Communications of the Association for Information Systems

Volume 23 Article 3

7-2008

An Overview and Tutorial of the Repertory Grid

Technique in Information Systems Research
Aaron M. Curtis
Indiana University, aacurtis@indiana.edu

Taylor M. Wells
Indiana University

Paul B. Lowry
Brigham Young University

Trevor Higbee
ExxonMobile

Follow this and additional works at: https://aisel.aisnet.org/cais

Recommended Citation
Curtis, Aaron M.; Wells, Taylor M.; Lowry, Paul B.; and Higbee, Trevor (2008) "An Overview and Tutorial of the Repertory Grid
Technique in Information Systems Research," Communications of the Association for Information Systems: Vol. 23 , Article 3.
DOI: 10.17705/1CAIS.02303
Available at: https://aisel.aisnet.org/cais/vol23/iss1/3

This material is brought to you by the AIS Journals at AIS Electronic Library (AISeL). It has been accepted for inclusion in Communications of the
Association for Information Systems by an authorized administrator of AIS Electronic Library (AISeL). For more information, please contact
elibrary@aisnet.org.
An Overview and Tutorial of the Repertory Grid Technique in Information
Systems Research

Aaron Mosiah Curtis

Indiana University
aacurtis@indiana.edu

Taylor Michael Wells

Indiana University

Paul Benjamin Lowry

Brigham Young University

Trevor Higbee

ExxonMobil

Interest in the repertory grid technique has been growing in the IS field. This article seeks to inform the reader on the
proper use and application of the technique in IS research. The methodology has unique advantages that make it
suitable for many research settings. In this tutorial, we describe the technique, its theoretical underpinnings, and
how it may be used by IS researchers. We conclude by detailing many IS research opportunities that exist in respect
to the repertory grid technique.

Keywords: repertory grid, repgrid, personal construct theory, cognitive mapping

Volume 23, Article 3, pp. 37-62, July 2008

Volume 23 Article 3
 An Overview and Tutorial of the Repertory Grid Technique in Information
Systems Research

I. INTRODUCTION
The Repertory Grid Technique (RGT) is a cognitive mapping approach [Walsh 1995] useful for understanding how
individuals and groups derive meaning from the artifacts and individuals in their environment. The RGT is based on
Personal Construct Theory (PCT) [Kelly 1955], a constructivist theory that contends that people experience,
organize, and describe their environment in terms of cognitive personal constructs that can be distilled into bipolar
verbal labels—labels such as helpful-hurtful, emotional-analytical, or trustworthy-unreliable. Complementary to PCT,
Kelly [1955] introduced the RGT for eliciting these personal construct systems from individuals. The RGT was
initially developed for use in a clinical setting to help diagnose psychiatric conditions and more successfully
prescribe treatments [McKnight 2000]. Understanding the meanings embedded within a patient’s personal construct
systems helped clinical patients in their recovery [Bannister 1965].

The RGT has evolved from the original clinical therapeutic methodology for interviewing patients to a set of general
guidelines used in a wide variety of application domains, including environmental studies [Dinsdale and Fenton
2006], education [Kreber et al. 2003; Pill 2005], healthcare [MacCormick et al. 2004; Mickan and Rodger 2005], and
business [Ginsberg 1989; Stewart et al. 1981]. Although the RGT has been utilized on a limited basis in research
published in recognized IS journals, its use remains limited. Recent research indicates a need for increased
exploration of the cognitive aspects of IS-related issues [Heninger et al. 2006; Orlikowski and Gash 1994; Tan and
Hunter 2002], an exploration that, we believe, can benefit from using the RGT. The primary aims of this paper are
(1) to provide a comprehensive tutorial and review of research that utilizes this technique in order to promote
enhanced cognitive IS research; (2) to aid researchers who may be new to this technique; and (3) to provide ideas
of future research streams that can most benefit from the cognitive mapping in general and the RGT specifically.

To accomplish these objectives, we build on existing descriptions of the RGT in IS research [e.g., Tan and Hunter
2002] and other extensive reviews of the RGT [Bell et al. 2002; Caputi and Reddy 1999; Fransella et al. 2004;
Hagans et al. 2000] by exploring the wide spectrum of variations that compose the RGT [Rugg et al., 2000] and by
presenting a flexible and modular framework to guide researchers employing the RGT in IS research.

The RGT is one of many cognitive mapping techniques, each of which has its own advantages and disadvantages
[Brown 1992; Byrd et al. 1992]. While no single technique is best for all types of cognitive mapping [Rugg et al.
2002; Walsh 1995], understanding the opportunities and challenges of the RGT can help researchers use the RGT
in conjunction with other research methods [Liou 1990]. The RGT differs from other cognitive mapping techniques in
that it is highly personal, easy to administer, structured, and facilitates qualitative and quantitative analysis. Some of
the key challenges in utilizing the RGT are that it focuses on differentiation in personal construct systems; it
sometimes creates monotony; it is often cognitively demanding; it requires a lot of time; it is limited to the specified
scope; and it describes rather than prescribes. These pros and cons are elaborated in the next section.

Advantages of the RGT

Highly Personal
Researchers using the RGT discover how participants construct their world, whereas many other survey instruments
seek only to confirm what the researcher understands to be the domain of interest. Because the personal constructs
in the repgrid (created through the RGT) are often elicited directly from research participants, the technique can
reduce researcher bias and is more transparent to interviewees than other techniques [Boyle 2005; Brown 1992;
Crudge and Johnson 2004; Hunter 1997; Hunter and Beck 2000]. When eliciting personal constructs, an interviewer
may focus the RGT questioning on concrete examples from a participant’s own experience, which makes the results
more meaningful to participants [Langan-Fox and Tan 1997].

Easy to Administer
The RGT’s highly personal nature makes it relatively easy to administer in practice [Hudlicka 1996]. Trained
interviewers can use the RGT to gather data in many different contexts. Many software packages exist to facilitate
RGT interviewing and analysis, but the RGT can be just as effectively conducted with simply a trained interviewer,
an informed participant, and a pencil and paper.

Volume 23 Article 3
38
Structured
Data contained in a repgrid are highly structured, and multiple repgrids can be gathered, analyzed, combined, and
compared [Xiangmin and Chignell 2001] using Web-based or stand-alone elicitation and analysis tools [Shaw and
Gaines 1996; Verlinden and Coenders 2000]. Researchers can create maps or diagrams from this structured
analysis in a simple and straightforward manner [Verlinden and Coenders 2000].

Facilitates Qualitative and Quantitative Analysis

The personal and structured nature of the RGT makes it amenable to both qualitative and quantitative analysis
[Brown 1992; Hunter and Beck 2000] and allows the researcher to draw on the strengths of each type of analysis.
The data gathered in a repgrid are personal and flexible enough to provide a rich view of a single participant’s
personal construct system relating to a specific topic, and are structured enough to enhance the comparability and,
consequently, the validity of analysis [Langan-Fox and Tan 1997]. The data can be used for quick exploratory
analysis—because some results can be obtained without detailed analysis [Hudlicka 1996]—or analyzed further
using advanced coding or statistical techniques.

Disadvantages of the RGT

Focuses on Differentiation in Personal Construct Systems

Due to the way that constructs are elicited during the RGT interview, traits of the knowledge domain that are
important but not differentiated may be overlooked [Hassenzahl and Wessler 2000].

Sometimes Creates Monotony

For some participants, the RGT construct-elicitation process can be monotonous and lead to higher participant drop-
out rates than other techniques [Brown 1992]; however, other participants find the process intriguing [Moynihan
1996] or even fun and challenging [Oppenheim et al. 2003]. The level of participant boredom may be dependent on
the participant’s level of understanding. Davis [1982] found that for experienced participants whose expertise is
housed in long-term memory, indirect methods like the RGT may become frustrating and direct questions may be
more effective.

Can be Cognitively Demanding

The RGT can be a cognitively taxing methodology for participants because they must simultaneously consider
repgrid elements, personal constructs, and their individual ratings. This may be exacerbated by a complex topic
domain or one with which participants are inexperienced.

Consumes a Lot of Time

The entire process of the RGT can be time consuming, with some researchers reporting that 90 minutes were
required to create a 9 x 9-inch repgrid [Whyte and Bytheway 1996]. Typical RGT interviews last one to two hours,
though other interviewing techniques often take longer, as reported earlier. However, depending on the goal of the
research, the researcher may reduce the time required for the interview by supplying the elements (objects the
participant is assessing) and/or personal constructs. Furthermore, a researcher can combine this technique with
other, less time-intensive techniques to reduce the total time requirements.

Limited to the Specified Scope

As personal constructs have a finite range of convenience, the breadth of information elicited in a single repertory
grid is narrow by nature. Additionally, personal constructs are dependent upon the scope of the elements used in the
construct-elicitation process. Too much similarity in elements may lead to constructs that are of little use.
Additionally, if theoretically interesting tangential topics arise during an interview, a researcher can note them and
determine if they are appropriate to include in the repgrid being built. A separate repgrid may be required to examine
different topics.

Describes not Prescribes

The RGT facilitates exploration of a knowledge domain as it currently is; thus, it is not the most effective tool for
generating alternatives or for capturing participants’ expectations and future needs [Verlinden and Coenders 2000].
Placing an imaginary “ideal” element is a technique that can reduce this limitation, as discussed later in this paper.

Given these characteristics of the RGT, we argue that the advantages outweigh the disadvantages in many
methodological contexts and that the technique may be fruitfully applied to many problems in IS research—
especially if a researcher actively counters some of the disadvantages of the technique. The remainder of the paper
is organized as follows. Section II begins by discussing PCT—the theoretical foundation of the RGT. Section III

Volume 23 Article 3
39
 explains how a repgrid is created and analyzed. Section IV reviews the use of the RGT in IS research and suggests
areas for additional research.

II. THEORETICAL FOUNDATIONS: PERSONAL CONSTRUCT THEORY

The RGT originated in clinical psychology as a tool to elicit the personal constructs that are at the heart of Kelly’s
PCT. Kelly describes each person as an “incipient scientist” [Kelly 1955, p. 12] who creates a subjective model of
the world—a construct system—from an objective reality and then tests and modifies that model based on daily
experience. This model is made entirely of interrelated yet distinct bipolar constructs (i.e., characteristics) that allow
a person to compare and contrast distinct objects.

A personal construct system is a unique hierarchical configuration of constructs that guides a person’s behavior
[Kelly 1955; Kelly 1963]. Even when the sets of constructs used by two individuals are similar, the way each
individual organizes (i.e., creates relationships among) constructs often differs. To the extent that two persons’
construct systems are similar, their cognitive models will also be similar.

The departure of the RGT from the realm of clinical psychology and its widespread use in nonclinical applications
has loosened the ties between PCT and the RGT. A researcher does not necessarily have to utilize PCT in order for
the RGT to be useful, or a researcher can adopt PCT in his or her work without utilizing the RGT. Others combine
the RGT and the theoretical lens of PCT. In a recent article, Davis and Hufnagel [2007] chose to use both the RGT
and PCT to provide insights about the personal constructs of fingerprint technicians around a technical automation
project. A comprehensive discussion of PCT can be found in Kelly’s own works [1955; 1963] and in Bannister and
Fransella [1986]. Appendix I contains more detailed information on PCT’s main postulate and corollaries.

