THE MYERS-BRIGGS PERSONALITY TYPE AND ITS RELATIONSHIP TO COMPUTER PROGRAMMING

Abstract

 

This research investigated whether college students with certain personality types (as measured by the Myers-Briggs Type Indicator) performed better in an introductory programming class than students with the opposite personality types. A pilot study of 24 students and a follow-up study of 114 college students taking an introductory programming class indicated that sensing students performed better on programming assignments than intuitive students, and that judging students achieved higher programming averages than perceptive students. Additionally, it was found that personality type influenced achievement in programming performance, but did not influence achievement on exams or overall average. (Keywords: personality, programming.)

Within the field of computer science there is some agreement that there are tremendous individual differences in students' achievement in programming. Teachers of computer programming witness first-hand the huge variability among students in learning and achievement. Schneiderman (1980) reported that researchers have found differences in programming performance as high as 100 to 1 among programmers of similar background, and Brooks (1980) reported that one of the problems in conducting studies on introductory programming classes is the tremendous variability in students' achievement.

Personality traits and cognitive styles may help explain some of the variability among both student and professional programmers. This study investigated whether students of certain personality styles were more successful in writing computer programs than students with the opposite styles. A secondary question investigated in the study was whether students of certain personality types were more likely to drop the course.

One of the most common measures of personality is the Myers-Briggs Type Indicator (MBTI), which measures an individual's preference on four dimensions. Several of the dimensions, such as whether a student is oriented toward thinking or feeling, seem especially relevant to computer programming. For instance, one would expect a student who has a logical-objective (thinking) orientation to perform better in writing computer programs than a student who has a feeling orientation.

This article is organized in three parts. The first section introduces the MBTI. The second reviews the literature combining MBTI and problem solving and outlines the conceptual framework for the study. The last section describes the current study and the implications of the findings.

THE MYERS-BRIGGS TYPE INDICATOR

In his classic book Psychological Types, the Swiss psychiatrist Carl Jung (1923) described in detail the systematic ways in which people differ. Jung believed that conscious mental activities could be placed in one of four categories: sensing, intuitive, thinking, and feeling. He provided a model that helps us understand the different ways people perceive information and make judgments.

The MBTI, developed in the early 1950s by Katherine Cook Briggs and Isabell Briggs Myers, was designed to make Jung's theory more explicit and practical in its application to people's everyday lives. Since its publication in 1955, the MBTI has been increasingly used in the education, counseling, business, government, and religious communities (McCaulley, 1987). Campbell and Davis (1988) reported that more than 1,100 dissertations, theses, books, and journal articles have been published on the MBTI. The MBTI is currently the most widely used inventory of psychological types in the world (Hirsh & Kummerow, 1989).

Dimensions and Poles

The MBTI measures preferences on four dimensions, which have been described by Hirsh and Kummerow (1989). They classify the four dimensions as:

1. Energizing: How and where you get your energy.

 

2. Attending: What you pay attention to when you gather information.

 

3. Deciding: What system you use when you make decisions.

 

4. Living: What type of life you will adopt.

The second and third categories refer to the mental powers or cognitive dimensions and are often considered the two most important dimensions. The first and fourth categories refer to attitudes. They describe where we gain our energy and how we deal with the outside world.

Each dimension has two poles. The first dimension, Energizing, refers to a person's orientation toward the world. The two poles of this dimension are introversion and extroversion. Extroversion describes an attitude in which attention is drawn our toward objects and people: Extroverts tend to draw energy from the external world of people and things. They prefer to communicate by talking and they often process information verbally. Introversion describes an attitude in which attention is drawn toward the inner world of ideas. Introverts tend to draw energy from the internal world of ideas, emotions, and impressions. They tend to process information inside their heads. Whereas extroverts often act without thinking, introverts often think without acting.

