Advances in Personality Assessment

Volume 10

James N. Butcher, Charles D. Spielberger

• Publisher: Psychology Press
• Year: 2014
• Print ISBN:
  9780805818048
• eBook ISBN:
  9781317779896
• Topic: Psychology
• Subtopic: History & Theory in Psychology
• Index: Psychology

The Advances in Personality Assessment Series began in the early 1980s to facilitate the rapid dissemination of important new developments in theory and research on all aspects of personality assessment. Impressed with the extensive research on test development and validation that was going on at that time, the editors were concerned with the limited publication resources devoted to personality assessment. With this series, they hoped to provide a publication opportunity and resource for reports of personality assessment research and/or clinical practice that might not conveniently fit in journal format because of length, focus, or content.

The first nine volumes have accomplished this goal exceptionally well by highlighting new empirical and theoretical developments, providing descriptions of new scale development, and in publishing timely reviews of important research. Volume 10 -- the last in the series -- continues in the same tradition as the previous volumes, with chapters devoted to scale construction, theoretical interpretation, and empirical analysis. The editors conclude the series knowing that an important void has been filled. They close with a feeling of both accomplishment and a slight sense of regret now that their efforts for more than a decade are at an end, as well as assurance that the torch has been passed on to others.

1	An Idiographic and Nomothetic Study of Personality Description

Ruth Schiller
Auke Tellegen
Jean Evens
University of Minnesota

For the past several decades, trait psychologists have used lexical personality descriptors to investigate the structure of personality. A great deal of this research has converged on a dimensional model of personality known as the Big Five (Digman, 1990; Goldberg, 1990, 1993; John, 1990). Although the five-factor model is not without its critics, consensus regarding the Big Five has evolved to where some researchers advance the model as fundamentally correct and propose moving the inquiry forward to its implications and applications (e.g., McCrae & John, 1992).

The widespread acceptance of five basic personality dimensions is the result of a long process of rational and empirical distillation. Initially, Cattell (1957) performed a reduction of Allport and Odbert’s (1936) well-known descriptor list that resulted in 12 oblique dimensions. Others have not been able to replicate Cattell’s factors. His methodology required visual rather than objective rotations, introducing an element of subjectivity, and his results reflected errors in computation (Digman & Takemoto-Chock, 1981). John (1990) observed that it is difficult to avoid the conclusion that Cattell’s variables and factors seem to represent the traits that Cattell considered most important. Subsequent studies by Tupes and Christal (1961), Norman (1963), and Borgatta (1964) all yielded five major factors. However, because these studies drew on Cattell’s reduced set of 35 descriptors, it was not clear whether the five factors were an accurate representation of lexical personality description. Subsequently, Norman (1967) provided an updated descriptor list, from which Goldberg (e.g., 1992) selected new sets of descriptors. In other contemporary studies (e.g., Digman & Inouye, 1986; McCrae & Costa 1985, 1987), the descriptor sets were likewise selected independently of Cattell’s list. These later studies have also replicated the five-factor structure.

Despite these developments, the issue of the representativeness of the Big Five has recently been raised again. Do these dimensions indeed capture natural-language personality characterization? Tellegen (1993) questioned the elimination of so-called “evaluative” and “state” terms from the descriptor sets used in previous studies. He maintained that the elimination of such terms may have resulted in an incomplete variable set (i.e., a set with fewer or different personality dimensions than actually represent natural-language personality description). Tellegen and Waller (in press) conducted a study using what they considered a truly representative set of lexical personality descriptors, including evaluative and emotional terms (the latter to be used as emotional trait descriptors) from the American Heritage Dictionary. They recovered seven major factors, including two large evaluative or valence dimensions: Positive Evaluation and Negative Evaluation. Positive Evaluation appears to be a socially oriented dimension ranging from excellent (dazzling, important) to nothing special. Negative evaluation appears to be a character-oriented dimension ranging from awful (evil, destructive, deceitful, immoral) to decent. Even with a new set of descriptors, four of the other five dimensions clearly resemble the Big Five. For the fifth factor, Tellegen and Waller offered Conventionality (vs. Unconventionality) as a reversed alternative to Digman’s (1990) and Goldberg’s (1990) interpretation of the factor as Intellect, and to McCrae and Costa’s (1987) interpretation of it as Openness to Experience.

In another study, descriptors were generated by the subjects rather than being selected from a dictionary. Chaplin and John (unpublished data, cited in John, 1990) asked 300 college students to describe desirable and undesirable characteristics of their own personalities. The 60 most frequent terms generated by these students were further analyzed in a separate sample of subjects, and five factors closely resembling the Big Five were recovered. Chaplin and John’s use of the descriptors generated by the individuals was an important advance in the study of natural-language personality description. They assembled their subjects’ descriptors into a single questionnaire, and then had a new group of subjects complete the questionnaire to derive the usual matrix of R-correlations characteristic of a nomothetic design. The dimensions underlying the natural-language descriptions of self and others, such as those assembled in Chaplin and John’s study, may “represent shared personal constructs of personality … person-perceptual schemata that are attuned to those available cues … of the social landscape that are relevant to the perceiver’s most important everyday social needs” (Tellegen, 1993, p. 126). If these interindividual studies are tapping into a person’s perception of personality, and if the Big Five or Big Seven dimensions recovered are a representation of how people actually think about personality, then these same dimensions should be recoverable from the personality descriptions generated by a single person.

