The George Kelly Society
  The Psychology of Personal Constructs
George Kelly Personal Construct Psychology The Society Resources
Interest Groups Conferences

  Repertory Grid Methods
Developed within personal construct theory, repertory grid methods represent a widely used set of techniques for studying personal and interpersonal systems of meaning. Because of their flexibility, repertory grids (or repgrids) have been used in approximately 3,000 studies of a broad variety of topics, ranging from children’s understandings of physical science principles and consumer preferences, to formal structures of self-reflection within cognitive science and the mutual validation of belief systems between friends. However, their most consistent area of application has probably been in the clinical domain, where grids have been used to assess the properties of meaning systems of different groups of persons (e.g., those diagnosed as “thought disordered” or “agoraphobic”), and how these change over the course of treatment.

The “role construct repertory test” (or reptest) was initially designed by George Kelly, the author of personal construct psychology (PCP), as a means of assessing the content of an individual’s repertory of role constructs —the unique system of interconnected meanings that define his or her perceived relationships to others. In its simplest form, the reptest requires the respondent to compare and contrast successive sets of three significant people (e.g., my mother, my father, and myself), and formulate some important way in which two of the figures are alike, and different from the third . For example, if prompted with the above triad, a person might respond, “Well, my mother and I are very trusting of people, whereas my dad is always suspicious of their motives.” This basic dimension, trusting of people vs. suspicious of their motives, would then be considered one of the significant themes or constructs that the person uses to organize, interpret, and approach the social world, and to define his or her role in it.  By presenting the respondent with a large number of triads of varying elements (e.g., a previous romantic partner, best friend, a disliked person, one’s ideal self; for details see element selection), the reptest elicits a broad sampling of the personal constructs that constitute the person’s outlook on life and perceived alternatives. These constructs can then be interpreted impressionistically, used as the basis for further interviewing of the respondent, or categorized using any of a number of reliable systems of content analysis, conducted either manually or using available computer programs.

While the results of the reptest are often revealing, most contemporary users prefer to extend the method beyond the simple elicitation of constructs, by prompting the respondent subsequently to rate or rank each of the elements (e.g., people) on the resulting construct dimensions. For example, using the triadic comparison method described above, a respondent might generate a set of 15 constructs (e.g., trusting vs. suspicious; moved by feelings vs. rational; has ambition vs. no goals; young vs. old), which might be arrayed in 15 rows on a sheet of paper. She might then be asked to assign a number to each of 10 elements (e.g., my mother, father, self, partner) arranged in columns going across the sheet, representing where each figure would fall on, say, a 7-point scale anchored by the poles of each construct. For instance, “mother” might be seen as 1 on trusting vs. suspicious, representing very trusting, whereas “father” might be placed at 5 on this same scale, representing moderately suspicious. The intersection of the 15 construct rows with the 10 element columns forms the “grid,” and the matrix of 150 specific ratings it contains is amenable to a wide range of analyses. In practice, repertory grids can be virtually any size, from 6 constructs and elements to literally hundreds of each for a given respondent. However, most research indicates that the amount of new information about the person’s meaning system in a domain (e.g., perceptions of acquaintances) begins to peak once approximately 15 to 20 constructs and a similar number of elements have been sampled. Although the repgrid was originally devised as an interview-based or paper-and-pencil measure, most contemporary users rely on any of a number of computer programs for their elicitation and analysis, such as the popular WebGrid V program available via the Internet.

Although the specific element ratings on important constructs are often informative in themselves (e.g., seeing that a respondent views her father as 'suspicious' and 'having no goals', but also as 'rational'), it is typically more helpful to conduct a comprehensive analysis of the grid to discern larger patterns. This might involve correlating and factor analyzing the matrix of ratings to see at a glance which constructs “go together” for the respondent (e.g., everyone who is trusting may also be seen as moved by feelings), or to learn what people are most and least alike in the respondent’s view. These linkages among constructs often suggest why people remain “stuck” in symptomatic patterns, as when a client resists reconstruing himself as happy instead of depressed, because the former is associated with being superficial as opposed to deep. Similarly, patterns of identification among elements in a grid can be clinically informative, with some of these (e.g., degree of correlation between actual self and ideal self) providing useful indices of progress in psychotherapy.

An interesting feature of grid technique is that it combines aspects of both idiographic assessment, which strives to reveal unique dimensions of a given respondent’s outlook, and nomothetic research, which seeks general patterns across people.  Thus, the format of the repgrid essentially guides the respondent in constructing his or her own questionnaire (by eliciting the individual’s own constructs and relevant elements or figures to rate), while permitting comparisons across different people or groups.  For example, depressed individuals, relative to others, tend to show not only distinctive themes in the content of their constructs (e.g., more self-references and more morally evaluative themes), but also distinctive overall structure (e.g., “tighter" intercorrelations among constructs, and more “polarized” or extreme perceptions). This blend of projective and objective testing has made grid technique useful to both clinicians and scientists seeking to understand how different persons and groups organize their view of themselves and the world.

Describing a few of the problems to which repgrids have been applied gives some idea about the range and flexibility of the method. Grids have been used to study the long-term adjustment of survivors of incest, who carry with them a sense of distance from other people decades after the sexual abuse. They have also been used to measure processes of identification with other clients and therapists within group therapy settings, and to predict who is most likely to benefit from this form of treatment. Grids have been applied to the study of the development and breakdown of romantic relationships and friendships, by looking at the degree of convergence between partners in the way they construe experiences at increasingly intimate levels. Other investigators have relied on grids to understand the distinctive differences in the knowledge structures of experts and novices in a given domain, and to refine the discriminations made by assembly line workers in detecting product flaws.  As the use of these methods continues to grow with the dissemination of ever more powerful computerized systems for their elicitation and analysis, it seems likely that repertory grids will become an increasingly popular tool for both helping professionals and social scientists.

© Jörn Scheer  2016
Last update 15 July 2016