The second dimension, Attending, refers to how a person perceives information. The two poles of this category are sensing and intuitive. A sensing person tends to perceive observable facts through the five senses. An intuitive person perceives information based on the meanings, relationships, or possibilities beyond the information gathered through his or her senses. Sensing persons are often described as being more practical whereas intuitive individuals are described as being more innovative. Keirsey and Bates (1984) reported that the perceiving dimension is the source of the most "miscommunication, misunderstanding, vilification, defamation, and denigration" (p. 17). As will be seen later in this article, the sensing/intuitive dimension plays one of the more important roles in problem solving.

The third dimension, Deciding, refers to how a person makes decisions. The two poles of this dimension are feeling and thinking. Feelers tend to be very attuned to their own feelings and the feelings of others. They base their decisions on what is important to themselves and others. Thinkers, on the other hand, base their decisions on an objective, impersonal, and logical analysis of a situation. They are often focused on cause-effect relationships and seek an objective standard of truth.

The last dimension of the MBTI, Living, refers to how one is oriented toward the outer world. The two poles of this dimension are judging and perceptive. Judgers are people who prefer to work in a linear, orderly method. They seek closure, tend to be organized, and want things settled. Those who prefer a perceptive experience would rather live a flexible, spontaneous life. They prefer to keep their options open and are often viewed as spontaneous.

The four dimensions and the two poles of each, can be combined to identify 16 different types. It should be emphasized that no type is better than another and that individuals use all preferences at one time or another. Type does not explain everything and does not measure abilities. The MBTI simply suggests one's preferences in dealing with the inner or outer world, perceiving information, making decisions, and style when dealing with the outer world.

There is an abundance of literature evaluating the reliability, validity, and usefulness of the MBTI (Carlson, 1989; Carlyn, 1977; Coan, 1978; Devito, 1985; Huber, 1983; Mendelsohn, 1965; Myers & McCaulley, 1989; Sundberg, 1965; Wiggins, 1989; Zemke, 1992). For the most part, researchers and reviewers conclude that the MBTI is a reasonably valid and reliable instrument that is potentially useful for a variety of purposes. Even so, the instrument is not without its critics. Several reviewers (Coan, 1978; Devito, 1985; Mendelsohn, 1965; Wiggins, 1989) question whether the dimensions should be treated as genuine dichotomies. Others (Sundberg, 1965; Zemke, 1992) challenge the item content of the questions. They suggest that the test seems obvious and easy to manipulate. Other criticisms (Huber, 1983; Schweiger, 1983) are not directed at the MBTI in particular but at issues related to research on cognitive styles and personality in general. Zemke (1992) asserts that even the best research available shows a very weak relationship between personality and performance and that other factors play a much greater role. In any event, the study of cognitive style and personality continues to evolve, and the MBTI is clearly one of the best researched and widely accepted of the instruments measuring cognitive style and personality instruments. Keen and Bronsema (1981) suggest that of the many cognitive style and personality instruments that exist, the MBTI should be the basis for cognitive style research in the area of information systems.

EMPIRICAL STUDIES

While there is an abundance of information related to using the MBTI in counseling, education, career guidance, and teamwork, there is considerably less investigating the relationship between the MBTI and computer programming. A literature search focused specifically on the MBTI and computer programming in the classroom yielded only a handful of articles (Carland & Carland, 1990; Corman, 1986; Evans & Simkin, 1989; Werth, 1986). Of these four studies, three, Corman, Evans and Simkin, and Werth, investigated whether type, as measured by the MBTI, could be used to predict achievement in an introductory programming class. Both Werth and Corman tried to predict academic achievement based on a number of variables. In both studies the dependent variable was overall class average, and dimensions of the MBTI as well as a variety of other personality and demographic measures were the independent variables. In study did the dimensions of the MBTI enter the regression equations as being significant variables. Evans and Simkin, on the other hand, followed a similar procedure and found a number of the MBTI variables to be significant in their regression predicting homework and test grades. The poles sensing, intuitive, thinking, introverted, and judging entered the regression equation with positive coefficients, and the feeling pole was significant with a negative coefficient. Carland and Carland administered the MBTI to business information systems students and discussed the implications of type with regard to the educational process.