III. CONDUCTING THE REPERTORY GRID TECHNIQUE

Figure 1 shows a sample repgrid, created as an illustration for this paper. The purpose of this grid might be to
understand how an individual perceives different email programs. Within the figure, the labels along the bottom of
the repgrid identify the elements. In the original RGT methodology created by Kelly, elements were always people;
however, over time researchers have examined many different objects as elements. In Figure 1, the elements are
email programs. Labels along the sides of the grid comprise the bipolar personal constructs, and the numeric values
within the grid represent the individual’s ratings of the elements along each individual construct. The RGT provides a
technique in which the interviewer and the participant create the repgrid by defining the elements, identifying the
constructs, and rating each element on each construct.

Construct

Rating

Element

Figure 1. Sample Repertory Grid

Volume 23 Article 3
40
The RGT consists of four major stages: pre-interview, interview, review, and analysis. Each stage consists of a
combination of activities that a researcher or analyst chooses and configures to suit his or her needs. These
processes are illustrated in Figure 2 and are outlined below.

1. Pre‐Interview
Understand the knowledge domain.

Interviews, Questionnaires, Research, Surveys, etc.  

Choose source of elements and constructs.

Researcher Participant
Identify elements.

Confirm participant’s element familiarity. Use element pool or role list to identify elements.

Identify constructs.

Minimum Context Form Full Context Form
Confirm participant’s construct familiarity.

Triads Dyads Multiple user‐

Difference  created 
Method groupings

Opposites 
Method

Link elements and constructs
Have participant rate, rank, or dichotimize elements.
2. Interview

Optional: Write elements on cards  and have participant physically separate them.

Ladder

Probe into construct meanings to elicit more constructs.

Conduct post‐interview confirmation
3. Post‐Interview

Confirm with participant that grid accurately reflects perceptions of the knowledge domain.

Review elements and constructs

Review elements and constructs for appropriate ranges  of convenience.

Analyze results

Analyze the data and draw conclusions.
4. Analysis

Figure 2. Four Stages in Creating a Repgrid

Stage 1. Pre-Interview
In this stage, the researcher becomes familiar with the knowledge domain so he or she can make informed
decisions about interview methods and direct the research participant in completing the repgrid. The two primary
activities of this stage are acquiring understanding of the knowledge domain and choosing the source of the
elements and constructs.

Volume 23 Article 3
41
 Acquire an Understanding of the Knowledge Domain
A researcher can become familiar with the knowledge domain by conducting interviews, distributing questionnaires,
performing preliminary research, forming focus sessions with groups or individuals [Grudnitski 1984; Langan-Fox
and Tan 1997], or reviewing relevant literature.

Choose the Source of Elements and Constructs

Although we treat element and construct elicitation separately, and in further detail later in the paper, the pros and
cons regarding the sources of both are similar. In light of the following guidelines, the researcher must choose the
source from which he or she will identify elements and constructs.

Identified by the Researcher. Before the interview begins, the researcher can identify repgrid elements and/or
constructs [Rentsch 1990]. Doing this reduces interview time and allows researchers to compare grids statistically
along common elements or constructs. However, this introduces researcher bias and offers less personal
representations of an individual’s understanding and preferences [Adams-Webber 1998]. Introducing elements may
be less biasing than supplying constructs if subjects are familiar with the elements.

Identified  by  Each  Individual. Elicitation of elements and/or constructs from individuals introduces the least
amount of researcher bias into the grid structure. Individual elicitation of constructs and elements is particularly
useful in preliminary or exploratory studies where a wide set of diverse attributes is desired [Whyte and Bytheway
1996].

Stage 2. Interview
During this stage, elements and/or constructs are elicited from a participant (unless the researcher supplies them),
and the interviewee enters ratings in the repgrid. If the researcher does not provide the elements or constructs, he or
she first introduces the domain of interest (email programs in our example) to the participant, who identifies
elements with which he or she is familiar. Once the elements are identified, the constructs are elicited. Triadic
elicitation, or triading, is the most common method to elicit constructs in the RGT. In this method, the researcher
presents three random elements to the participant and asks the participant how two of the elements are similar to
each other and different from the third. Once a bipolar distinction between the three elements is made (e.g., fast–
slow), that distinction becomes a construct on which all other elements are rated. All personal constructs require
both the left- and right-hand poles. This process of distinguishing between triads and rating all elements on the
elicited construct continues until the participant cannot think of any new constructs. Some software packages require
a minimum number of elements or constructs to perform some statistical analyses. A researcher may use a
laddering technique during the interview process to help the participant think of new constructs. An analyst performs
laddering by asking “how,” “why,” and “what” questions about an elicited construct that allow the analyst to
understand both the broader context and the deeper meaning of an idea [Reynolds and Gutman 1984; Reynolds
and Gutman 1988].

With this overview of the interview stage, we will now discuss specific practices during the interview stage such as
interviewee familiarization, element identification, construct identification, and linking. Throughout the process, the
interviewer can employ a laddering technique to clarify and expound upon participant meanings.

Interviewee Familiarization
If the researcher chooses the elements or constructs during the pre-interview stage, he or she can hold a
familiarization session with the participants, as Crudge and Johnson [2004] did in their research on search engine
usability. If the participant is not familiar with the elements or constructs, the researcher can change or clarify them,
use a subset of elements or constructs, or drop the participant.

Element Identification
If the researcher did not identify elements during the pre-interview process, he or she must elicit elements from the
group or from the individual during the interview process. The researcher can elicit elements through an element
pool, a role list, or a discussion. An element pool identifies a general domain from which the interviewee can identify
a group of elements. Such a pool could as simple as directing the participant to “list six email programs with which
you are familiar.”

A role list is a list of short sentences (one for each element) that describe typical elements in the domain. Early role
lists used descriptions of people such as “best friend” or “disliked person” to elicit elements. In our example, a role
list could include roles such as “an email program you use often,” “a Web-based email service,” and “an email
program you no longer use.” The participant then matches each role description to an actual entity, such as Gmail or

Volume 23 Article 3
42
Eudora Lite. The role list can be created from scratch by doing preliminary interviews, or it can be put together from
modified versions of lists used in previous research [Carlsson 1988]. With a role list, the participant is forced to cover
select elements within the domain of interest but is free to choose those elements.

Discussions between the researcher and interviewee can also yield a set of negotiated elements and allow the
researcher to direct the element-identification process interactively.

Construct Elicitation
If the researcher did not identify constructs during the pre-interview process, he or she must elicit them from the
group or individual during the interview process. The researcher can elicit constructs by using either full-context form
or a variation of minimum-context form.

Full‐Context Form. Using a full-context form, the participant or group identifies groupings of similar and dissimilar
elements from the list of elements and provides the researchers with a verbal label of each of the groupings [Tan
and Hunter 2002]. For example, given a set of six cards, each with the name of an email program written on them, a
participant might divide them into two piles: Web-based and stand-alone. These two labels become the poles of one
construct. The participant repeats this sorting process until the participant can no longer think of meaningful
construct groupings.

Minimum‐Context Form. Using a minimum-context form of construct elicitation, participants identify groupings of

elements from a random subset of two or three elements rather than from among the entire set of elements. For
example, a researcher presents a participant with a random group of three email programs (a triad) and asks him or
her to describe how two of the elements are similar (convergent or left-hand pole) and a way that the other is
different (divergent or right-hand pole). The researcher repeats this sorting process until the participant can no
longer think of meaningful distinctions or similarities among triads.

Research has shown that small variations in the way in which a researcher carries out the minimum-context form
can affect the resulting repgrid. In particular, studies suggest that element triads are more desirable than element
dyads [Caputi and Reddy 1999; Hagans et al. 2000] and that construct labels distinguished through examination of
element differences are more desirable than those distinguished through element opposites.

Dyads  vs.  Triads. In some research, construct elicitation is performed by prompting the interviewee to identify
distinctions between two (instead of three) elements. Although this dyadic elicitation technique is useful in
experiments where time is a priority or the participants are small children, the general consensus is that constructs
elicited using the dyadic method are not as useful as those elicited using the triadic method [Caputi and Reddy
1999]. Constructs elicited through triads have greater variation, are better able to discriminate among many
elements, and generally produce results that are more cognitively complex [Caputi and Reddy 1999], in part
because they are elicited through comparison of both similarities and differences. Dyadic elicitation focuses only on
differences.

Difference or Opposites. When a researcher presents elements as triads, a participant can identify construct poles
by either the difference method or the opposites method. Both methods prompt the participant to identify and
verbally label how two of the three elements are similar. Using the difference method, the participant would then
identify how the third element is different from the other two. Using the opposites method the participant is prompted
to identify the opposite of the label used to describe the first two elements.

Although these two techniques vary just slightly in their elicitation approach, there are at least three differences in
the outcomes of constructs elicited using the difference and opposites methods [Hagans et al. 2000].

The first distinction is that the difference method creates more bent constructs than the opposites method. Bent
constructs are constructs that do not contrast two opposites and may not be mutually exclusive (see Figure 3). The
problem with bent constructs is that the construct may be describing two ideas embodied in one construct. For
example, the construct happy-sad is mutually exclusive, opposite, and embodies one construct. The construct
happy-confused, however, could be a mix of two constructs: happy-sad and clear thinking–confused. A skilled
interviewer may be able to help a participant determine if a participant means a single construct or multiple.

The second difference between the two methods is that the opposites method creates poles that may be too
extreme to describe the constructs. In describing attributes of email programs, for example, a participant using the
difference method might say that two of the three programs are friendly while the other is less friendly. Using the
opposites method, however, the participant might identify two of the programs as friendly, and choose hostile as the

Volume 23 Article 3
43
 construct’s other pole. The opposites method creates a construct whose pole (hostile) is too extreme to be
appropriately applied to any the elements [Neimeyer et al. 2005].

Figure 3. Graphical Representation of a Bent Construct

The third difference between methods is that constructs elicited through the opposites method may not apply to any
of the elements. Using the example above, none of the programs can be described as hostile, but some are
definitely less friendly than others are. Thus, using the difference method, each pole of the bipolar construct is sure
to apply to at least one of the elements in the set [Hagans et al. 2000]. For these reasons, many researchers prefer
the difference method to the opposite method and will probe further (using laddering or other questions) to examine
bent constructs.

Number of Constructs.  Although no specific rules exist on the number of constructs that should be created, most
studies range from 10 to 25 [Bannister 1968; Fransella et al. 2004], and most salient constructs can be elicited from
just 7 to 10 triads [Reger and Huff 1993].

Linking of Elements and Constructs

After the researcher or participant identifies each construct, or after he or she creates the entire list of constructs, the
participant can link elements and constructs by (1) dichotomizing, (2) ranking, or (3) rating. In this process of linking,
the researcher can employ laddering to elicit additional constructs and/or to understand the meaning of a person’s
constructs more fully.

Dichotomizing.  Kelly’s original repgrid applied a dichotomized (bimodal) evaluation to each element on each
construct. On a construct of simple layout–cluttered, for example, a person would place a tick or checkmark,
indicating that the email program had either a simple design or a cluttered design. If an element needed to be rated
in the middle of a bipolar construct, then the element needed a different construct to explain the ambiguity [Beail
1985]. One problem with this approach is that it can force the research participant to make crisp distinctions where
they might not actually exist.

Ranking. An alternative to the semantic scale is to rank each of the elements on a continuous scale. Stewart and
Stewart [1981] suggest that this method helps the participant decide which elements are similar and which are
different according to the specific construct. Hunter and Beck [2000] use cards to help participants rank attributes of
systems analysts. Unfortunately, this technique introduces method variance, which may lead to inaccurate analysis.
This leaves the rating technique as the preferred method for linking elements and constructs.

