A comparison of the implication and repertory grid techniques

Abstract

The 'repertory grid ' and the 'implication grid' techniques are compared using criteria based on an indirect measure of construct matching scores. The implication grid proved superior to the repertory grid under conditions designed to compare stability of matching scores and under conditions designed to compare sensitivity to subjects' changes in construing. The significance of a further result, that the implication grid was superior for reflecting construct bipolarity, is considered in the light of conceptual problems with an assumption of bipolarity. The set of results is especially noteworthy since 'indirect' measures were formerlp the exclusive province of the repertory grid. The discussion raises general issues about the relative utility of two grid formats, and the unique measurement possibilities of the implication grid are illustrated. It is concluded that the. implication grid deserves careful consideration for use in studies concerned only with construct relationships.Kelly (1955) developed his 'repertory grid' method in the context of his work as a psychotherapist. The repertory grid (RG) provided a systematic framework for a client to describe his ways of discriminating (the constructs in a grid) between people important to him (the elements in a grid). Kelly assumed that each construct applied to a limited number of people and that each was bipolar. The most important RG measure was the degree of association between construct pairs. For example, the construct 'aggressive-quiet ' would have a high degree of association (high matching score) with the construct 'tense-relaxed ' if the people construed as 'aggressive ' were usually construed as 'tense ' and the people construed as 'quiet' were usually construed as 'relaxed'. The RG has now been applied to a variety of problems, some far removed from psychotherapy, but the emphasis on computing construct-matching scores remains.Working within the framework of Kelly's 'Personal Construct Theory ', Hinkle (1965) developed the 'implication grid ' (IG) for examining the relations between constructs. Subjects answered questions of the form, 'If you changed from being aggressive to quiet, what other constructs would have cause to be changed (e.g. tense to relaxed, ambitious to contented) by a change in yourself on this one construct alone'. This process was repeated so that each construct was paired twice with each of the others in order to generate a square matrix of interconstruct implications. Hinkle's IG has unique measurement advantages over the RG (Bannister & Mair, 1968, ch. 3), but Ryle (1974) notes that it is a complex task for subjects, and is one that has attracted little research.Fransella (1972) goes some way to meeting these criticisms by developing a modified form of the IG in the context of her construct theory approach to the treatment of stutterers. Fransella employed a card-sorting task in which each construct pole was considzred separately. Subjects were presented with one card at a time (e.g. 'aggressive') and told that all they know about...

The Reliability and Validity of Structural Measures Derived From Repertory Grids

“…For example in one study (Bannister 1962) the re-test correlation was 0.35. Exactly the same correlation was reported recently by Honess (1978) for Intensity in a rank order grid. Honess' subjects were children and the test re-test interval was four weeks" (p. 84).…”

Theory-Appropriate Analysis of Repertory Grid Data

“…Consequently, in a repertory grid that shows the constructs relate to elements, there must also be an underlying organization among constructs. Honess (1978) has shown imp grids to be superior to rep grids with respect to stability and bipolarity, although ten Kate (1981) has questioned the logical basis of bipolarity in imp grids, and stability could well be due to the direct nature of the imp grid procedure. As Downloaded by [Simon Fraser University] Bannister and Mair (1968), and Landfield and Epting (1987) have pointed out, the implication grid procedure requires the client to have insight into the relationships among constructs, whereas the repertory grid does not.…”

Analyzing Implications Grids Using Hierarchical Models

“…Included in this category are the techniques proposed by Fransella (1972) and Honess (1978Honess ( , 1982. Fransella's (1972) approach was to calculate the binomial probabilities of matching and mismatching ticks (entries equal to one) between pairs of rows and columns.…”

“…For computational details the reader is directed to Fransella (1972). Honess (1978) also explored interconstruct relations by constructing dissimilarity scores. In contrast to Fransella's technique, Honess' approach is applicable to both binary and fuzzy-valued data.…”