Because there were so few studies that specifically related the MBTI to computer programming, the literature relating the MBTI to problem solving provided the foundation for this study. Problem solving is clearly a general characteristic of computer programming. The studies that investigated problem solving focused on the perceiving dimension (sensing/intuitive) and the judging dimension (thinking/feeling). Because these two dimensions are considered the cognitive dimensions of Jung's theory, they play the greater role in problem solving (Campbell & Davis, 1988; Hellriegel & Slocum, 1975; McCaulley, 1987; Yokomoto & Ware, 1982).

Boreham (1987) and Yokomoto and Ware (1982) considered fire perceiving dimension (sensing/intuitive) to be the most relevant to problem solving. Yokomoto and Ware suggested that the "sensing-intuitive scale of the MBTI seems to have the strongest impact on problem solving style" (p. 163). Sensors and intuitives have different sets of strengths and weaknesses when solving problems. Sensors focus on facts and immediate realities, and they often see the trees but not the forest. They tend to move from specific to general (inductive) and are very oriented to the present. Intuitives are good at seeing new possibilities, viewing a problem in different ways, and seeing the implications of the big picture. They move from the general to the specific (deductive) and are future-oriented. Yokomoto and Ware found that sensing students in a linear circuit engineering class tended to emphasize details, routines, and procedures. McCaulley (1987) further explained that sensors focus on facts and realities. Computer programming, especially at the novice level, also requires a focus on facts and realities, and it is a very hands-on, practical activity. One runs a program and the results of the run are tangible. The process of debugging (a significant part of the novice's programming process) is clearly detail-oriented and routine. Given the findings of Yokomoto and Ware and given McCaulley's description of sensors, one would expect sensing students to perform better on elementary programming exercises than their intuitive counterparts.

A person with a thinking preference has an analytical and objective orientation, is concerned with definitions and connections, and is often impersonal analyzing situations. Feeling persons are more likely to place emphasis on how deeply they care about the gains or losses associated with alternatives. Hellriegel and Slocum (1975) explained that in managers' problem-solving styles, feeling managers are more likely to make decisions that will win approval of peers, subordinates, and superiors. Thinking managers are more likely to plan and look for a method to solving problems by careful defining constraints, and they have a tendency to solve problems by using standardized procedures. Computer programming requires an impersonal, sequential and logical analysis (the exact description of the thinking pole in the MBTI). While feeling has a place in the design of user interfaces and in dealing with the people who will use the system, it has little place in the actual programming of a machine. Given that the nature of novice programming assignments is generally well specified, the novice programmers' activities are focused on programming the machine. As noted previously, Evans and Simkin (1989) found that feeling characteristics produced a negative coefficient in predicting achievement in a computer programming course. It may follow that thinking students would enjoy and perhaps stick with the process of computer programming more than would feelers.

While the cognitive functions appear to play the more important role with respect to problem solving, the four poles associated with cognitive functions also influence one's approach to solving problems. An introvert will place greater weight on inner ideas and concepts. Introverts are more likely to need quiet concentration and have a greater need for working without interruptions (Hellriegel & Slocum, 1975). When solving problems, extroverts place greater weight on the views of others and are more likely to discuss and communicate during the problem-solving process. Several studies have found that introverts tend to perform better in mathematical and scientific fields (Eysenck, 1964; Kagan & Douthat, 1985). Because computer programming is considered both a scientific and mathematical field, one would expect introverts to perform better than extroverts in programming-related activities.