Rating. As a solution to the forced distinction problem and to reduce method variance, numerous studies have
modified Kelly’s repgrid to use a semantic scale to rate elements on each construct. This is widely used and
accepted by most repgrid researchers but is not without criticism. Eden and Jones [1984] point out that this
approach may be contrary to Kelly’s original theory [1955] of dichotomous constructs. Another problem is that a
rating in the middle of the scale could be interpreted by the researcher incorrectly as either indifference or
uncertainty [Eden and Jones 1984; Yorke 1983]. Despite these critiques, our contention is that rating scales provide
the best option of the three when linking elements with constructs.
 
Laddering. Laddering is a technique that allows the researcher, together with the participant, to explore the
rationale for, the meanings of, and the relationships between different constructs. Applied to repgrids, this technique
requires the researcher to qualitatively identify and further elicit similar elements/constructs and construct
hierarchies. A researcher can use this technique in any stage of the repgrid interview process to elicit more
information about the constructs and explore their hierarchical relationships [Rugg et al. 2002]. To perform laddering,
the researcher uses a set of standard questions to move sideways, upward, or downward from a seed item, or
starting point. The researcher can then use a standard notation set to record the research participant’s responses.

Laddering can be especially useful in understanding evaluative but ambiguous constructs such as friendly–less
friendly. This construct, while evaluative, gives the researcher little actionable insight into how to make an email
program more or less appealing.

Volume 23 Article 3
44
To probe downward, the researcher might ask, “What characteristics of the program make it more or less friendly?”
To which the participant might respond with new constructs like familiar logo branding–unfamiliar branding, or
current constructs like intrusive advertisements–no advertisements. These new constructs help researchers
understand a person’s construct system and give insight into hierarchical relationships between constructs.

In a similar fashion, a researcher can use laddering to discover higher-level constructs. For example, if the
participant had provided the construct familiar branding–unfamiliar branding, then the interviewer could ask, “Why is
that distinction important to you?” and the subject may have provided the friendly–less friendly construct.

Sideways laddering uncovers items similar to the one in question. Using the construct split into panes–no panes as
the seed item, the researcher might say, “This construct describes the structural layout of the program. What other
pertinent attributes can you think of that describe structural layout?” Young et al. [2005] use laddering to go deeper
into interviewee meanings of their communicative and noncommunicative constructs. For a full treatment of the
laddering technique, see [Reynolds and Gutman 1984; Reynolds and Gutman 1988].

Stage 3. Post-Interview
In the post-interview stage a researcher or analyst can review the created repgrid with a participant to refine the
elements or constructs as needed. This can be done by confirming the grid and/or reviewing the elements and
constructs.

Confirm Grid
This optional step allows the researcher and participant to quickly confirm that what was recorded during the repgrid
interview represents the true beliefs, feelings, meanings, perceptions, or interpretations of the research participant
and that those attributes are accurately understood by the analyst [Hunter 1994]. During this process, the researcher
allows the participant to change any ratings that he or she desires.

Review Elements and Constructs

After the elements and/or constructs have been elicited, the participant can review them to ensure that they
appropriately cover the desired range of the knowledge. This can be useful in determining missing elements or
constructs. If the range of elements is too wide, a researcher can work with the research participant to eliminate
elements or can group elements from tangential domains into separate grids [Eden and Jones 1984] and link them
[Bradshaw et al. 1987]. If the element scope is too narrow, the researcher can use laddering to explore more
elements, use a role list, or create imaginary “ideal” and “non-ideal” elements to spread the range of elements to the
extremes [Crudge and Johnson 2004; Hunter and Beck 2000; Stewart et al. 1981; Whyte and Bytheway 1996;
Young et al. 2005]. The addition of ideal and non-ideal elements also lets the researcher anchor comparisons
between elements and constructs [Hunter 1994].

During and after the post-interview process, participants should be able to change ratings, constructs, or elements
until they feel that the repgrid accurately represents their understanding of the knowledge domain.

Stage 4. Analysis
During the analysis phase, the researcher or analyst submits the completed repgrid to either statistical or qualitative
analysis. Analysis helps the researcher understand the content and organization of the participant’s construct
system. Researchers can analyze the content of the grid (its element and construct labels) or the grid ratings.
Content analysis can be used to understand textually the construct labels. Ratings analysis provides insight on
content organization. Researchers often use cluster analysis, principle components analysis (PCA),
multidimensional scaling (MDS), logical analysis, and conceptual mapping to analyze ratings. They also occasionally
employ various summary measures to capture a single aspect of the grid structure. Various software packages are
available for analyzing and creating repgrids. We have listed many of these in APPENDIX II.

Once a researcher and a participant create a repgrid and rate elements on constructs, the researcher has a matrix
of labels and numbers that yields two areas of analysis: the construct labels and the element-construct ratings.

First, the element and construct labels themselves are a type of data that can help the researcher draw qualitative
interpretations of how the participant views the domain in question. Alone, a researcher can group and analyze
these construct labels without regard to the ratings using content analysis or some other type of qualitative coding.

Second, the numerical evaluations of the elements on the constructs give the researcher a deeper look into the
meaning of and relationships between constructs and elements. Analysis of element-construct ratings reveals

Volume 23 Article 3
45
 relationships (1) between elements, (2) between constructs, and (3) between elements and constructs jointly [Bell
1990]. Analysis of these element-construct ratings falls into one of four categories:

1. Statistical analysis (cluster analysis, PCA, or MDS)

2. Conceptual mapping

3. Logical analysis—Gaines and Shaw method, uncertainty coefficient, phi coefficient, Proportionate-
Reduction-in-Error measure, etc.

4. Summary measures—Landfield's Ordination, Bieri’s Cognitive complexity, etc.

Multiple analysis techniques can be combined in the same study [Boyle 2005] to enhance validity of results [Latta
and Swigger 1992]. We describe each of these techniques in the following paragraphs. We then discuss their
extension to the analysis of group-level data.

Eden and Jones [1984] suggest that the element and construct elicitation phases are perhaps the most valuable
stages of the repgrid technique and that rating elements is often too time consuming to be worth the effort. In fact, in
his study on personal constructs for systems projects’ risk, Moynihan [1996] elicits elements and constructs from his
participants but forgoes the linking step altogether. When analyzing elements or constructs, researchers can also
use frequency counts to find common trends among individuals [Boyle 2005].

A researcher can analyze elements and constructs without their corresponding ratings using a technique called
content analysis. Content analysis [Feixas et al. 2002; Hunter 1997; Moynihan 1996] is a subjective qualitative
analysis by which elements and/or construct labels are placed into common categories or main issues and
interpreted for meaning [Langan-Fox and Tan 1997].

Applied to repgrids, this would involve qualitative identification of groupings of thematically similar elements or
constructs. Using our email example as an illustration, a researcher might assign elicited constructs into the
following groups: features, overall perceptions, layout, and other. He or she could then group elements into stand-
alone, Web-based, and hybrid categories. These subjective groupings are useful for the generalization of constructs
and elements, for identifying trends between classifications and for creating questionnaires. It is also useful, as we
will discuss later, in comparing different grids whose elements and/or constructs are not similar.

Statistical Analysis of Ratings

Many statistical techniques are appropriate for the analysis of repgrid ratings. We will discuss cluster analysis, PCA,
MDS, conceptual mapping, logical analysis, and other summary measures in this section.

Cluster/Correlation  Analysis. Cluster analysis enables the researcher to analyze relationships between
constructs or between elements, but not between elements and constructs together. Cluster analysis of repgrids
involves the quantitative identification of groups of similar elements or similar constructs based on highest statistical
correlations. An analyst can express these groupings visually on a tree diagram, as in Figure 4.

The tree diagram visually portrays possible relationships between the elements and between the constructs. By
glancing at this figure, the researcher will recognize three main groups of closely correlated elements: group one
consists of Ideal and Gmail; group two consists of Yahoo, AIM, Hotmail, and Non-Ideal; and group three consists of
Outlook Express and Free Eudora. The first six elements also form a cluster of elements that are more similar to
each other than they are to Outlook Express and Free Eudora.

These groupings also give the researcher insight into the hierarchical structure of a participant’s personal constructs
[Shaw and Gaines 1998], in theoretical accordance with Kelly’s organization corollary (see Appendix I).

Most of the groupings in Figure 4 make sense on their face. Programs that are cool will also be friendly, and those
that are fast will naturally have less advertisements and a simple layout. However, when seemingly unrelated
elements or constructs show high correlations (such as constructs landing page–no landing page and individual
emails–threaded conversations), a researcher may want to explore the reasoning behind anomalies with the
research participant.

An exploratory variation on cluster analysis is simple linkage clustering, as described in [Eden and Jones 1984]. This
method computes the absolute value of the sum of the numerical differences between pairs of constructs or

Volume 23 Article 3
46
elements. These values are normalized on a 0-to-100 scale where 0 indicates a complete dissimilarity of constructs
or elements and 100 means a complete similarity.

Figure 4. A Tree Diagram from Gaines and Shaw’s Rep IV Software, Showing Relationships between
Elements and Constructs

For example, the absolute value of the sum of numerical differences between two statistically similar programs like
Outlook Express and Free Eudora is 10. The difference between two statistically different programs like Gmail and
Hotmail is 47. The lower the number, the more closely the pair of elements is related. This type of cluster analysis is
less complicated and more transparent than its statistical variant and can be done easily by hand in grids with less
than seven elements.

Another variation on correlation analysis involves measuring how strongly a construct is correlated with an
individual’s ideal [Whyte and Bytheway 1996]. This analysis requires the researcher to supply an imaginary “ideal” or
“successful” element that the participant will use to make ratings.

Cluster analysis is conceptually straightforward and enables the researcher to quickly identify relationships between
constructs or between elements. There are, however, drawbacks to cluster analysis [Easterby-Smith et al. 1996].
First, cluster themes are often difficult to label or identify. Second, specific correlation algorithms may differ from one
procedure to another. Most cluster analyses of repgrids use the FOCUS algorithm, created by Shaw [1980]. Third,
constructs may correlate mathematically, but may not make logical sense. Fourth, the technique cannot show
element-construct relationships. Fifth, clusters show correlations but do not show which factors in a cluster might be
most important in a decision-making process [Phythian and King 1992]. 
 
Principle  Components  Analysis. The second main type of statistical repgrid analysis is performed using a
technique called principle components analysis (PCA) [Bell 1990; Easterby-Smith et al. 1996; Leach 1980; Tan and
Hunter 2002]. PCA is a data reduction technique used to find dimensions of maximum variability in data. Some
analysis packages (e.g., Rep IV) utilize PCA to produce plots that represent relationships among the constructs and
elements spatially. Spatial distance between and among elements and constructs suggests how they might be
related to each other.

A graphical representation of PCA is shown in Figure 5. Using this method, the researcher can easily notice clusters
of common elements and constructs. He or she can also observe that the general cluster of Outlook Express and
Free Eudora is spatially closest to those constructs that effectively describe the stand-alone class of email
programs—not integrated into a portal, split into panes, and stand alone.

Volume 23 Article 3
47
 Figure 5. PCA of Email Programs

Multidimensional Scaling (MDS): MDS is an analytical technique designed to create graphical representations of
the similarities and differences in the ratings of multiple objects. MDS is an alternative for mapping the distance
between ratings of constructs and elements in multidimensional space. Because PCA is built into many of the
software packages for analyzing repgrids, it is the most common method of performing exploratory quantitative
analyses on repgrid data. However, MDS is designed to examine spatial distances between grid ratings, which can
be easily performed using statistical packages such as SPSS or SAS [Bell 1994; Bell 1997; Bell 2003; Van der Kloot
1981]. As a result, MDS is gaining popularity as a technique to analyze repgrid data graphically.

