Relating Cognitive Problem-Solving Style to User Resistance

INTRODUCTION

This chapter explores cognitive problem-solving style and its impact on user resistance, based on the premise that the greater the cognitive difference (cognitive gap) between users and developers, the greater the user resistance is likely to be. Mullany (1989, 2003) conducted an empirical study demonstrating this. This study contradicts the findings of Huber (1983) and supports Carey (1991) in her conclusion that cognitive style theory, as applied to IS, should not be abandoned. Mullany’s findings, in fact, are the opposite. Kirton (1999, 2004) supported Mullany’s results. In particular, Mullany made use of Kirton’s (2004) adaption-innovation theory. The emergent instrument, called the Kirton adaption-innovation inventory (KAI; Kirton, 1999, 2004), was used by Mullany as his measure of cognitive style.
Mullany’s study also investigated the relationship between user resistance and user ages and lengths of service in the organisation. It failed to show any relationship between these factors and user resistance. This countermands the findings of Bruwer (1984) and dismisses any intimation that older or longer-serving employees are necessarily more resistant to change as myths.

BACKGROUND

Ever since the early 1980s, experts have identified user resistance to new systems as an expensive time overhead (see studies by Hirschheim & Newman, 1988, and Markus, 1983). Some authors suggest the greater importance of age and length of service. Bruwer (1984), for instance, claimed to have demonstrated that the older or longer-serving an employee, the more resistant he or she is likely to be to a new computer system. Clarification of issues surrounding user resistance has also highlighted cognitive style theory as potentially important, but to date, its impacts have only been sparsely researched in relation to user resistance, many of the prior studies being open to question. This research, on the other hand, proposes that a system will fail when the developer and user differ significantly in their problem-solving approaches. To reduce user resistance, it thus makes sense to recommend system designs that suit the user’s approach to problem solving.
This issue appears only to have been studied empirically by Mullany (1989, 2003). He formulated the research question, “Is there a relationship between user resistance to a given information system and the difference in cognitive style between the user and the developer?” With the aid of his own instrument for measuring user resistance and the Kirton adaption-innovation instrument (Kirton, 1999) to measure the cognitive styles of users and associated system developers, he found a highly significant relationship between developer-user cognitive style differences and the level of user resistance to systems.
Why no other studies along similar lines have been reported in credible current research is difficult to explain. One possibility is that the literature contains speculative studies, such as that by Huber (1983), that discredit cognitive-style theory as a tool in understanding system success. Other studies, such as that by Carey (1991), while encouraging the continued use of cognitive-style theory in studying system phenomena, do not demonstrate its predictive success in information systems (IS). The remainder of this chapter thus examines the meaning and measure of cognitive style, the measure of user resistance, the specific findings of Mullany (1989, 2003), and outlooks for the future in this area of research.

THE MEANING AND MEASURE OF COGNITIVE PROBLEM-SOLVING STYLE

Liu and Ginther (1999) defined cognitive style as, “An individual’s consistent and characteristic predispositions of perceiving, remembering, organizing, processing, thinking and problem-solving.” Schroder, Driver, and Streufert (1967), in a discussion of human information processing, suggested that organisms “either inherit or develop characteristic modes of thinking, adapting or responding and go on to focus upon adaptation in terms of information processing.” In short, an individual exhibits characteristic ways of processing information (and, hence, solving problems), known as his or her “cognitive style.” Table 1 gives an historic summary of key experts over the years who have endeavoured to name and measure the construct of cognitive style. Of these, the MBTI (Myers-Briggs type indicator) is the most used in current, credible research literature, followed by the KAI (Kirton, 1976, 1984). As previously stated, the only evident effort made to relate cognitive style to user resistance was carried out by Mullany (1989) using the KAI. The reason for his preferred use of the KAI stemmed from its ability to provide a near-continuous, bipolar scale, convenient for finding correlations and associations. The MBTI, by contrast, yields only certain cognitive classifications, where no mutual order is evident. The correlation with other factors would then have been more difficult to show statistically.

Table 1. Cognitive-style constructs: Key studies.