Judgers are concerned with organizing and structuring their problem-solving processes and are eager to seek closure. Perceptive types are much more adaptable, curious and spontaneous in solving problems. Computer programming research has indicated that the weakest phases in a beginning programmer's problem-solving process involves problem representation and design (Allwood, 1986). Some students immediately begin writing code without any advance planning. This may be especially true for perceptive students, who prefer to follow a more spontaneous method of solving a problem. The structured methodology that the computer programming field as a whole has endorsed is very oriented toward a judging orientation. Structured methods require a step-by-step, organized, planned process. The tools (flowcharts, structure charts, and pseudocode) that many instructors utilize may be more appealing, and hence more helpful to judging students. Because the description of a judger more closely fits the structured methodology of writing programs, it follows that students who are already oriented toward such structured processes will be more successful in writing computer programs.

METHOD

Pilot Study

In the spring of 1991, 34 students in two sections of an introductory programming class at a branch campus of a medium-size midwestern university participated in an exploratory study. Of the 34 students who took the introductory programming class, 25 completed the class. The MBTI-Form G was administered the first day of class. Program grades, quiz grades, and test grades were collected and analyzed at the end of the class.

Two questions guided the exploratory study. First, did personality type have any relationship to programming achievement, test achievement, or overall achievement? Second, did those who completed the class differ from those who dropped the class on any of the personality dimensions measured by the MBTI?

The results of the pilot study combined with the theory of type and previous empirical work provided the basis for the following four hypotheses for testing in a confirmation study.

H1: Introverts will have a higher program average than Extroverts.

 

H2: Sensors will have a higher program average than Intuitives.

 

H3: Judgers will have a higher program average than Perceptives.

 

H4: Thinkers are more likely to complete the class than Feelers.

For the first three hypotheses, the program assignment average was used as the dependent variable and measured student achievement on programming assignments. Specifically, the program assignment average was the students' weighted average on eight out-of-class programming assignments. For the last hypothesis, the number of students completing the class was the dependent variable.

Confirmatory Study

A total of 114 students taking the same computer programming class as that taken by students in the pilot study participated in the confirmatory study. (The students in the confirmatory study were attending a different campus of the university.) These students were in four different sections of the same class. The same instructor taught all four sections. A questionnaire asking about previous programming experience was administered to all students. Of the 114 students, four were not included in the analysis because they reported experience levels above that of a novice. Only one other student was dropped from the analysis. That student was dropped because he was ill during the majority of the term and the grade did not reflect his typical achievement. A total of 16 students either formally dropped the class or stopped attending. The remaining 93 students completed the class. The majority of the students were either systems or engineering majors. Table 1 shows the distribution of students on each of the four dimensions of type. The table reveals that students were evenly split between extroverted and introverted. Approximately 60% of the class was sensing, 60% thinking, and 60% judging.

Procedure

In the spring semester of 1992, the instructor taught four sections of the introductory programming class. Program assignments, quizzes, and exams were the same among the different sections. Although the instructor was aware of the general goal of the research, he was not aware of the specific hypotheses, which eliminated any instructor bias either during grading or teaching.

Following a procedure similar to the one used in the pilot study, the MBTI-Form G was administered in the first week of the class. Each of the tests were scored, and the preference scores were converted to continuous scores, as recommended by Myers and McCaulley (1989). Using this procedure yields a single score for each dimension. For example, for the first dimension the preference scores are converted to a continuous score ranging from 33 to 167. Scores falling below 100 identify an extrovert; scores above 100 indicate an introvert. It is not possible to score 100 using the conversion procedure identified by Myers and McCaulley (1989). During the class eight programs and three exams were administered, graded, and collected. Students in the confirmatory study, unlike those in the pilot study, worked in pairs to complete most programs. The pairs were instructor assigned, and pairings were based on the location of the students' names in the alphabet. Students switched pairs for each assignment. Usually each pair received the same grade, although occasionally the grades differed. Occasionally, a group of three would work on a program. Even though students worked with a variety of other students during the term, an individual's program grade was a good reflection of an individual's programming achievement.