Conceptual Mapping
Phythian and King [1992] use conceptual mapping in their construction of a decision-based system for evaluating
tender enquiries. After performing a cluster analysis with two experts, the analysts map the statistically strongest
relationships and discuss the maps with the two experts. The mapping process has three advantages. First, it
confirms and fleshes out the relationships with the user. Second, it allows the depiction of causal relationships.
Third, it can help the interviewee think of other important relationships that might have been overlooked during the
primary elicitation process. The individual maps are then negotiated between the two experts, and a single map is
created.

Logical Analysis
Cluster analysis, PCA, and MDS are distance-based methods. In other words, they reveal the degrees of relative
equality between and among elements and constructs by calculating the statistical distance between them.

Logical analysis, in contrast to the distance-based methods, seeks to identify not only the degree of equality but also
the directional implications between and among elements and constructs. In the analysis of repgrids, this technique
involves the quantitative identification of logical, inductive relationships among constructs and elements. In the field
of IS, these directional implications are useful for revealing relationships within repgrid structure that can be directly
translated into machine-readable logic for rapid prototyping of knowledge-based systems [B. R. Gaines 1993].
Asymmetrical relationships also give researchers a way of assessing one-way relationships without resorting to
laddering or implications grids [Bell 2004].

Distance-based analyses of a grid that are designed to elicit information for a system supporting a decision to use
one program over another, for example, might show the correlation (degree of relative similarity) between the

Volume 23 Article 3
48
constructs Web based–stand alone and slow–fast. Logical analysis can deepen the researcher’s understanding of
this relationship by suggesting its direction—that stand alone entails fast, but that fast does not necessarily entail
stand alone. By looking at a repgrid, we know that all stand-alone applications are fast. However, one Web-based
application—Gmail—was also fast. Stated technically, the right-hand pole (RHP) rating of the Web based–stand-
alone construct on all elements necessarily leads to an RHP rating of the slow–fast construct for all elements.
However, the opposite is not always true because a fast email application might be Web based. This asymmetrical
relationship can be written as follows [Gaines and Shaw 1981, pp. 155-157]:

RHPm RHPn

where m = Web based–stand-alone and n = slow–fast.

These logical relationships can also be transitive. If LHPm LHPn and LHPn LHPo, then LHPm LHPo [Gaines
and Shaw 1981].

The interpretation of these logical relationships forms the basis for expert and decision support systems created with
the repertory grid. Figure 6 is a graphical representation of construct entailment and shows both asymmetry (one-
way arrows) and equivalence (two-way arrows) in this participant’s construct space. For a more in-depth look at how
this logic can be analyzed in grids and be useful for the creation and prototyping of knowledge-based systems, see
[Gaines and Shaw 1980].
RHP7

LHP4

RHP9 LHP2 LHP6

RHP1

Figure 6. Graphical Representation of Construct Entailment, from [Gaines and Shaw 1980]

Asymmetrical logical relationships can also be explored using an uncertainty coefficient [Gaines and Shaw 1986],
the phi coefficient, or a Proportionate-Reduction-in-Error measure (PRE) [Smithson 1987], as mentioned by Bell
[2004].

Summary Measures
Beyond clustering, factor analysis, content analysis, and logical analysis, many repgrid software packages also offer
a variety of summary metrics. Such quantitative measures assign numerical numbers to repgrids, often for
comparison with other grids. Measures, such as Landfield's Ordination [Gallifa and Botella 2000; Landfield and
Cannell 1988], Bieri’s measure of cognitive complexity [Baldwin et al. 1996; Bieri 1955], or Gallifa and Botella’s
[2000] structural quadrants, sum up a repgrid in a small set of numbers. Similarly, construct or element measures
such as discriminative power or extremity ratings can summarize various attributes of constructs and elements.
These summary measures can be used to explore other relationships in the data and can inform future research.

Extending Analysis to Groups

Repgrid and Groups. The repgrid can be used to investigate understanding at the group level by identifying
semantic areas of consensus, conflict, correspondence, and contrast [Shaw and Gaines, 1989]. Repgrids can be
used in a group setting to elicit group constructs or to compare the constructs of individual group members with
other team members.

PCT and Groups. PCT suggests that an individual’s constructions are unique to that one person. However, PCT’s
commonality and sociality corollaries suggest that construct systems can be similar and individual constructs can be
shared with others. In a study of the commonality of constructs, Latta and Swigger [1992] validated the repgrid’s
reliability and integrity for modeling communal understanding. They use a number of analyses to test the degree of
similarity between students’ construction of information relevant to the domain of information retrieval systems.

Analyzing Group Repgrids. There are three ways that repgrids can be analyzed in a group setting. The first way
that a repgrid can yield group information is to have a single group complete a single repgrid whose elements and/or
constructs have been identified by the group itself or by the researcher. Analysis of this single group-created grid
can be performed using any of the techniques described above. However, because of the personal nature of

Volume 23 Article 3
49
 constructs, conclusions and interpretations of a group’s repgrid should be verified with the team and its individual
members where possible [Easterby-Smith et al. 1996].

A second way of analyzing repgrids to acquire information from individuals, and then combine the grids into one
single grid for analysis. An advantage of this type of group analysis is that ambiguities arising from the single group
grid can be explored through deeper analysis of the single grids. An example of how grids can be combined is
Phythian and King’s [1992] case study of the development of an expert system. To facilitate this process, there are
specialized repgrid software packages designed to perform cluster or factor analysis on multiple individual grids
whose constructs and/or elements are the same. Of the packages referenced in Appendix II, IdeoGrid, Rep IV, and
GridSuite each support some form of collective grid analysis.

Where the elements and/or constructs of multiple repgrids are not the same, they can be compared to the global
measures mentioned previously. Multiple grids can also be analyzed by performing a content analysis on each and
identifying or interpreting common emergent themes [Whyte and Bytheway 1996]. An additional way of observing
trends across repgrids is to perform a frequency count of constructs. Those constructs that appear most frequently
may be significant to a developed understanding of the knowledge domain [Boyle 2005; Whyte and Bytheway 1996].

Of course, other designs are possible. Wright [2002; 2004a] provided his participants with elements and, once they
were done, aggregated their constructs and ratings into one grid, which was analyzed as a single group grid.

IV. POTENTIAL USES OF THE RGT IN IS RESEARCH

Having provided an overview of how the RGT is conducted and its many applications, we will now elaborate on how
the RGT has been used in IS research in the past and how it may be applied in the future. IS researchers can apply
the RGT to a variety of areas, bounded mainly by the research participant’s domains of knowledge and perception
[Hassenzahl and Wessler 2000]. The RGT is an effective tool for beginning to understand a problem [Chao et al.
1999], exploring the problem’s scope, and subjectively evaluating a problem. It allows researchers to discover basic
understanding of individual perceptions, meanings, and cognitive models that may be helpful in solving IS problems
[Chao et al. 1999; Phythian and King 1992]. Additionally, the RGT has been shown to be useful as a tool to monitor
changes in a participant’s understanding or perceptions over time [Carlsson 1988].

Because of its focus on understanding and problem solving, the RGT has wide potential application in IS research.
The most popular use of the RGT in IS literature has been from a design science perspective [Hevner et al. 2004]
Specifically, the RGT has been used as a knowledge elicitation technique in the design of expert systems [Boose,
1989] and a variety of other systems [Byrd et al. 1992]. The RGT has also been used to help researchers
understand users’ mental models within the context of performing information storage and retrieval tasks [Dillon and
McKnight 1990], as well as the cognition of those performing general systems analysis and design [Hunter 1994]

RGT in Decision-Based Research

Decision-based research using the RGT goes back at least two decades. Furthermore, as overviewed by Boose
[1989], hundreds of decision-based systems have been created using the RGT. In Boose’s overview, he suggests
several uses for the RGT in creating decision-based systems, including creating decision aids, system development
and delivery aids, group decision aids, online knowledge bases, project exploration and feasibility tools, teaching
aids, expert system-building tools, and tools that provide situation insight [1989, p. 215]. Other examples of research
using the RGT to create decision-based systems and to support knowledge acquisition for expert systems include
[Castro-Schez, Castro et al. 2004; Castro-Schez, Jennings et al. 2004; Hwang 1992; Liou 1990; Phythian and King
1992]. More recently, research has proposed using the RGT to elicit buyer and seller information to support the
creation of automated negotiation agents [Castro-Schez, Castro et al. 2004; Castro-Schez et al., 2005]. Finally, a
recent study compared the efficacy and usefulness of the Kepner-Tregoe (KT) approach and the RGT as methods of
eliciting decision criteria and alternatives for multi-criteria decision-making processes [Scheubrein and Zionts 2006].
They found that KT elicited more criteria, while the RGT elicited more alternatives. In addition, the repgrid was
completed 33 percent quicker than the KT approach.

RGT in Information Storage and Retrieval Research

Information storage and information retrieval research has also benefitted from the RGT. The earliest study in this
area used the RGT to explore how readers perceive different types of text, with the purpose of suggesting how to
format and design electronic texts [Dillon and McKnight 1990]. Another used the RGT to model a querying user’s
view of relationships between types of documents [Bhatia 1992]. A couple of studies used the RGT for advancing
information storage and retrieval of medical information [Jung and Gudivada 1995; Madigan et al. 1994]. Other
notable RGT studies in information storage and retrieval research include [Bhatia and Deogun 1998; Bhatia and Yao
1993; Gudivada et al. 1994]. Finally, Xiangmin and Chingell [2001] investigated how educational and professional

Volume 23 Article 3
50
status, first language, academic background, and computer experience influence a person’s mental model of
information retrieval systems. They were able to classify different types of users, which is helpful in understanding
each class’s perceptions and needs in a system.

RGT in Systems Analysis, Design, and Development

Given the problem-defining and problem-solving nature of the RGT, it is perhaps not surprising that the most
extensive application of the RGT in the IS literature has been in conjunction with improving systems analysis,
design, and development.

Systems Analysis
In terms of systems analysis, the earliest study that we found used the RGT to acquire managers’ most important
sources of information [Grudnitski 1984]. Hunter [1994; 1998; 1997; 2000] performed several studies on different
perceptions and assessments of the systems analyst, highlighting key perceptions [Hunter, 1994] and cross-cultural
views [Hunter and Beck 2000] of the system analyst. These types of studies explore job descriptions to help systems
analysts understand the social and behavioral aspects of other analysts within organizations. Another study explored
the concept of IS fit with an actual system in the field to help the researchers understand and communicate to
management what was perceived as wrong with the system and what was wanted from the system from
stakeholders [Kanellis et al. 1999].

Design
In terms of design, one study used the RGT to rate design prototypes, which provided strong insights into design-
relevant factors of user interfaces. [Hassenzahl and Wessler 2000]. Another study used the RGT to understand key
aspects of user interface design on the Internet [Verlinden and Coenders 2000]. Finally, a study used the RGT to
explore user-centered design evaluations of search engines [Crudge and Johnson 2004].

Development
In the systems development research domain, one study explored project risk planning from the vantage point of
expert IS project managers [Moynihan 1996]. Another study examined extreme programming team membership
roles using the RGT [Young et al. 2005].

Though research in these areas has been more extensive than other areas, the technique could help inform many of
the analysis, design, and development topics that have yet to be addressed with the RGT. As an example, we
illustrate the possibilities in using the RGT to develop organizational-level requirements.

Organization-level requirements define the broad strategy and infrastructure of the entire information system across
an organization. They are broader in scope and less detailed than application-level requirements. Effective analysts
consider organizational and policy issues when they analyze systems requirements [Davis 1982]. Organization-level
requirements establish the overall portfolio of the organization’s applications, dictate common communication
interfaces across applications, specify the priority of application development, and set the structure of common data
requirements.

