(c) 1992
Version 3.5 incorporates a number of small improvements over Version 3.0, released in 1990. There is more flexibility in the prompts employed for the first two phases of the procedure: Gathering Targets and Gathering Descriptors. Most important, it is now possible to interrupt the third phase of the procedure, Gathering Ratings, and restart the procedure where it left off at a later time. This will prove very helpful in the case of large targets x descriptors matrices.
PERSPACE, a computer-controlled procedure for assessing the individual's view of his or her social world, was developed for the Program on Conscious and Unconscious Mental Processes of the John D. and Catherine T. MacArthur Foundation, directed by Mardi J. Horowitz of the Department of Psychiatry, University of California, San Francisco, and located at the Langley Porter Psychiatric Clinic. PERSPACE was created and designed by John F. Kihlstrom of the Amnesia and Cognition Unit, Department of Psychology, University of Arizona, under a subcontract from the Program; programming was by Dave Olsen at the University of Arizona, and by Randolph L. Cunningham and Paul H. DuBois of the University of Wisconsin.
A major item on the agenda of those who seek to foster the integration of personality, social, and clinical psychology is the development of techniques that can be used with individual patients to assess their social schemata -- the organized mental structures, pertaining to self and others, that guide experience, thought, and action in an interpersonal context. To this end, a number of techniques have been developed for rating protocols derived from recorded psychotherapy sessions (e.g., Horowitz, 1990). At the same time, however, it seems important to remember that rating schemes are interpretive schemes, in which some piece of experience, thought, or action is given meaning by assigning it to one category or another. In the act of interpretation, the rater's own cognitive structures are brought into play, in order to make inferences about the target's real intentions, or the actual origins of his or her action. This situation, in turn, raises the very real risk that the judge's schemata will be confused with those of the subject and that whatever interpretive scheme is applied becomes less a tool for understanding the individual and more a Procrustean bed for slotting him or her into preconceived categories.
As is well known, this sort of criticism has long been directed at insight-oriented psychotherapy and the psychodynamic theories of personality on which it is based -- one never quite knows whether it is the patient's fantasies or the therapist's that are being discussed. But it is also a criticism that has been directed against traditional psychometric theories of personality, where instances of experience, thought, or action are classified into trait categories, or construed as reasons for assigning a person to one or another personality type (Mischel, 1968). Over the last 30 years, personologists and psychotherapists of a cognitive-behavioral persuasion have argued that subjects should be allowed to speak for themselves, and that traditional psychometric or psychodynamic approaches to personality assessment have no privileged status as routes to either conscious or unconscious mental processes (see Cantor & Kihlstrom, 1987).
In the hands of traditional behaviorists, this critique of interpretation meant that subjects should keep quiet too -- that only their overt behavior was of interest, and that it should only be tabulated and cross-tabulated with objectively recorded environmental variables (Skinner, 1953). But we are a long way from that time now. Beginning with Rotter (1954) and Kelly (1955), and continuing with Bandura (1986) and especially Mischel (1968), cognitively oriented personologists and psychotherapists have argued that understanding the meanings that subjects assign to environmental events, and their own actions, are critical for understanding personality. In terms of the social intelligence view of personality (Cantor & Kihlstrom, 1987), this means that social-cognitive structures and processes lie at the heart of individual differences in experience, thought, and action, whether these are adaptive or maladaptive. Personality theories of this sort require new assessment instruments that permit people to speak for themselves, going beyond mere tabulation but stopping short of schemes that impose the investigator's interpretations on the subject or patient (e.g., Kihlstrom & Nasby, 1981; Nasby & Kihlstrom, 1986).
Since the time of George Kelly, one of the most important ideas in cognitive approaches to personality and psychopathology has been the personal construct -- the person's idiosyncratic repertoire of concepts. Kelly developed an assessment instrument, the Role Construct Repertory Test, for assessing an individual's personal constructs. Briefly, the subject was asked to name persons who exemplified each of a broad range of social roles such as boss and rejecting person. Then three of these targets were sampled at a time, and the subject was asked to indicate some way in which two of these individuals were alike but different from the third. After eliciting a number of constructs in this manner, every target is rated on every construct. A similar procedure can be followed for social situations. By applying a number of intuitive and mathematical techniques, the investigator can determine the content of the constructs, the relations among them, and the complexity of the individual's personal construct system. Thus he or she is able to enter into the subjective social world of the subject -- begins to understand how the subject categorizes, compares, and contrasts those whom he or she encounters in the ordinary course of everyday living.
