30 to 40 years old, the researchers gathered information about life outcomes, such as education, work, marriage, and parenthood. They then examined whether the personality characteristic of ill-temperedness, measured in childhood as O-data, predicted significant life outcome two to three decades later , measured as L-data.
The results proved to be remarkable. For the men, early temper tantrums were linked with many negative outcomes in adult life. The men who had exhibited temper tantrums in childhood achieved significantly lower rank in their military service. They tended to have erratic work lives—changing jobs more frequently and experiencing more unemployment than those who had not been judged to be ill-tempered as children. Furthermore, such men were less likely than their even-tempered counterparts to have a satisfying marriage. Fully 46 percent of the ill-tempered men were divorced by age 40, whereas only 22 percent of the men in the low temper -tantrum category were divorced by the age of 40.
For the women, early temper tantrums did not have a bearing on their work lives, in contrast to the men. However , the women who had had temper tantrums as children tended to marry men who were
significantly lower than themselves in occupational status; fully 40 percent of th women who had showed temper tantrums as children "married down," compared with only 24 percent of the women who had been even-tempered as children. As with the men, childhood temper tantrums were linked with frequency of divorce for the women. Roughly 26 percent of the women who had had childhood tantrums were divorced by age 40, whereas only 12 percent of the even-tempered women were divorced by that age.
In addition to empirical studies, such as those that predict later divorce from childhood personality, life-outcome data are used in real ways that af fect our everyday lives. Our driving records, including speeding tickets and traf fic accidents, ar used by insurance companies to determine how much we pay for car insurance. Our histories of credit card usage are sometimes tracked by businesses to determine our behavioral preferences, which influence the advertisements we get sent. And more recently, advertisers sometimes track the websites we visit and use e-mail "spam" and pop-up advertisements based on our patterns of Internet surfing. Thus, driving records, credit card usage, and patterns of Internet usage have become modern sources of L-data. Do you think we can predict these patterns of publicly traceable data from personality variables, such as impulsivity (more driving accidents), status striving (credit card purchase of prestige possessions), and sex drive (more frequent visiting of pornography websites)? Future studies of L-data will shortly answer these questions.
In sum, L-data can serve as an important source of real-life information about personality. Personality characteristics measured early in life are often linked to important life outcomes several decades later . In this sense, life outcomes, such as work, marriage, and divorce, are, in part, manifestations of personality . Nonetheless, it must be recognized that life outcomes are caused by a variety of factors, including one' s sex, race, and ethnicity and the opportunities to which one happens to be exposed. Personality characteristics represent only one set of causes of these life outcomes.
Think of a personality characteristic that you find interesting. For example, you might consider such characteristics as activity level, risk taking, temper, or cooperativeness. Using the four main data sources, think of ways that you might gather information on this characteristic. Give specific examples of how you could assess this characteristic using S-, O-, T-, and L-data as sources of information on people's level of this characteristic. Be specific in providing examples of how and what you might do to assess your chosen personality characteristic.
Now that we have outlined the basic data sources, it is useful to take a step back and consider two broader issues in personality assessment. The first issue involves usin two or more data sources within a single personality study . What are the links among the various sources of personality data? The second issue involves the fallibility of personality measurement and how the use of multiple data sources can correct some of the problems associated with single data sources.
A key issue that personality psychologists must address is how closely the finding obtained from one data source correspond to findings from another data source. If for example, a person rates herself as dominant, do observers, such as her friends and spouse, also view her as dominant? Do findings obtained from mechanical recordin devices, such as an actometer , correspond to data obtained from observer reports or self-reports of activity level?
Depending on the personality variable under consideration, agreement across data sources tends to range from low to moderate. Ozer and Buss (1991) examined the relationships between self-report and spouse-report for eight dimensions of personality . They found that the degree of agreement varied depending on the particular trait and on the observability of the trait. Traits such as extraversion showed moderate agreement across data sources. The trait of "calculating," on the other hand, showed low self-spouse agreement. Traits that are easily observable (such as extraversion) show a higher degree of self-observer agreement than do traits (such as calculating) that are difficult to observe and require inferences about internal mental states
One of the central advantages of using multiple measures is that each measure has unique idiosyncrasies that have nothing to do with the underlying construct of interest. By using multiple measures from various data sources, researchers are able to average out these idiosyncrasies and home in on the key variable under study .
A major issue in evaluating linkages among the sources of personality data is whether the sources are viewed as alternative measures of the same construct or as assessments of dif ferent phenomena. A person self-reporting about her relative dominance, for example, has access to a wealth of information—namely , her interactions with dozens of other people in her social environment. Any particular observer—a close friend, for example—has access to only a limited and selective sample of relevant behavior. Thus, if the friend rates the woman as highly dominant, whereas the woman rates herself as only moderately dominant, the disagreement may be due entirely to the dif ferent behavioral samples on which each person is basing his or her ratings. Thus, lack of agreement does not necessarily signify an error of measurement (although it certainly might). It may instead signify that observers are basing their conclusions on dif ferent behavioral samples.
In summary, the interpretation of links among the sources of personality data depends heavily on the research question being posed. Strong agreement between two sources of data leads researchers to be confident that their alternative measures ar tapping into the same personality phenomenon, as proves to be the case with extraversion and activity level. Lack of strong agreement, on the other hand, may mean that the dif ferent data sources are assessing dif ferent phenomena, or it may indicate that one or more data sources are fallible or have problems—an issue to which we will now turn.
Each data source has its own problems and pitfalls that limit its utility . This is true of all methods in science. Even so-called objective scientific instruments, such as tel escopes, are less than perfect because minor flaws, such as a slight warping in th lens, may introduce errors into the observations. The fallible nature of scientific mea sures is no less true in personality research.
One powerful strategy of personality assessment, therefore, is to examine results that transcend data sources—a procedure sometimes referred to as triangulation. If a particular effect is found—for example, the influence of dominance on the assumptio of leadership—does the ef fect occur when dominance is measured with self-report as well as with observer -reports? If extraverts are more easily driven to boredom than are introverts, does this show up when boredom is assessed with physiological recording devices (e.g., brain waves suggesting the person is almost asleep) as well as via self-report?
Throughout this book, as we discuss the empirical findings that have accumu lated within each domain of personality , we will pay special attention to findings tha transcend the limitations of single-data-source assessment. If the same results are found with two or more data sources, then researchers can have greater confidence i the credibility of those findings
Was this article helpful?