In management, researchers use the repgrid to understand managers’ models of cognition in a diverse and varied
set of business units across the corporation [Ginsberg, 1989]. In that same arena, other researchers use the repgrid
to model high- and low-performing managers’ cognition of the overall strategy-making process [Wright 2004b].
Laddering is used to involve stakeholders in the identification of elements of IS structure that enhance performance
and lead to accomplishing organizational objectives. Yet, little similar work has been done in IS.

Recognizing this opportunity, Tan and Gallupe [2006] recently used the repgrid to understand an aspect of business-
IT alignment. They build on Reich and Benbasat’s [2000] alignment literature, which suggests that alignment can be
increased through shared understanding of business and IS executives. Tan and Gallupe [2006] use the repgrid to
measure the relationship between business and IS managers’ cognitive commonality and business-IS alignment.
Their research finds that increased shared cognition between business and IS executives generally correlates with
greater business-IS alignment. Still others have used the repertory grid to elicit user-perceived reasons behind a
misfit of business needs and IT [Kanellis et al. 1999].

These studies suggest that PCT techniques in general and the repertory grid in particular could be useful in the
development of organization-level IS strategy and alignment. The repgrid’s ability to detect minute differences
between alternatives could also help prioritize application development or help executives weigh the pros and cons
of various enterprise-wide system configurations and strategies.

Volume 23 Article 3
51
 Other IS Research Using the RGT
Aside from the aforementioned IS research areas, IS researchers have found several other creative areas of RGT
application. One study examined how users perceive the success of information systems in terms of an
organization’s developed systems [Whyte and Bytheway 1996]. Another used the RGT to explore user attitudes
toward technology [Frewer et al. 1998]. Another showed how the repgrid can be helpful in exploring cognition in
information systems research [Tan and Hunter 2002]. Clases et al. [2003] used the RGT to explore the subjective
meaning of trust in two Swiss virtual organizations, which knowledge was used to improve a survey they were
constructing. Oppenheim et al. [2003] used the RGT to elicit from top managers their personal views of information
assets. The RGT has also been used to explore Web site evaluation criteria [Tan and Tung, 2003], analyze Web
sites [Hassenzahl and Trautmann 2001], and measure Web site appeal [Wessler and Ortlieb 2002]. Boyle [2005]
suggests using the RGT to elicit from various IT team members (e.g., programmers, technical specialists, IT
managers) the desirable and undesirable traits of other team members in various positions in order to improve team
performance. Finally, in a recent article, Davis and Hufnagel [2007] chose to use both the RGT and PCT to provide
insights about the individual meanings of fingerprint technicians around a technical automation project.

Other Potential applications of the RGT in IS Research

In addition to the aforementioned applications of the RGT in IS research, we believe there are several additional
potential uses for the RGT in IS research, including modeling of practitioner problems, building taxonomies,
understanding group conflict, surveying information space, verifying or validating constructs, instrument
development, and theory building or hypothesis testing.

Communication of Requirements
One potential novel use of the repgrid in IS research is to use it to communicate results of requirements
determination with stakeholders and other analysts, because the repgrid can be an effective communication
medium. In fact, Kelly’s [1955] own original use of the repgrid—eliciting and then communicating back to the patient
his or her perception of elements and constructs—suggests that communication may be a primary advantage of the
finished or analyzed repgrid. Although we know of no empirical studies that specifically test the repgrid’s ability to
represent information effectively with stakeholders, Verlinden and Coenders [2000] suggest that the ease with which
repgrid data is transformed into visual diagrams makes the repgrid “one of the best ways to communicate user
experiences to all stakeholders in the development of web sites” (p. 144). Other research also shows that the repgrid
is an efficient tool for normalizing and standardizing a group’s vocabulary, which enhances communication [Shaw
and Gaines 1989]. Future research could explore the effectiveness of the repgrid in representing the results of the
determination of requirements to other analysis or to key stakeholders.

Modeling of Practitioner Problems

The repgrid provides a way for a researcher to analyze, decompose, and further understand a problem [Eden and
Jones 1984; Poulymenakou et al. 1990]. It can be “used to represent what people know about problems”
[Poulymenakou et al. 1990, p. 187]. The technique is also effective for defining and organizing semi-structured or
ambiguous problems [Scheubrein and Zionts 2006], because it is useful in eliciting tacit knowledge that a person
being interviewed cannot easily articulate [Shadbolt and Milton 1999].

Building Taxonomies
Whether one is trying to classify facial features [Gudivada et al. 1994], medical documents [Madigan et al. 1994], or
electronic texts [Dillon and McKnight 1990], the repgrid forces the research participant to verbalize similarities and
differences between elements. This can be helpful in classification or taxonomy building, which is increasingly
common in IS research.

Understanding Group Conflict

A researcher can use the repgrid to help resolve conflicts or bridge gaps in understanding by eliciting mental models
of a particular domain from multiple individuals and then comparing those models [Wright 2002]. This process
reveals the sources of conflicts and can help to resolve them. This could be particularly helpful in collaboration,
group, and computer-mediated communication research in IS, where interpersonal conflict can be a tremendous
source of group process losses. Similarly, the RGT could be used to better understand group mental models, group
processes, consensus, and so forth.

Surveying Information Space

Information space, or “the objects (real or virtual) to which the individual turns to acquire information” [McKnight
2000, p. 730], can be modeled through repgrid interviews [McKnight, 2000]. In the aforementioned study [McKnight
2000], the RGT was used to understand the interviewee’s personal construct system related to the online and offline

Volume 23 Article 3
52
sources of information (e.g., email, newsgroups, electronic journals, conferences). Understanding an individual’s
meanings associated with different sources of information may be useful in understanding the cognitive stopping
rules [Browne et al. 2007] associated with information search activities. The personal constructs that provide
meaning to artifacts in an individual’s information space may also be helpful in understanding how individuals make
conclusions regarding the reliability of information encountered within an information system.

Verifying or Validating Constructs

The repgrid is an efficient tool for researchers who need to generate user-centered constructs. User-centered
constructs can be compared to established and commonly employed constructs in order to identify gaps in the
literature or deficiencies in established constructs’ abilities to measure the intended phenomenon correctly [Crudge
and Johnson 2004; Whyte and Bytheway 1996; Wright 2004b]. While this approach has been utilized in other fields,
the use of RGT for construct creation is rare in IS.

Instrument Development
The RGT may be used fruitfully to help generate questions for survey instruments. Generating scale items from
subjects may help overcome the problem of researcher bias that often results in scale development [Hallsworth
1988]. A researcher using the RGT can gather information on important aspects of a particular domain that can
inform questionnaire development. This approach has been suggested in education [Lambert et al. 1997], social
work [Hutchinson 1998], and health care research [Stewart et al. 2003] and could be used to strengthen the validity
of IS survey instruments.

Other RGT Research Issues and Future Directions

Aside from the above areas of potential research, there are several other research issues with the RGT that are
highly pertinent to IS researchers. Researchers have created many different ways of implementing repgrid
techniques. Although the flexibility of the technique lends itself to a wide range of situations, such a wide range also
makes comparison and verification of results difficult. We have discussed how researchers used the RGT in design
science and other IS research. Studies suggest that various techniques are best suited to specific domains [Byrd et
al. 1992] or different types of knowledge [Chao et al. 1999]. Hence, there is a need for more research on the optimal
use of the RGT in various IS research domains and for eliciting specific types of knowledge.

Liou [1990] suggested that a researcher could use a combination of techniques to elicit knowledge and that the
repertory grid is a useful supplementary technique. Chao, Slvendy et al. [1999] similarly suggested that different
knowledge elicitation techniques could be combined for larger coverage of a knowledge domain. Along these same
lines, Byrd et al. [1992] suggested future research be done on the “synergistic effects of elicitation techniques” [p.
118]. Future research could also examine which techniques are best suited for combination with the repertory grid to
maximize the amount of knowledge elicited in specific tasks.

Poulymenakou, Cornford et al. [1990] suggest that different types of managerial problems and decisions require
different methods of problem construction. More specifically, they state that as problems evolve from the operational
level to the tactical level and the strategic level, the methods of problem construction should correspondingly vary.
Future research on the effectiveness of the repertory grid as a tool for IS problem construction could yield important
insights.

Separately, since the results from many grids can be compared easily, the RGT can be useful for inductive theory
building [Hunter 1994] via Grounded Theory [Glaser and Strauss 1967]. During the early stages of knowledge
elicitation, the repgrid can help turn participants’ attention to previously unexplored topics [Grudnitski 1984;
Hassenzahl and Wessler 2000; King and Phythian 1992; Phythian and King 1992], and help explore objectives in
fields of research that have not yet developed testable hypotheses [Clases et al. 2003]. Again, such applications are
rare in IS research.

Moreover, future literature could explore RGT’s ability to confirm theory-based hypotheses. To this end, future
research could determine which perspectives and theories the repgrid best complements. They might suggest
theories based on action-theory perspective (which states that human actions are driven by subjective meanings
and by their interpretation of events, and that human action can be interpreted only in light of these perceptions). Or
they might suggest theories based on the grounded theory perspective, the socio-technical perspective, and the
service perspective of IS user satisfaction.

Finally, the RGT may also be useful in comparing the effectiveness of conceptual models developed using
competing modeling grammars (e.g., ontologically clear vs. unclear) as a communication tool between the

Volume 23 Article 3
53
 stakeholder and the analyst. RGT could thus be used as alternative comparison methodology as compared to
instigating information recall or problem-solving tasks.

V. CONCLUSION
The repgrid technique is a useful and powerful method for understanding individual perceptions, meanings, and
mental models. In this paper, we provided an accessible overview of the repertory grid technique, its underlying
theoretical background, and specific procedures and issues that researchers should understand when utilizing the
method. The RGT has been presented in an organized four-phase framework, and the options for executing the
RGT in each of these phases have been reviewed. We have also provided a review of the extant IS research that
employs the RGT and have suggested several areas of research that may benefit from use of the RGT. As
increasing attention is given to the cognitive processes and the meanings construed by those who develop and use
information systems, the utility of cognitive mapping approaches such as the RGT increases. It is our hope that the
guidelines and suggestions featured in this tutorial will encourage further research to aid those scholars who are
interested in rigorously employing the RGT to enhance their work.

ACKNOWLEDGEMENTS
We want to thank Joey F. George, Katie Newbold, Laura Rawlins and the anonymous reviewers for their reviews
and outstanding help writing this paper. We also acknowledge funding received from the Information Systems
Department, Marriott School, Brigham Young University.