RESULTS

The following sections describe the statistical procedures used to test each hypothesis and the results obtained:

H1: Introverts will have a higher program average than Extroverts. A t-test indicated that this hypothesis was not significant (t = .25, df = 91, p </= .80). Table 2 shows that the overall program average for extroverts was 80.08 and the overall program average for introverts was 79.58. The results of the analysis indicate that there is insufficient evidence to conclude that introverts will have higher program grades than extroverts.

 

H2: Sensors will have a higher program average than Intuitives. A t-test indicated that this hypothesis was significant (t = 2.51, df = 91, p </= .014). Table 2 indicates that the program score for sensors was 81.93 while the program average for intuitives was 76.93. The results of the t-test indicate that there is a significant difference between sensing and intuitive students in the direction hypothesized.

 

H3: Judgers will have a higher program average than Perceptives. A t-test indicated that this hypothesis, while not significant at the .05 level, does approach significance (t = 1.91, df = 91, p </= .067). Table 2 shows that the overall average of the judging students was 81.42, while the overall average for the perceiving students was 77.63.

 

H4: Thinkers will be more likely to complete the class than Feelers. A one-sample chi-square test showed no significant difference between the student drop-out rate for between thinkers and feelers (chi[sup 2] = .06, df = 1, p </= .80)

Hypotheses 1-3 were also tested using an analysis of covariance, using GPA as the covariate. This analysis did not change any of the findings.

DISCUSSION

The first hypothesis, that introverts would have higher programming averages than extroverts, was not supported. The two groups scored essentially the same. This finding differs from the results of the pilot study and some previous empirical studies (Kagan & Douthat, 1985). According to type theory, introverts are more likely to need quiet concentration and to have more need for working without interruptions. They are also less likely to communicate and discuss their ideas with others. In the present study the students worked in pairs on almost all assignments. The process of working in pairs requires a style less preferred by introverts. It may be that the pairings eliminated any effects that may have occurred had the students worked independently.

The second hypothesis, that sensors would have higher programming averages than intuitives, was supported. Sensors prefer to work with observable facts, tend to be focused toward practicality, have a memory for details, and prefer concrete, tangible experiences. Introductory computer programs tend to be quite concrete and provide tangible feedback when the program runs. The entire process of writing, entering, and debugging programs is oriented toward a person focused on facts and experiences. Our findings provide further support for the findings of Yokomoto and Ware (1982) that the sensing/intuitive scale has the strongest impact on the problem-solving process. In our study, the sensing/intuitive difference was the only one significant at the .01 level, indicating that it has a stronger impact than the other dimensions.

Although the third hypothesis--that Judgers would have higher programming averages than Perceptives--was not statistically significant, a t-test showed that the hypothesis approached significance (p </=.07). The process of designing a computer program is focused on a structured methodology. Students were encouraged to preplan and are required to write structured programs. It follows that students who are already oriented toward such structured procedures would be more successful at completing computer programs. Additionally, judgers prefer to organize their activities, and a stronger program grade may be a reflection of the Judgers' preference for organizing and appropriately managing their time. It may be that the difference in scores in this study is linked to the fact that novice students do little advance planning. MBTI theory predicts that perceptive students fail to preplan and therefore they create poorer programs.

In the pilot study, 46% of feeling types dropped the class while only 18% of thinking types dropped the class. However, the confirmatory study hypothesis that thinkers would be more likely than feelers to complete the class was not supported. This indicates that other factors, such as the difficulty of the class, student scheduling difficulties, or spring fever, may have overridden any effects of personality types. Indeed, further analysis revealed that no personality type or combination of personality types was more likely than any other to drop the class. Although cognitive style may not influence a student to the point of dropping a class, it may indeed influence the student's satisfaction with the class. For instance, it would seem that students having a thinking orientation would enjoy a computer programming class more than would those students having a strong feeling orientation.