Turning to the theory behind the KAI, Kirton (1999) identified two extremes of cognitive style; namely, the adaptor and the innovator. The adaptor tends to follow traditional methods of problem solving, while the innovator seeks new, often unexpected, and frequently less-accepted methods. The adaptor tends to “do well” within a given paradigm, where the innovator tends to “do differently,” thus transcending accepted paradigms. The adapter is prepared to be wedded to systems, solving problems “in the right way,” but is often seen as “stuck in a groove.” The innovator has little regard for traditions, is often seen as creating dissonance, and elicits comments such as, “He wants to do it his own way, not the ‘right’ way.” All humans, Kirton proposed, can be located on a continuum between the extremes of these two cognitive styles.
Both cognitive extremes can be highly creative, can resist change, and can act as agents for change. Adaptors support changes to the conservative, back to the “good old ways,” and resist changes to novel methodologies. Innovators support changes toward unprecedented systems and technologies and resist changes to the traditional.
Kirton’s instrument, the KAI, has been widely demonstrated to be a successful measure of his construct of cognitive problem-solving style. The instrument takes the form of a questionnaire, on which the respondent has to rate himself or herself against 33 character traits. KAI scores can range from 32 to 160, with a mean of 96 and a standard deviation of about 16. A person scoring above the mean of 96 is considered to be an innovator; conversely, a person scoring below 96 is rated as an adaptor. However, in the range of 80 to 112 (that is, within one standard deviation of the mean), a third cognitive style can be identified—the mid-scorer. Such persons tend to have human rather than technical problem-solving preferences and can relate better to the extreme scorers than either can to the other.

A DESCRIPTION AND MEASURE OF USER RESISTANCE

Mullany (1989) measured user resistance at personal interviews with the key user of each system selected for investigation. The user was asked to list the problems that he or she recalled had occurred during the system’s development and implementation. They were asked, in effect, to make complaints, in confidence, against the system and its manner of implementation. Then they were requested to rate the severity of each complaint on a seven-point scale (with seven representing the most severe weighting). The sum of severities of all the complaints measured the respondent’s resistance score or R-score. Obvious criticisms of the R-score method are as follows:
1. It may be highly influenced by the cognitive style of the interviewer.
2. At an interview, the user might forget certain crucial problems that had been experienced.
Mullany refuted (1) on the grounds that the same person (himself) did all the interviewing in his study. He assumed (2) to be of limited impact, because the object of the R-score method is to observe the user in the process of complaining. Consequently, the resistant user is capable of exaggerating or even inventing complaints, making the issue of those forgetten less relevant. However, he conceded the limitation that there are covert forms of resistance, such as absenteeism and withdrawal, that are not necessarily related to overt complaints.
To investigate a relationship between cognitive-style differences and user resistance, Mullany (1989) set out to collect data from a suitable sample of computer system developers and users. Bivariate data were to be collected, namely, the analyst-user KAI difference and the R-score for each user. The association between these was then to be measured. For his association measure, he used both the Kendall-t and the more traditional Spearman-r, which are equally reliable for significance testing (Liebetrau & Kendall, 1970). According to Kendall (1970; who invented the Kendall-t measure of association) and as confirmed by Liebetrau (1983), sample sizes of10 to 20 are sufficient for such tests. The author thus selected a much larger sample size of 34 systems in 10 South African organizations. However, the following further criticisms were identified and addressed:
1. The sample size is small compared with some other studies in IS.
2. A user who champions a system may point out deficiencies in the hopes of improving that system.
Referring to the first of these criticisms, one should be alerted to the fact that sample representivity is more important than size in obtaining reliable results. In fact, the larger the sample, the less the researcher is likely to be able to guarantee a lack of significant bias. For example, suppose that with the aid of a suitable instrument, one sets out to measure the diligence of some human population. If a large sample size is sought through a postal, Web-based, or e-mail survey (the only practicable methods for really large samples), only the most diligent respondents are likely to respond, giving a bias to the more diligent and, thus, casting serious doubt on the results. To reduce this effect, Mullany (1989) collected all the data at personal interviews with the analysts and users. Furthermore, organisations he approached were requested to provide a fair spread of systems in use. He thus used legitimate power lent to him by the organisations to interview those as he required.
The second criticism was addressed by obtaining approval to keep all employees’ responses confidential and to make this clear at each interview. This meant that a user would be unlikely to complain to Mullany in the hopes of achieving some system improvement, as he or she knew that no information would be relayed to the rest of the organisation. Every effort was made to preserve standard interviewing conditions: these being freedom from pressure or interruption and complete assurance of confidentiality. In short, interviewing conditions similar to those of face-to-face counselling were achieved. Furthermore this technique of measuring user resistance has been confirmed by respected researchers. First, both Markus (1983) and Hersheim et al. (1988) identified complaint as an overt symptom of, if not even a form of, resistance. Kirton (2004), in a discussion of Mullany’s study, confirmed the technique as valid.