REFERENCES
Adams-Webber, J. R. (1998). "Differentiation and Sociality in Terms of Elicited and Provided Constructs,"
Psychological Science 9(6), pp. 499-501.
Baldwin, D. A., J. N. Greene, R. E. Plank, and G. E. Branch. (1996). "Compu-Grid: A Windows-Based Software
Program for Repertory Grid Analysis," Educational and Psychological Measurement 56(5), pp. 828-832.
Bannister, D. (1965). "The Rationale and Clinical Relevance of Repertory Grid Technique," The British Journal of
Psychiatry 111(479), pp. 977-982.
Bannister, D. (1968). The Evaluation of Personal Constructs. London, New York: Academic Press.
Bannister, D. and F. Fransella. (1986). Inquiring Man: The Psychology of Personal Constructs (3 ed.). Dover: Croom
Helm.
Beail, N. (1985). "An Introduction to Repertory Grid Technique," In N. Beail (Ed.), Repertory Grid Technique and
Personal Constructs (pp. 1-26). Cambridge, Massachusetts, United States: Brookline Books.
Bell, R. C. (1990). "Analytic Issues in the Use of Repertory Grid Technique," Advances in Personal Construct
Psychology 1(1990), pp. 25 - 48.
Bell, R. C. (1994). Using SPSS to Analyse Repertory Grid Data.
Bell, R. C. (1997). "Using SPSS to Analyse Repertory Grid Data," Paper presented at the Proceedings of the
Eleventh International Congress on Personal Construct Psychology, Barcelona, Spain, pp. 2-71.
Bell, R. C. (2003). "The Repertory Grid Technique," In F. Fransella (Ed.), International Handbook of Personal
Construct Psychology (pp. 95-103). Chichester, West Sussex, England: John Wiley and Sons.
Bell, R. C. (2004). "Predictive Relationships in Repertory Grid Data: A New Elaboration of Kelly's Organization
Corollary," Journal of Constructivist Psychology 17(4), pp. 281-295.
Bell, R. C., J. Vince, and J. Costigan. (2002). "Which Vary More in Repertory Grid Data: Constructs or Elements?"
Journal of Constructivist Psychology 15(4), pp. 305-314.
Bhatia, S. K. (1992, April). "Selection of Search Terms Based on User Profile," Paper presented at the 1992
ACM/SIGAPP Symposium on Applied Computing, Kansas City, Missouri, United States, pp. 224-233.
Bhatia, S. K. and J. S. Deogun. (1998). "Conceptual Clustering in Information Retrieval," IEEE Transactions on
Systems, Man & Cybernetics: Part B 28(3), pp. 427-436.
Bhatia, S. K. and Q. Yao. (1993). "A New Approach to Knowledge Acquisition by Repertory Grids," Paper presented
at the Proceedings of the second international conference on Information and knowledge management
Washington, D.C., United States pp. 738-740.
Bieri, J. (1955). "Cognitive Complexity-Simplicity and Predictive Behavior," Journal of Abnormal and Social
Psychology 51(1955), pp. 263-268.

Volume 23 Article 3
54
Boose, J. H. (1989, 3-6 Jan.). "Using Repertory Grid-Centered Knowledge Acquisition Tools for Decision Support,"
Paper presented at the Proceedings of the Twenty-Second Annual Hawaii International Conference on
System Sciences, Kailua-Kona, Hawaii, United States, pp. 211-220.
Boyle, T. A. (2005). "Improving Team Performance Using Repertory Grids," Team Performance Management
11(5/6), pp. 179.
Bradshaw, J. M., J. H. Boose, S. P. Covington, and P. J. Russo. (1987). "How to Do with Grids What People Say
You Can’t," Knowledge Acquisition: An International Journal of Knowledge Acquisition for Knowledge-Based
Systems 1(4).
Brown, S. M. (1992). "Cognitive Mapping and Repertory Grids for Qualitative Survey Research: Some Comparative
Observations," The Journal of Management Studies 29(3), pp. 287.
Browne, G. J., M. G. Pitts, and J. C. Wetherbe. (2007). "Cognitive Stopping Rules for Terminating Information
Search in Online Tasks," MIS Quarterly 31(1), pp. 89-104.
Byrd, T. A., K. L. Cossick, and R. W. Zmud. (1992). "A synthesis of research on requirements analysis and
knowledge acquisition techniques," Management Information Systems Quarterly 16(1), pp. 117-138.
Caputi, P. and P. Reddy. (1999). "A Comparison of Triadic and Dyadic Methods of Personal Construct Elicitation,"
Journal of Constructivist Psychology 12(3), pp. 253-264.
Carlsson, S. A. (1988). "A Longitudinal Study of Spreadsheet Program Use," Journal of Management Information
Systems 5(1), pp. 82-100.
Castro-Schez, J. J., J. L. Castro, and J. M. Zurita. (2004). "Fuzzy Repertory Table: A Method for Acquiring
Knowledge about Input Variables to Machine Learning Algorithm," IEEE Transactions on Fuzzy Systems
12(1), pp. 123-139.
Castro-Schez, J. J., N. R. Jennings, X. Luo, and N. R. Shadbolt. (2004). "Acquiring Domain Knowledge for
Negotiating Agents: A Case of Study," International Journal of Human-Computer Studies 61(1), pp. 3-31.
Castro-Schez, J. J., L. Jimeneza, J. Moreno, and L. Rodriguez. (2005). "Using Fuzzy Repertory Table-Based
Technique for Decision Support," Decision Support Systems 39(3), pp. 293-307.
Chao, C.-J., G. Salvendy, and N. J. Lightner. (1999). "Development of a Methodology for Optimizing Elicited
Knowledge," Behaviour and Information Technology 18(6), pp. 413-430.
Clases, C., R. Bachmann, and W. Theo. (2003). "Studying Trust in Virtual Organizations," International Studies of
Management and Organization 33(3), pp. 7-27.
Crudge, S. E. and C. Johnson. (2004). "Using the Information Seeker to Elicit Construct Models for Search Engine
Evaluation," Journal of the American Society for Information Science and Technology 55(9), pp. 794-806.
Davis, C. J. and E. M. Hufnagel. (2007). "Through the Eyes of Experts: A Socio-Cognitive Perspective on the
Automation of Fingerprint Work," Management Information Systems Quarterly 31(4), pp. 681-703.
Davis, G. B. (1982). "Strategies for Information Requirements Determination," IBM Systems Journal 21(1), pp. 4-30.
Díaz De Leó, E. and Guild, P. (2003), "Using repertory grid to identify intangibles in business plans," Venture Capital
5(2), pp. 135-160.
Dillon, A. and C. McKnight. (1990). "Towards a Classification of Text Types: A Repertory Grid Approach,"
International Journal of Man-Machine Studies 33(6), pp. 623-636.
Dinsdale, E. A. and D. M. Fenton. (2006). "Assessing Coral Reef Condition: Eliciting Community Meanings," Society
and Natural Resources 19(3), pp. 239-258.
Easterby-Smith, M., R. Thorpe, and D. Holman. (1996). "Using Repertory Grids in Management," Journal of
European Industrial Training 20(3), pp. 3-30.
Eden, C. and S. Jones. (1984). "Using Repertory Grids for Problem Construction," Journal of the Operational
Research Society 35(9), pp. 779-790.
Feixas, G., H. Geldschlager, and R. A. Neimeyer. (2002). "Content Analysis of Personal Constructs," Journal of
Constructivist Psychology 15(1), pp. 1-19.
Fransella, F., R. Bell, and D. Bannister. (2004). A Manual for Repertory Grid Technique (2nd ed.). Hoboken, New
Jersey, United States: Wiley.

Volume 23 Article 3
55
 Frewer, L. J., C. Howard, and R. Shepherd. (1998). "Understanding Public Attitudes to Technology," Journal of Risk
Research 1(3), pp. 221-235.
Gaines, B. R. and M. L. G. Shaw. (1981). "New Directions in the Analysis and Interactive Elicitation of Personal
Construct Systems," Recent Advances in Personal Construct Technology, pp. 147-182.
Gaines, B. R. and M. L. G. Shaw. (1986). "Induction of Inference Rules for Expert Systems," Fuzzy Sets and
Systems 18(1986), pp. 315-328.
Gaines, B. R. and M. L. G. Shaw. (1980). "New Directions in the Analysis and Interactive Elicitation of Personal
Construct Systems," International Journal of Man-Machine Studies 13(1980), pp. 81-116.
Gaines, B. R. and M. L. G. Shaw. (1993). "Knowledge Acquisition Tools Based on Personal Construct Psychology,"
The Knowledge Engineering Review 8(1), pp. 49-85.
Gallifa, J. And L. Botella. (2000). "The Structural Quadrants Method: A New Approach to the Assessment of
Construct System Complexity via the Repertory Grid," Journal of Constructivist Psychology 13(1), pp. 1-26.
Ginsberg, A. (1989). "Construing the Business Portfolio: A Cognitive Model of Diversification," Journal of
Management Studies 26(4), pp. 417-438.
Glaser, B. G. and A. L. Strauss. (1967). Discovery of Grounded Theory. New York, New York, United States: Aldine
de Gruyter.
Grudnitski, G. (1984). "Eliciting Decision-Makers' Information Requirements: Application of the Rep Test
Methodology," Journal of Management Information Systems 1(1), pp. 11-32.
Gudivada, V. N., V. V. Raghavan, and G. S. Seetharaman. (1994). "An Approach to Interactive Retrieval in Face
Image Databases Based on Semantic Attributes," Paper presented at the Third symposium on Document
Analysis and Information Retrieval, pp. 319-335.
Hagans, C. L., G. J. Neimeyer, and R. C. J. Goodholm. (2000). "The Effect of Elicitation Methods on Personal
Construct Differentiation and Valence," Journal of Constructivist Psychology 13(2), pp. 155-173.
Hallsworth, A. G. (1988). "Repertory Grid Methodology and the Analysis of Group Perceptions in Retailing,"
International Journal of Retailing 3(4), pp. 43-54.
Hassenzahl, M. and T. Trautmann. (2001). "Analysis of Web Sites with the Repertory Grid Technique," Paper
presented at the Conference on Human Factors in Computing Systems Seattle, Washington, United States,
pp. 167-168.
Hassenzahl, M. and R. Wessler. (2000). "Capturing Design Space from a User Perspective: The Repertory Grid
Technique Revisited," International Journal of Human-Computer Interaction 12(3/4), pp. 441-460.
Heninger, W. G., A. R. Dennis, and K. M. Hilmer. (2006). "Individual Cognition and Dual-Task Interference in Group
Support Systems," Information Systems Research 17(4), pp. 415-424.
Hevner, A. R., S. T. March, J. Park, and S. Ram. (2004). "Design Science in Information Systems Research," MIS
Quarterly 28(1), pp. 75-105.
Hudlicka, E. (1996). "Requirements Elicitation with Indirect Knowledge Elicitation Techniques: Comparison of Three
Methods," Paper presented at the International Conference on Requirements Engineering, Colorado Springs,
Colorado, United States, pp. 4-11.
Hunter, G. M. (1994). "'Excellent Systems Analysts: Key Audience Perceptions," Computer Personnel 15(1), pp. 15-
31.
Hunter, G. M. (1998). "Managing Information Systems Professionals: Implementing A Skill Assessment Process,"
Paper presented at the Proceedings of the 1998 ACM SIGCPR conference on Computer personnel research
Boston, Massachusetts, United States, pp. 19-27.
Hunter, M. G. (1997). "The Use of Repgrids to Gather Interview Data about Information Systems Analysts,"
Information Systems Journal 7(1), pp. 67-81.
Hunter, M. G. and J. E. Beck. (2000). "Using Repertory Grids to Conduct Cross-Cultural Information Systems
Research," Information Systems Research 11(1), pp. 93-101.
Hutchinson, J. R. (1998). "Using the Role Repertory Grid Technique for Item Generation in a Survey of Knowledge
Use," Journal of Constructivist Psychology 11(2), pp. 149-162.
Hwang, G. (1992). "Knowledge Elicitation and Integration from Multiple Experts," Paper presented at the
International Conference on Computing and Information, Toronto, Canada, pp. 208-211.