Kelly's proposal stimulated much interest, but application of his ideas was hampered by the fact that the appropriate mathematical techniques were cumbersome and expensive to apply (Bannister & Fransella, 1971; Cole & Landfield, 1977). The advent of computer technology has changed this situation somewhat, and a number of investigators have begun to finding ways of doing precisely on a computer the sorts of things that Kelly had to do roughly on paper. Among the most inventive of these investigators are Seymour Rosenberg (1977, 1986, 1988; Rosenberg & Gara, 1985; Rosenberg & Jones, 1972; Rosenberg & Sedlak, 1972a, 1972b) and Lawrence Pervin (1976, 1977, 1983), both at Rutgers.
Rosenberg's technique involves three phases. First, each subject listed at least 100 people from their lives. The subjects then prepare, for each target, a list of the physical and psychological traits thought to be characteristic of that person, and the feelings elicited in them by him or her. In addition, the subjects also freely describe three views of themselves: "me-now", me-past", and "me- ideal". A computer collates all of the person and attribute entries, and the subjects then rate each of the target persons, including the three "selves", on each of the trait and feeling attributes. The resulting two-way matrices, one for persons by traits and the other for persons by feelings, are analyzed by means of hierarchical clustering and multidimensional scaling -- multivariate techniques similar to factor analysis. Simplifying for purposes of exposition, the dimensions emerging from the scaling solution can be taken as analogous to factors, which are in turn defined by the clusters of related traits and feelings that load highly on them.
Rosenberg's analysis focused on the relations among the traits and feelings, expressed either in terms of co-occurrence or correlation. As might be expected from research on implicit personality theory (Schneider et al., 1979), certain traits and feelings were fairly consistently found to co-occur across the subjects. But other dimensions were quite idiosyncratic, appearing in one subject but not in the others. The differences between the subjects are, from a cognitive point of view, the differences between the individual's personalities. The entire set of trait co-occurrences shows, for each subject, how he or she perceives a major portion of the social world as organized, both in terms of the traits of other people and his or her own reactions to them.
Pervin (1976, 1977, 1983) employed a similar technique geared toward the assessment of situations rather than people. Subjects were first asked to list the important situations in their current lives. They were then asked to describe each situation, as well as how they felt, and their characteristic behavior, in each. The lists for each situation were collated and edited for redundancy, and then the subjects rated each situation on each attribute. The resulting matrix was factor analyzed for each subject separately, yielding the basic dimensions in the perception of interpersonal situations. Again, this procedure revealed the various dimensions that these individuals use to organize their perceptions of the situations in which they live their lives.
For the reasons outlined, above, it appears that the Rosenberg-Pervin adaptation of Kelly's Rep Test is a powerful technique of choice for tapping the content and organization of people's mental schemata for the social world. Its primary advantage is that it allows subjects to speak for themselves, without interpretation or inference on the part of the investigator. While engaged in the procedure, subjects achieve considerable insight about themselves in the course of completing the procedure. In the process of listing the people, situations, and events that are important to them, they are led to reflect on their lives in ways that may escape them in the ordinary course of everyday living. And although the descriptors provided in the subsequent feature-listing phase might (but not necessarily) come "off the top of one's head", the final target-by-feature rating phase really forces us to think about the entities in our interpersonal and intrapsychic worlds. If the mental representations elicited by this technique are not unconscious (in the strict sense of being in principle inaccessible to introspection; see Kihlstrom, 1984, 1987, 1989, 1990), then at least they may be preconscious. The simple fact that subjects are intrigued, and surprised, by what they see in the cluster analyses clearly indicate that the procedure is telling them -- and us -- something they didn't know about themselves.
We have developed an adaptation of the Rosenberg/Pervin technique for the purposes of assessing schemata for self and others in clinical settings. The adaptation differs from its original inspirations in two major ways. First, the general assessment strategy is applied to a wider variety of entities in interpersonal space. For example, pilot research in our laboratory is focused on the subject's self-concept, and is intended to indicate how context-specific self-concepts are organized in the mind of the person (see also Gara, 1985; Gara & Rosenberg, 1979; Rosenberg & Gara, 1985). Following the lead of Kelly and Mayman, the procedure can also be adapted to study the perceived relations among personal experiences, as represented (for example) in early memories of childhood. Second, and perhaps more important, it seems that clinical assessment should as interested in the relations among the entities as in the relations among their features. Therefore, we employ cluster analysis rather than factor analysis or multidimensional scaling to represent the perceived relations among self, others, social interactions, and the situations in which they take place.