To test this hypothesis, it would be necessary to examine the way in which each individual uses his or her terms to describe other people (i.e., to employ an idiographic design). An idiographic study would have each subject generate his or her own descriptors, as well as use these descriptors to describe a large number of known others. Separate analyses of P-correlations could then be employed for each dataset. Only an idiographic design can actually provide access to an individual’s dimensional personality constructs. Zevon and Tellegen (1982) pointed out that idiographic studies are an important complement to nomothetic designs, and that the use of multiple P-data studies can lead to nomothetic inferences. The present investigation is based on the assumption that the generation and use of descriptors by single subjects is important, and even essential, for identifying the actual dimensional structure of people’s everyday personality descriptions. Embedded in these idiographic structures are the underpinnings of the nomothetic folk concepts of personality (Tellegen, 1993).

The aim of this study is to explore how people, using their own words, describe other people. Will it be possible to integrate a series of idiographic studies and make nomothetic sense of them? Will five major dimensions similar to the Big Five consistently emerge in and across individual studies? Will evaluative dimensions also emerge? To address these questions, 12 subjects were asked to generate their own personality descriptors and to use these descriptors in rating 100 acquaintances. Separate P-analyses were conducted on each of the 12 datasets, and 23 judges were recruited to rate the content of the factors from all 12 sets. The rating task was designed to reveal whether naive judges could identify, with any degree of regularity, representatives of the Big Five or Big Seven among the idiographic factors. If they did, then, at last, one could conclude that the Big Five capture important dimensional features of how individuals actually characterize personality differences.

METHOD AND RESULTS

The 12 Idiographic Studies

Subjects. Twelve subjects participated in this phase, including eight female University of Minnesota undergraduate psychology students recruited from a personality class and four friends (two female) of the first author. The participating psychology students received extra credit toward their course grade (all subjects also received a book at the conclusion of the study, but were not told about this ahead of time).

Procedure. The participants were given a brief overview of the study and the tasks involved. Then each subject individually completed the study tasks at home in three steps.

First, each subject made a list of 100 real-life acquaintances, identifying them by first name and initial only of last name. Acquaintances were individuals whom the subject knew well enough personally to have some opinions about (thus excluding public and mass media figures). The investigator handed each subject several sheets of paper containing 100 numbered blocks, one block per acquaintance, and asked her or him to generate as many descriptive terms or short phrases as were needed to describe each acquaintance. Subjects were told that they could, but need not, repeat previously used terms to describe subsequent persons on their list. If a descriptive term was very unclear or complex, the investigator recommended that the subject’s think of simpler terms to describe the concept. Each subject generated 150-300 descriptors for her or his 100 acquaintances. After every subject returned the completed list of acquaintances and descriptors to the investigator, she transferred each distinct descriptor generated by each subject to a 3 × 5 card. This resulted in several stacks of cards, one per subject, with each stack containing only those descriptors generated by a given subject.

Second, each subject was given her or his stack of descriptor cards and was instructed to sort the cards into groups of highly similar descriptors that could be considered interchangeable. The investigator explained that the groupings were necessary to make the task in the last step manageable. Once the groups were formed, she asked the subjects to select one to three descriptors from each group that most accurately summarized the meaning of that group. She then entered the list of these summary descriptors and the subject’s list of 100 acquaintances in the rows and columns of a large matrix, respectively. Care was taken to place somewhat dissimilar descriptors adjacent to each other, and to intermix positive and negative descriptors.

Third, the investigator instructed the subjects to use the matrix of summary descriptors and acquaintances to rate each acquaintance on each descriptor using the following 4-point rating scale: (1) very inaccurate or very uncharacteristic, (2) slightly inaccurate or slightly uncharacteristic, (3) slightly accurate or slightly characteristic, and (4) very accurate or very characteristic. Each subject was given 2 weeks to complete this lengthy task, and each was encouraged to rate only a small number of acquaintances in any one session.

Factor Analyses. Exploratory factor analyses were carried out on the 12 idiographic datasets. Principal components were extracted from each matrix of P-correlations among descriptors, followed by normalized varimax rotations. Initially, four rotated solutions were obtained for each dataset: a solution based on the Kaiser criterion (rotation of all components with eigenvalues greater than or equal to one); a five-factor solution; a seven-factor solution (the former two were selected on the basis of previous research); and an (n + 5)-factor solution, where n equals the number of factors chosen according to the Kaiser criterion (this rotation ensured the extraction of all factors of any substance).

On the basis of inspections of the four solutions for each dataset, the authors adopted the Kaiser solution as the most satisfactory across all 12 datasets. The five- and seven-factor soluti...

Table of contents

• Cover
• Title Page
• Copyright Page
• Table of Contents
• Preface
• 1. An Idiographic and Nomothetic Study of Personality Description
• 2. Development of an MMPI–2 Scale to Assess the Presentation of Self in a Superlative Manner: The S Scale
• 3. Measuring Alexithymia: Reliability, Validity, and Prevalence
• 4. Relations Between Mood and Personality: Findings From the Israeli Mood Studies
• 5. A Comparison of the Benefits of Two Therapeutic Community Treatment Regimens for Inner-City Substance Abusers
• 6. MMPI–2 Measures of Substance Abuse
• 7. Measuring the Prosocial Personality
• 8. Rorschach Susceptibility to Malingered Depressive Disorders in Adult Females
• Author Index
• Subject Index