Main Page
Alphabetical Index

Hints for prints


Element selection
Elements form a crucial component of all repertory grid applications. They are central to the construing process of construct elicitation. The elements used in grid applications (albeit supplied or elicited) determine the type of constructs you will elicit from your interview respondents. Because elements are so central to grid work, much care must be taken in identifying / eliciting them.
Characteristics of elements
  • Homogeneous – elements must all be of the same type (or homogeneous). That is, they must either be all people, all objects, all situations, all experiences, or all events…  Ensuring the elements are of the same group makes the elicitation of personal constructs much smoother and simpler. Some examples of elements used in past grid studies have looked at different types of managers, leaders, job applicants, work activities, work roles, work teams, relationships, events in organizational life, information systems, organizations, products, bank loan applications to different countries, inter alia. However, if your purpose is to elicit more complex cognitions about the way people construe, interpret and make judgments about their experiences that may be based on people, objects, situations and events, you may consider re-wording your elements into "doing-words" so that they all become verbs (and hence homogeneous). Such extensions in the way elements are worded can widen the powers of the grid.
  • Discrete and representative – elements should provide a reasonable coverage of most aspects of whatever is being investigated. They should literally represent the four corners of your subject matter: the key things that make up your area of study. They must be discrete in order to elicit a wide range of constructs about a chosen topic. It is not recommended to use elements that are sub-sets of another, as this will make construct elicitation problematic. Moreover, if your elements can be converted into opposites, you most likely have parts of constructs instead of discrete elements in their own right.
  • As short as possible – elements must be specific and easily understood by the respondent. In this respect, about nine elements is an adequate number for most applications. Past applications have used between 7-25 elements. In the end, it all depends on how much time you and your grid respondent is prepared to sit down and talk about the topic under investigation. Usually, a typically grid with 9 elements can take anywhere between 1-2 hours, solid! More elements will take more time. Nine elements is usually recommended (though not essential) because it also allows you to make a decent enough coverage of your area of investigation; And it is also practically convenient as you can use the typical triadic comparisons of each element twice (e.g.: 123, 456, 789, 147, 258, 369).
  • Previously experienced – When choosing elements, they must be well known to the person to whom the grid is being administered. A general rule is for the respondent to have had actual experience (current or recent experience) with each of the elements so that the personal constructs generated from the grid interview are relevant and meaningful.
Supply or elicit elements?

Naturally, in the true spirit and intent of Kelly’s Personal Construct Psychology, eliciting both elements and constructs, and getting the respondents to rate their own elements based on their own constructs make the technique free from observer bias. Depending on how much researcher intrusion you are prepared to live with in your grid application, grid designers / interviewers have the choice of supplying the elements or eliciting them from respondents. In essence there area three (3) ways to do this:
  1. Interviewer can supply them
  2. Ask the grid interviewee to generate a list of elements based on the focus of the study, or
  3. Elicit the elements from the interviewee based on a set of questions (usually in matching pairs), the answers to which will be the elements
  • Supply elements: supplying elements give you the advantage of choosing those items you want the interviewee to consider because these are the elements that you are interested in finding out how people make sense of them based on their actual experience. The disadvantage however is that there is a chance your respondent has no or very little experience with them which could generate constructs that are not meaningful to the respondent. In this respect, you could do a bit of research of your own (say a pilot test) that helps identify elements that your intended respondents are bound to have had experience with. And it would be wise to ensure that your interviewee has experience with the supplied elements prior to the interview taking place.
  • Ask grid interviewee to generate a list of elements by free-response: This approach lets the respondent nominate what is considered the key elements that make up the subject matter you want to investigate. This can be very revealing as new insights can be gained by allowing the respondent to define his / her own parameters (given the focus of study). However, this approach has its own disadvantage in that the grid respondent may mention items / elements that are most familiar to them or deliberately disregard items that they do not like or feel are not important. If you are comfortably with living with this type of bias in the respondent’s favor then this may be the element choice strategy for you.
  • Elicit the elements from the interviewee based on a set of questions: This way of eliciting elements from grid respondents is similar to how Kelly elicited his elements using "role titles". If for example you want to investigate how subordinates perceive their working relationship with past and current supervisors, you may ask them to name: a supervisor you liked; a supervisor you didn’t like; an effective supervisor; an ineffective supervisor; a supervisor similar to you; a supervisor dissimilar to you; your current supervisor etc… (One caution is that you want to ensure that the interviewee gives you a different name / element for each question you ask). And if you are dealing with people (which can be quite personal and sensitive) you may like to suggest that the respondent use initials or only first names to identify them for themselves without revealing the true identity of the person to a third party.
Both the last two strategies for eliciting elements are more consistent with Kelly’s intentions of keeping the grid as much free from observer bias as possible. In the final analysis, you need to choose the approach that best suits the purpose of your study in eliciting the grid. As a case in point, if one of your goals is to compare between elicited grids, you will need to keep the elements (or constructs) constant. 


  • Bell, R.C , Vince, J., Costigan, J. (2002). Which vary more in repertory grid data: Construct or elements? Journal of Constructivist Psychology, 15, 305-315.
  • Easterby-Smith, M. (1980). The design, analysis and interpretation of repertory grids. International Journal of Man-Machine Studies, 13, 3-24.
  • Easterby-Smith, M., Thorpe, R., Holman, D. (1996). Using repertory grids in management. Journal of European Industrial Training, 20, 3-30.
  • Fransella, F, Bannister, D. (1977). A Manual of repertory grid technique. United Kingdom: Academic Press.
  • Fransella, F. (Ed.). (2003). International handbook of personal construct psychology. England: John Wiley.
  • Jankowicz, A. D. (1990). Applications of personal construct psychology in business practice. in. R. A. Neimeyer, G. Neimeyer (Eds.) Advances in Personal Construct Psychology 1: 257-287.
  • Jankowicz, A. D. (2003). The easy guide to the repertory grid. Chichester: Wiley.
  • Kelly G. 1955. The psychology of personal constructs. New York: Norton.
  • Stewart V, Stewart A. (1981). Business applications of repertory grid. England: McGraw-Hill.
  • Wright, R.P., Lam, S. S.K. (2002). Comparing apples with apples: The importance of element wording in grid applications Journal of Constructivist Psychology, 15, 109-119.

Robert P. Wright

Establ. 2003
Last update: 15 February 2004