At times, the interviewee is unable to answer a question or suggestion. To ease the questioning process some authors suggest that the client:
"...be encouraged to say whatever comes into his mind and not be concerned if he gives repetitive responses later in the procedure. When the client feels he should react more rapidly or finds it difficult to think of the most appropiate words or expressions, he is reassured by being told that the task is a new experience and most people have this difficulty (which is true)" (Landfield, 1971, p. 135).
Fransella and Bannister (1977) provide the following suggestions:
These suggestions can be very useful, but we must always remember to elicit the subject's constructs and not our own. Minimizing the interviewer's influence on a client's responses is of utmost importance.
When this line of questioning is followed, many respondents begin to show fatigue and become "saturated." They may find it progressively more difficult to develop new constructs or may start producing superficial responses. If enough elements have been presented by this stage, administration can be terminated and the next stage initiated. However, some subjects, because of their cognitive or verbal richness, provide several constructs for each comparison and do not seem to reach this saturation point. When this happens, the interviewer feels the process could continue almost indefinitely with the end result being too many constructs. This raises the issue of whether a maximum number of constructs should be established. Although our protocol allows for a maximum of twenty constructs, a few more could be added. Analysing grids of more than twenty-five constructs, however, is very complex and exceeds the parameters of some computer programs (see GRIDCOR's manual for limitations).
The interviewer may also need to consider excluding certain constructs which are not significant to the purpose of the study. Similar hair colour, for example, is usually not relevant in a clinical context and does not aid in further identifying personal meanings. Hunt (1951), in his doctoral thesis supervised by Kelly, created a classification system for those constructs which are of little use:
Considering these exclusion criteria, it can be said that clinically valid constructs are generally those that discriminate adequately between elements and describe personality characteristics.
In Daniel's case, we selected two elements, for example, "Mother" and "Daniel" and asked him, "Do you see Mother and Daniel as more similar or more different?" Daniel replied, "More similar." We then asked, "In what important way are these elements similar to you?" or, rather, "What characteristic would make these two elements alike?" Daniel answered saying that both were "anxious." We then asked for the characteristic that he considered to be opposite of anxious which elicited the construct "anxious-relaxed." This construct was written down on the first row of the protocol sheet. The first pole was placed to the left and the second pole to the right. We then asked, "In what way are your mother and father different?" Daniel answered, "My father tends to be irresponsible and my mother is more committed." In this way, the second construct "irresponsible-committed" was obtained. This procedure was then repeated with the dyad SELF/FATHER. The sequence of dyads presented was the following:
By the time all these dyads had been presented, Daniel had exhausted his capacity to produce any more relevant constructs. During the last few comparisons, the constructs had been repeating themselves indicating that his saturation point was reached, so we ended this stage of the process.