Volume 23 Article 3
56
Jung, G. S. and V. N. Gudivada. (1995). "Automatic Determination and Visualization of Relationships among
Symptoms for Building Medical Knowledge Bases," Paper presented at the Proceedings of the 1995 ACM
symposium on Applied computing Nashville, Tennessee, United States pp. 101-107.
Kanellis, P., M. Lycett, and R. J. Paul. (1999). "Evaluating Business Information Systems Fit: From Concept to
Practical Application," European Journal of Information Systems 8(1), pp. 65-76.
Kelly, G. (1955). The Psychology of Personal Constructs. New York, New York, United States: W.W. Norton and
Company Inc.
Kelly, G. (1963). A Theory of Personality: The Psychology of Personal Constructs. New York, New York, United
States: W.W. Norton.
King, M. and J. G. Phythian. (1992). "Validating an Expert Support System for tender Enquiry Evaluation: A Case
Study," Journal of the Operational Research Society 43(3), pp. 203-214.
Kreber, C., H. Castleden, N. Erfani, J. Lim, and T. Wright. (2003). "Exploring the Usefulness of Kelly's Personal
Construct Theory in Assessing Student Learning in Science Courses," Teaching in Higher Education 8(3), pp.
431-445.
Lambert, R., M. Kirksey, M. Hill-Carlson, and C. McCarthy. (1997). “The Repertory Grid as a Qualitative Interviewing
Technique for Use in Survey Development,” Annual Convention of the American Educational Research
Association (pp. 1-17). Chicago, Illinois, United States.
Landfield, A. W. and J. Cannell. (1988). "Ways of Assessing Functionally Independent Construction,
Meaningfulness, and Construction in Hierarchy," In J. Mancuso and L. G. Shaw (Eds.), Cognition and
personal structure. New York, New York, United States: Praeger.
Langan-Fox, J. and P. Tan. (1997). "Images of a Culture in Transition: Personal Constructs of Organizational
Stability and Change," Journal of Occupational and Organizational Psychology 70(3), pp. 273-293.
Latta, G. F. and K. Swigger. (1992). "Validation of the Repertory Grid for Use in Modeling Knowledge," Journal of the
American Society for Information Science 43(2), pp. 115-129.
Leach, C. (1980). "Direct Analysis of a Repertory Grid," International Journal of Man-Machine Studies 13(1980), pp.
151-166.
Liou, Y. I. (1990). "Knowledge Acquisition: Issues, Techniques, and Methodology," Paper presented at the 1990
ACM SIGBDP Conference on Trends and Directions in Expert Systems, Orlando, Florida, United States.
MacCormick, A., A. MacMillan, and B. Parry. (2004). "Identification of Criteria for the Prioritisation of Patients for
Elective General Surgery," Journal of Health Services Research and Policy 9(1), pp. 28-33.
Madigan, D., C. R. Chapman, J. Gavrin, O. Villumsen, and J. Boose. (1994). "Repertory Hypergrids: An Application
to Clinical Practice Guidelines," Paper presented at the Conference on Hypertext and Hypermedia, Edinburgh,
Scotland, pp. 117-125.
McKnight, C. (2000). "The Personal Construction of Information Space," Journal of the American Society for
Information Science 51(8), pp. 730-733.
Mickan, S. M. and S. A. Rodger. (2005). "Effective Health Care Teams: A Model of Six Characteristics Developed
from Shared Perceptions," Journal of Interprofessional Care 19(4), pp. 358-370.
Moynihan, T. (1996). "An Inventory of Personal Constructs for Information Systems Project Risk Researchers,"
Journal of Information Technology 11(4), pp. 359-371.
Neimeyer, G. J., J. Z. Bowman, and J. Saferstein. (2005). "The Effects of Elicitation Techniques on Repertory Grid
Outcomes: Difference, Opposite, and Contrast Methods," Journal of Constructivist Psychology 18(3), pp. 237 -
252.
Oppenheim, C., J. Stenson, and R. M. S. Wilson. (2003). "Studies on Information as an Asset II: Repertory Grid,"
Journal of Information Science 29(5), pp. 419-432.
Orlikowski, W. J. and D. C. Gash. (1994). "Technological Frames - Making Sense of Information Technology in
Organizations," ACM Transactions on Information Systems 12(2), pp. 174-207.
Phythian, J. G. and M. King. (1992). "Developing an Expert Support System for Tender Enquiry Evaluation: A Case
Study," European Journal of Operational Research 56(1), pp. 15-29.
Pill, A. (2005). "Models of Professional Development in the Education and Practice of New Teachers in Higher
Education," Teaching in Higher Education 10(2), pp. 175-188.

Volume 23 Article 3
57
 Poulymenakou, A., T. Cornford, and E. A. Whitley. (1990). "Knowledge Acquisition to Facilitate Organizational
Problem Solving," Paper presented at the Proceedings of the 1990 ACM SIGBDP conference on Trends and
directions in expert systems, Orlando, Florida, United States, pp. 181-188.
Reger, R. K. and A. S. Huff. (1993). "Strategic Groups: A Cognitive Perspective," Strategic Management Journal
14(2), pp. 103-124.
Reich, B. H. and I. Benbasat. (2000). "Factors that Influence the Social Dimension of Alignment between Business
and Information Technology Objectives," MIS Quarterly 24(1), pp. 81-113.
Rentsch, J. R. (1990). "Climate and Culture: Interaction and Qualitative Differences in Organizational Meanings,"
Journal of Applied Psychology 75(6), pp. 668-681.
Reynolds, T. J. and J. Gutman. (1984). "Laddering: Extending the Repertory Grid Methodology to Construct
Attribute-Consequence-Value Hierarchies," In R. E. Pitts and A. G. Woodside (Eds.), Personal Values and
Consumer Psychology (pp. 155-167). London: Lexington Books.
Reynolds, T. T. and J. Gutman. (1988). "Laddering Theory, Method, Analysis and Interpretation," Journal of
Advertising Research 28(1), pp. 11-31.
Rugg, G., M. Eva, A. Mahmood, N. Rehman, S. Andrews, and S. Davies. (2002). "Eliciting Information about
Organizational Culture via Laddering," Information Systems Journal 12(3), pp. 215-229.
Rugg, G., P. McGeorge, and N. Maiden. (2000). "Method Fragments," Expert Systems 17(5), pp. 248-257.
Scheubrein, R. and S. Zionts. (2006). "A Problem Structuring Front End for a Multiple Criteria Decision Support
System," Computers & Operations Research 33(1), pp. 18-31.
Shadbolt, N. and N. Milton. (1999). "From Knowledge Engineering to Knowledge Management," British Journal of
Management 10(1999), pp. 309-322.
Shaw, M. L. G. (1980). On Becoming a Personal Scientist: Interactive Computer Elicitation of Personal Models of the
World. London; New York: Academic Press.
Shaw, M. L. G. and B. R. Gaines. (1989). "Comparing Conceptual Structures: Consensus, Conflict,
Correspondence, and Contrast," Knowledge Acquisition 1(1989) pp. 341-363.
Shaw, M. L. G. and B. R. Gaines. (1996). "Webgrid: Knowledge Elicitation and Modeling on the Web," Paper
presented at the WebNet 96 - World Conference of the Web Society, San Francisco, California, United States.
Shaw, M. L. G. and B. R. Gaines. (1998). "WebGrid-II: Developing Hierarchical Knowledge Structures from Flat
Grids," Paper presented at the Proceedings of the 11th Knowledge Acquisition Workshop.
Smithson, M. (1987). Fuzzy Set Analysis for Behavioral and Social Sciences. New York, New York, United States:
Springer-Verlag.
Stewart, J., R. Bregazzi, and C. O'Halloran. (2003). "Accident and Emergency Medicine Specialist Registrars'
Perceptions of the 'Essential' Aspects of a Training Placement," Learning in Health and Social Care 2(1), pp.
28-36.
Stewart, V., A. Stewart, and N. Fonda. (1981). Business Applications of Repertory Grid. Great Britain: McGraw-Hill.
Tan, F. B. and B. Gallupe. (2006). "Aligning Business and Information Systems Thinking: A Cognitive Approach,"
IEEE Transactions on Engineering Management 53(2), pp. 223-237.
Tan, F. B. and M. G. Hunter. (2002). "The Repertory Grid Technique: A Method for the Study of Cognition in
Information Systems," Management Information Systems Quarterly 26(1), pp. 39-57.
Tan, F. B. and L. L. Tung. (2003, Dec 12-13). "Exploring Web Site Evaluation Criteria Using the Repertory Grid
Technique: A Web Designers' Perspective," Paper presented at the Second Annual Workshop on HCI
Research in MIS, Seattle, Washington, United States.
Van der Kloot, W. A. (1981). "Multidimensional Scaling of Repertory Grid Responses: Two Applications of
HOMALS," In Recent Advances in the Theory and Practice of Personal Construct Psychology. London:
Macmillan.
Verlinden, J. C. and M. J. J. Coenders. (2000). "Qualitative Usability Measurement of Web Sites by Employing the
Repertory Grid Technique," Paper presented at the Conference on Human Factors in Computing Systems,
The Hague, The Netherlands.
Walsh, J. P. (1995). "Managerial and Organizational Cognition: Notes from a Trip Down Memory Lane,"
Organization Science 6(3), pp. 280-321.

Volume 23 Article 3
58
Wessler, R. and M. Ortlieb. (2002). "A User Centred Approach to Measure a Web Site's Appeal Using the Repertory
Grid Technique," Paper presented at the Proceedings of the 1st European Usability Professionals Association
Conference, London, pp. 70-78.
Whyte, G. and A. Bytheway. (1996). "Factors Affecting Information Systems' Success," International Journal of
Service Industry Management 7(1), pp. 74-93.
Wright, R. P. (2002). "Perceptual Dimensions of Performance Management Systems in the Eyes of Different Sample
Categories," International Journal of Management 19(2), pp. 184.
Wright, R. P. (2004a). "Mapping Cognitions to Better Understand Attitudinal and Behavioral Responses in Appraisal
Research," Journal of Organizational Behavior 25(3), pp. 339-374.
Wright, R. P. (2004b). "Top Managers' Strategic Cognitions of the Strategy Making Process: Differences between
High and Low Performing Firms," Journal of General Management 30(1), pp. 61-78.
Xiangmin, Z. and M. Chignell. (2001). "Assessment of the Effects of User Characteristics on Mental Models of
Information Retrieval Systems," Journal of the American Society for Information Science and Technology
52(6), pp. 445-459.
Yorke, D. (1983). "Straight or Bent? An Inquiry into Rating Scales in Repertory Grids," British Educational Research
Journal 9(2), pp. 141-151.
Young, M. S., H. M. Edwards, S. McDonald, and B. Thompson. (2005). "Personality Characteristics in an XP Team:
A Repertory Grid Study," Paper presented at the Proceedings of the 2005 Workshop on Human and Social
Factors of Software Engineering St. Louis, Missouri, United States, pp. 1-7.

ABOUT THE AUTHORS

Aaron M. Cutis is an information systems Ph.D. candidate at the Kelley School of Business at Indiana University.
He graduated with a Master of Information Systems Management degree from the Marriott School at Brigham Young
University, where he was also enrolled in the Information Systems Ph.D. Preparation Program. His research
interests include technology-mediated collaboration, human-computer interaction, knowledge management, and
health care information systems. He has articles published in the Journal of the Association for Information Systems;
IEEE Transactions on Professional Communication; and Journal of Business Communication.

Taylor M. Wells is a Ph.D. student in the Kelley School of Business at Indiana University. He graduated from the
Marriott School at Brigham Young University with a Master of Information Systems Management degree, where he
was also enrolled in the Information Systems Ph.D. Preparation Program. His research interests include technology-
mediated collaboration, e-commerce, knowledge management, and human-computer interaction.