THE RELATIONSHIP BETWEEN USER RESISTANCE AND THE DIFFERENCES IN COGNITIVE STYLES BETWEEN THE USER AND THE DEVELOPER

The key developer and key user of each were interviewed. In each case, measures were obtained for the developer KAI score, user KAI score, and user R-score. At the same time, demographic data were collected; most particularly, the ages and lengths of service of the respondents, in order to test the findings of Bruwer (1984). A relationship as an association was found for the user R-scores versus the absolute differences between developer and user KAI scores. The association (with p < 0.005) proved to be strong, suggesting that user resistance can be minimized by matching a user with a developer of similar cognitive style. However, no significant associations were found between the ages and lengths of service of users and their R-scores, in contradiction of Bruwer’s (1984) results. Rosen and Jerdee’s (1976) study, which sought a similar result for occupational groups in general, agrees with Mullany’s findings in this respect.
An interpretation of the R-score was demonstrated based on a near-perfect direct proportion that proved to exist between the weighted and nonweighted numbers of the users’ complaints. In this relationship, the constant of proportionality was found to be 3.913 (that is, nearly 4). The R-score can thus be described as approximately four times the number of complaints a user will make retrospectively, in private, concerning a system and its manner of implementation.

FUTURE TRENDS

This study reignites the issue of cognitive style as an important issue in IS and completely countermands the conclusions of Huber (1983). It substantially strengthens the case made by Carey (1991) that cognitive-style issues in IS research should not be abandoned. Further, it suggests that user resistance and the related constructs of user dissatisfaction and system success can be predicted from cognitive-style measures (that is, KAI scores) prior to system development.
Areas for further research centre upon the main limitation of this study. For instance, there is little known regarding how the developer-user cognitive gap influences the system development life cycle (SDLC) over a significant passage of time, and neither this study nor any other found in the literature has achieved this. In fact, the literature is devoid of any attempts to conduct such research. A longitudinal study where SDLC curves are compared with the developer-user cognitive gap would be of immense importance and interest. New rules for system development based on cognitive-style testing would be expected to emerge.

CONCLUSION

It is clear that cognitive problem-solving style, as defined by Kirton and measured using the KAI, impacts user resistance. The greater the cognitive gap between users and developers, the greater the user resistance is likely to be. This contradicts the findings of Huber (1983) and supports Carey (1991) in her conclusion that cognitive-style theory, as applied to IS, should not be abandoned. Mullany’s findings, in fact, are the opposite.
The failure to show any relationship between users’ ages and lengths of service, and their resistance ratings, countermand the findings of Bruwer (1984) and suggest that organisations should be alerted to the danger of discriminating against older or longer-serving users or dispensing with their services on such grounds.
As mentioned above, areas for further research centre upon the main limitation of this study. A longitudinal study where SDLC curves are compared with the developer-user cognitive gap would be of great importance and interest.

KEY TERMS

Adaptor: An adaptor tends to follow traditional methods of problem solving, tending to “do well.” He or she is often seen as “stuck in a groove” (Kirton, 1999).
Association: A relationship between two statistical variables. Unlike a correlation, an association does not yield a quantitative result but is contingent upon the ranking of the bivariate data values only.
Cognitive Gap: The difference in cognitive problem-solving style between two people, especially two people who are obliged to interact as members of a group or team.
Cognitive Problem-Solving Style: The position an individual occupies between two extremes of cognitive problem-solving style personality; namely, the adaptor and the innovator.
Cognitive Style: An individual exhibits characteristic ways of processing information and, hence, solving problems, known as his or her “cognitive style.”
Innovator: The innovator seeks new, often unexpected, and frequently less acceptable methods. He or she has little regard for traditions, is often seen as creating dissonance, and elicits comments such as, “He wants to do it his own way, not the ‘right’ way” (Kirton, 1999).
KAI (Kirton Adaption-Innovation Inventory): An instrument that measures cognitive problem-solving style. It takes the form of a questionnaire, on which the respondent is asked to rate himself or herself against 32 character traits.
R-Score (Resistance Score): A method of measuring user resistance where, at personal interviews with the key user of a given system, the user is asked to list system problems and then to rate the severity of each on a seven-point scale. The sum of severities of all the complaints measures his or her R-score (Mullany, 1989).
User Resistance: Any user behaviour, action, or lack of action that inhibits the development, installation, or use of an information system.