two. For the rest of this chapter we will illustrate some of the tools and
concepts we have been using to link consumers and foods.
REPERTORY GRID—A TOOL TO ELICIT CONSUMER
DESCRIPTORS AND PERCEPTIONS
The Repertory Grid Method (RGM) is another technique that has been
recently used to look at apples (Bhanji et al., 1997). The method was
originally developed by Kelly (1955) to identify the constructs that peo-
ple use to structure their perceptions of the social world. The basis for
this method is the elicitation of “constructs.” A construct can be defined
as a way in which two things are alike and, in the same way, different
from a third (Kelly, 1955). Applying the idea of the RGM to food ac-
ceptability was first proposed by Olsen (1981).
Objects are arranged into groups of three (triads) such that each ob-
ject appears in at least one triad and that one object from each triad is
carried over to the next triad. The subject is asked to state in which way
two objects are alike and different from a third, and descriptions for like-
ness and difference are recorded (Sampson, 1972). Once the subject has
exhausted all his/her constructs for a given combination, the two re-
maining combinations within the triad are similarly presented and con-
structs elicited in the same way. The constructs may then be associated
with 100 mm visual line scales or a box scale anchored with the ex-
tremes of the construct as described by the subject (Raats, 1992). The
method has been applied in a number of food-related studies (Thomson
and McEwan, 1988; and Scriven et a1., 1989).
The data from the RGM can be analyzed by Generalized Procrustes
Analysis. Generalized Procrustes Analysis (GPA) is an empirical mul-
tivariate method that can be used to explore the relationship between
samples and attributes on a multivariate space (McEwan, 1989). When
applying GPA to data, each configuration is a matrix of scores from one
individual who has assessed a number of samples for a given set of at-
tributes or constructs. Samples must be identical for each of the sub-
jects. GPA allows each individual to have a unique set of attributes or
constructs (Raats, 1992).
The data transformation involves three steps, which ensure that the
results can be meaningfully compared across subjects: translation, rota-
tion/reflection and scaling. Thus, GPA takes into account three sources