Paul Benjamin Lowry is an assistant professor of Information Systems at the Marriott School, Brigham Young
University and a Kevin and Debra Rollins Faculty Fellow. His research interests include Human-Computer
Interaction (HCI) (collaboration, culture, communication, adoption, entertainment), e-business (privacy, security,
trust, branding, electronic markets), and scientometrics of Information Systems research. He received his Ph.D. in
Management Information Systems (MIS) from the University of Arizona. He has articles published in the Journal of
Management Information Systems; Journal of the Association for Information Systems; Communications of the
ACM; Communications of the Association for Information Systems; Decision Support Systems; IEEE Transactions
on Systems, Man, and Cybernetics; IEEE Transactions on Professional Communication; Small Group Research;
Information Sciences; Journal of Business Communication; and others. He serves as an associate editor at
Communications of the Association for Information Systems.

Trevor Higbee is a software developer for ExxonMobil in Houston, Texas. He graduated with a Master of
Information Systems Management degree from the Marriott School at Brigham Young, where he was also enrolled
in the Information Systems PhD Preparation Program.

Volume 23 Article 3
59
 APPENDIX I. PCT: POSTULATE AND CORROLARIES
The table below explores in detail the main tenants of PCT—its fundamental postulate and selected supporting
corollaries.

Name Kelly’s Exact Wording Explanation Example Implication for

[Kelly, 1955] RepGrid/Application
Fundamental The ways in which a A person’s behavior is n/a n/a
Postulate person anticipates driven by the way in which
events psychologically he/she anticipates events
channels that person’s or objects.
processes.
Construction A person anticipates A person anticipates A meeting is an event A repgrid represents the
Corollary events by construing events and other objects that one might specified domain as it is
their replications. in the world by interpreting anticipate. No two in the participant’s mind,
the repetitions of past meetings are the same, based on experience, not
events. yet past meetings help as it could, should, or will
us understand the be.
general attributes
(constructs) that define
future meetings.
Individuality Persons differ from Persons differ from each An information source Each repgrid participant
Corollary each other in their other in their interpretation might have meaning for could use different
construction of events. of events or objects. Two one employee that is constructs or different
people may view the not shared by other labels for the same
same thing in different employees. To one constructs on the same
ways, because they employee, the source elements.
construe (interpret) the could be helpful and
event differently (i.e., they succinct, and to another Constructs are best
use different constructs to it could be confusing. elicited by the user.
identify it).
Grids should be
administered individually
Organization Each person To conveniently order and Web mail and stand- Laddering can be useful
Corollary characteristically anticipate events, a alone mail could be in the exploration of
evolves, for their person creates a subordinate constructs construct hierarchies.
convenience in construction system over of a higher email
anticipating events, a time. Constructs in this system construct, which If participants verbalize
construction system system can have a along with instant their thought process
embracing ordinal hierarchal relationship. messaging may be while discriminating
relationships between subordinate constructs between elements, the
constructs. of the higher researcher can make
communication systems notes of hierarchies of
construct, etc. All of elements.
these constructs and
their relationships
evolve over time and
make up a construct
system.
Dichotomy A person’s construction A person’s construction There are only so many Constructs must be
Corollary system is composed of system is composed of a constructs that can be bipolar.
a finite number of finite number of bipolar used to distinguish
dichotomous constructs. constructs. Constructs are events email programs.
bipolar because nothing in If speed were an
context is affirmed without attribute of email
denying something else. programs, its
dichotomous corollary
might be the construct
fast-slow.
Range A construct is A construct is useful for The construct of Rated elements should
Corollary convenient for the the anticipation of a finite threaded all belong to the same
anticipation of a finite range of events only. The conversations– range of convenience. If
range of events only. finite range is dictated by individual emails is the range is too large,
the organization and useful only (convenient) the constructs become
permeability of the if it is used in the nonsensical.
construct. context of email

Volume 23 Article 3
60
 Name Kelly’s Exact Wording Explanation Example Implication for
[Kelly, 1955] RepGrid/Application
systems. This construct
does not make much
sense when applied to
other elements such as
word processing
software. The range of
threaded
conversations–
individual emails is
narrowly limited.
Experience A person’s construction People modify their The initial introduction Repgrid is a good
Corollary system varies as they construct systems with of a new type of email measure for longitudinal
successively construe experience. As people system might meet with studies to find changes in
the replication of learn, their construct resistance initially, until perceptions.
events. system changes to align a person’s construct
itself with what they system makes room for
interpret to be objective new constructs that are
reality. used to identify and
evaluate the new
system.
Modulation The variation in a Permeability refers to a The range of Some constructs may be
Corollary person’s construction construct’s ability to convenience for the applied to different
system is limited by the accept other constructs construct threaded contexts to gain
permeability of the into its range of conversations– additional insight, but
constructs within whose convenience. As new individual emails others will be limited to a
range of convenience variations in a person’s encompasses email single context.
the variants. world arise, the variation programs, but does not
must be defined by extend to much else.
already-available However, the construct
constructs. The less fast–slow can certainly
permeable a person’s be applied to many
construct system is, the other areas, and its
more it varies. range of convenience is
therefore larger. If we
encounter a new
software program
(variant), it may lie
within the range of
convenience of
permeable constructs
like fast–slow but not
within the range of
convenience of the less
permeable constructs.
The less permeable a
person’s construct
system is, the more it
varies.
Fragmentation A person may Construct subsystems that A person is entertained These help the person
Corollary successively employ a seem incompatible with by violence in a movie, transcend contradictions
variety of construction each other at first can be and then be deeply in their logic.
subsystems, which are compatible if higher-level disturbed by violence
inferentially construct subsystems are toward his own person.
incompatible with each taken into consideration. The two concepts of
other. violence may be
inferentially
incompatible unless
consideration is taken of
the higher constructs of
entertainment and
reality.
Commonality To the extent that one To the extent that two Two people who both If two people share the
Corollary person employs a people use the same view the payroll process same constructs (note:
construction of constructs to distinguish as the same set of constructs differ from
experience that is between events, their functions to attain a set labels), their views can
similar to that employed psychological processes of goals will, in regards be compared.

Volume 23 Article 3
61
 Name Kelly’s Exact Wording Explanation Example Implication for
[Kelly, 1955] RepGrid/Application
by another, their and behavior will also be to these constructs,
processes are similar for that event. behave similarly toward
psychologically similar the process. The
to those of the other reverse is also true.
person.
Sociality To the extent that one A person can relate to Employee A may not Sharing grids in a group
Corollary person construes the others without employing view the payroll system setting and resolving
construction process of their same construct in the same way as differences between
another, they may play system. The Employee B, but to the constructs can lead to
a role in a social understanding (construal) extent to which mutual understanding
process involving the of others’ construct Employee A can and greater collaboration
other person. systems allows a person understand Employee between group
to be involved with and B’s construction of the members.
relate to them. system, the Employee A
can relate to Employee
B on the construct of
the system.

APPENDIX II. REPGRID SOFTWARE FOR ELICITATION AND ANALYSIS

Package Platform Cost Web site

Name
FlexiGrid CA $300 finn.tschudi@psykologi.uio.no
Chris Evans’ Win Free (SAS Required) http://www.psyctc.org/grids/default.htm
SAS Analysis
Program
Rep IV WinXP/OSX Personal: FREE http://repgrid.com/RepIV/
Research: US $700
OmniGrid DOS/MAC Free http://www.psyctc.org/grids/omnigrid.htm
GridStat DOS Free http://www.repgrid.unimelb.edu.au/grids.htm
GridCor WinXP Individual: €150 http://www.terapiacognitiva.net/record/
Institution: €240
IdioGrid Win95-XP Free http://www.idiogrid.com/
GridScal DOS Free http://www.repgrid.unimelb.edu.au/grids.htm
EnquireWithin Win US $49 http://www.enquirewithin.co.nz/
GridSuite Win/Mac Individual: €15–460 http://www.uni-
(Java) Site: €920 stuttgart.de/pae/gridsuite/index.php
WebGrid III All (Internet) Free http://tiger.cpsc.ucalgary.ca/

The latest information on these packages can be found at http://www.pcp-net.de/info/comp-prog.html.

Prices and versions change often.

Copyright © 2008 by the Association for Information Systems. Permission to make digital or hard copies of all or part
of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for
profit or commercial advantage and that copies bear this notice and full citation on the first page. Copyright for
components of this work owned by others than the Association for Information Systems must be honored.
Abstracting with credit is permitted. To copy otherwise, to republish, to post on servers, or to redistribute to lists
requires prior specific permission and/or fee. Request permission to publish from: AIS Administrative Office, P.O.
Box 2712 Atlanta, GA, 30301-2712 Attn: Reprints or via e-mail from ais@aisnet.org

Volume 23 Article 3
62
.
ISSN: 1529-3181
EDITOR-IN-CHIEF
Joey F. George
Florida State University
AIS SENIOR EDITORIAL BOARD
Guy Fitzgerald Joey F. George Kalle Lyytinen
Vice President Publications Editor, CAIS Editor, JAIS
Brunel University Florida State University Case Western Reserve University
Edward A. Stohr Blake Ives Paul Gray
Editor-at-Large Editor, Electronic Publications Founding Editor, CAIS
Stevens Inst. of Technology University of Houston Claremont Graduate University
CAIS ADVISORY BOARD
Gordon Davis Ken Kraemer M. Lynne Markus Richard Mason
University of Minnesota Univ. of Calif. at Irvine Bentley College Southern Methodist Univ.
Jay Nunamaker Henk Sol Ralph Sprague Hugh J. Watson
University of Arizona University of Groningen University of Hawaii University of Georgia
CAIS SENIOR EDITORS
Steve Alter Jane Fedorowicz Chris Holland Jerry Luftman
U. of San Francisco Bentley College Manchester Bus. School Stevens Inst. of Tech.
CAIS EDITORIAL BOARD
Michel Avital Dinesh Batra Ashley Bush Erran Carmel
Univ of Amsterdam Florida International U. Florida State Univ. American University
Fred Davis Gurpreet Dhillon Evan Duggan Ali Farhoomand
Uof Arkansas, Fayetteville Virginia Commonwealth U Univ of the West Indies University of Hong Kong
Robert L. Glass Sy Goodman Mary Granger Ake Gronlund
Computing Trends Ga. Inst. of Technology George Washington U. University of Umea
Ruth Guthrie Juhani Iivari K.D. Joshi Chuck Kacmar
California State Univ. Univ. of Oulu Washington St Univ. University of Alabama
Michel Kalika Claudia Loebbecke Paul Benjamin Lowry Sal March
U. of Paris Dauphine University of Cologne Brigham Young Univ. Vanderbilt University
Don McCubbrey Fred Niederman Shan Ling Pan Kelly Rainer
University of Denver St. Louis University Natl. U. of Singapore Auburn University
Paul Tallon Thompson Teo Craig Tyran Chelley Vician
Loyola College in Maryland Natl. U. of Singapore W Washington Univ. Michigan Tech Univ.
Rolf Wigand Vance Wilson Peter Wolcott Ping Zhang
U. Arkansas, Little Rock University of Toledo U. of Nebraska-Omaha Syracuse University
DEPARTMENTS
Global Diffusion of the Internet. Information Technology and Systems.
Editors: Peter Wolcott and Sy Goodman Editors: Sal March and Dinesh Batra
Papers in French Information Systems and Healthcare
Editor: Michel Kalika Editor: Vance Wilson
ADMINISTRATIVE PERSONNEL
James P. Tinsley Robert Hooker Copyediting by Carlisle
AIS Executive Director CAIS Managing Editor Publishing Services
Florida State Univ.

Volume 23 Article 3