The assessment procedure is implemented in PERSPACE, a general-purpose computer program for mapping interpersonal space (Kihlstrom & Cunningham, 1990). A preliminary version of the software, geared solely to the assessment of context-specific self-concepts (known as SITUATE, Version 1.10) was written in BASIC by Paul H. DuBois of the University of Wisconsin, to run under TRSDOS on the Radio Shack TRS80/IV microcomputer. A much-expanded program (PERSPACE, Version 2.20, 1987) was written in Turbo Pascal 3.01A by Randolph L. Cunningham, also at Wisconsin, for the IBM PC and compatibles. The most recent release of PERSPACE (Version 3.5, 1990), for the IBM PC, AT, PS/2, and compatibles, was programmed in Turbo Pascal 5.5 by Dave Olsen at the University of Arizona.
From the Subject's point of view, a complete PERSPACE assessment consists of three sessions.
The procedure begins by asking the subject to produce a list of targets for rating. McGuire (1984; McGuire & McGuire, 1988) has cogently argued that freely generated lists targets and descriptors reduce the potential for interference and distortion by concepts and expectancies imposed by the clinician or investigator, and thus come closest to revealing what is on the mind of the patient or subject (for a related argument, see Cantor & Kihlstrom, 1987; Kihlstrom & Cantor, 1984; Kihlstrom & Nasby, 1981; Nasby & Kihlstrom, 1986). However, there are circumstances in which specific probes might be necessary or desirable. For that reason, Version 3.5 provides both free-generation and probed-response options. The free-response probe can also be customized.
The program-supplied lists of Target probes are provided in Appendix 1. Future versions will include a wider list of probed-response targets and additional descriptor lists, as well as a custom facility.
After the targets have been listed, they are output one by one, in a random order, for feature listing. These individual attribute lists are then automatically edited for redundancy. That is, exact (case-sensitive) duplicates are removed and the resulting list is ordered alphabetically for ease of subsequent manual editing.
The target and descriptor lists are collated, and every pairwise combination of target and descriptor is presented to the subject for a final rating. Alternatively, Phase 2 can be skipped and each target can be rated on an investigator-supplied rating scale. The program-supplied rating scales are provided in Appendix 2. Future versions will include a wider selection of options.
There is also a utility for preparing a custom set of investigator-supplied rating scales. The numerical rating scales themselves range from two to nine points, with a variety of options (including a custom utility) available for defining endpoints and midpoints.
As documented below, this phase of PERSPACE can consume considerable subject time. Accordingly, Phase 3 can be interrupted, and restarted at a later time.
The compiled version of the Program runs under MS-DOS 2.0 or higher in 256K of RAM in systems with one 1.44 mb disk drive, or one diskette drive of any size and hard disk.
Copies of PERSPACE are available in 3-1/2" format from John F. Kihlstrom, Department of Psychology MC 1650, University of California, Berkeley, 3210 Tolman Hall, Berkeley, California 94720-1650.
PERSPACE is unsupported, but the source code is available. Users who make alterations to the source code are asked to send copies to both Kihlstrom and Horowitz.
The PERSPACE Software System will run on IBM PC compatible machines. It requires DOS 2.0 or higher and 110 KBytes of available RAM. To install the PERSPACE Software System on a hard disk, simply copy all of the files and subdirectories on the installation disk into the desired directory. It is suggested that a separate subdirectory off the root be created to hold the system files. If the version of DOS is 3.30 or higher, one can copy the entire system with just one command, "xcopy". To do this, place the floppy disk in drive A: (or B:) and type
"xcopy a:\perspace\*.* c:\perspace\*.* /s"
or
"xcopy b:\perspace c:\perspace /s".
The entire system will need approximately 220 KBytes, including the manual, plus 460 KBytes (maximum) for each subject who keeps his/her data on the hard disk.
To start up the PERSPACE System Software, first change into the newly created "PERSPACE" subdirectory
"cd \perspace".
Now simply type
"perspace"
and the system will start.
This page last revised 04/08/10 02:58:48 PM.