Additional Observations

One of the most interesting additional findings in both the pilot study and the follow-up study was the lack of significant relationships between exam grades and overall grade in contrast to the significant relationships that were observed with program achievement. There were no significant differences between personality types on any of four personality dimensions between exam grades or final grades. Additionally, in both data sets there was no correlation between program grade and exam grade. This has some interesting implications for both educators and researchers.

Much of the research involving personality and cognitive style has used an overall class grade as a measure of achievement. Typically, a number of demographic, aptitude, and cognitive style measures are taken, and these measures are used to predict achievement in computer programming classes. This study suggests that interesting relationships may exist at a lower level--the level of computer programming--and that the use of an overall class grade as a measure may be too broad. The research reported in this article indicates that the cognitive styles that have a relationship with achievement in completing computer programs are different than those related to overall achievement in a computer class or to computer programming exams. Perhaps investigating the relationship at the programming level would lead to a different set of results.

From an educator's perspective, the fact that program grades did not significantly correlate with exam grades (r = .13) presents some interesting issues. Which is the better measure of a good programmer, exam grades or program grades? On one hand, our findings may be explained by the frequent student collaboration and the assistance students received from the instructor and their peers. On the other hand, one of the main goals of assigning programs is to allow students to get a chance to practice the abstract structures and techniques they learn in class. It is reasonable to assume that students who could create a working program had a better understanding than those who could not. In short, it seems clear that the task of writing a successful program takes a different set of skills and a different set of cognitive styles than does the task of taking tests.

SUMMARY

This study of college students enrolled in an introductory programming class indicated that while personality may have little to do with overall achievement in computer programming classes, it may be more closely related to achievement on one of the components of the class grade--programming assignments. This study revealed that sensing students scored significantly higher than intuitive students on writing computer programs, and judging students achieved higher grades on computer programs than did perceiving students (although the results were not statistically significant to the .05 level). These findings indicate that the act of creating and debugging computer programs is a feat in its own right and should be considered independently of overall grade and exam grade. Additionally, personality type does not appear to be an important factor in predicting whether a student will drop a class. While personality type may not cause a student to drop a class, it may influence a student's evaluation of a class.

Contributors

Catherine Bishop-Clark is an assistant professor in the Systems Analysis Department at Miami University of Ohio. She holds a B.S. degree in computer science, and an M.S. degree in information systems. She is a doctoral candidate in the Educational Foundations department at the University of Cincinnati. Daniel D. Wheeler is a cognitive psychologist with a primary interest in the applications of computers in education. He is currently an associate professor at the University of Cincinnati, with a joint appointment in the College of Education and the Department of Psychology. (Address: Catherine Bishop-Clark, Miami University, 4200 East University Boulevard, Middletown, OH 45042.)

Table 1 Student Distribution by Type (N = 93)

Table 2 Comparison of Groups by Personality Type

References

Allwood, C.M. (1986). Novices on the computer: A review of the literature. International Journal of Man-Machine Studies, 25, 633-658.

Boreham, N.C. (1987). Causal attribution by sensing and intuitive types during diagnostic problem solving. Instructional Science, 16, 123-136.

Brooks, R. (1980). Studying programmer behaviors experimentally: The problems of proper methodology. Communications of the ACM, 23, 207-213.

Campbell, D.E., & Davis, C.L. (1988). Improving learning by combining critical thinking skills with psychological type. Wright-Patterson AFB, OH: School of Systems and Logistics, Air Force Institute of Technology. (Eric No. ED 306 250)

Carland, J.A., & Carland, J.W. (1990). Cognitive styles and the education of computer information systems students. Journal of Research on Computing in Education, 23, 114-126.

Carlson, J. (1989). Affirmative: In support of researching the Myers-Briggs Type Indicator. Journal of Counseling and Development, 67, 484-486.

Carlyn, M. (1977). An assessment of the Myers-Briggs Type Indicator. Journal of Personality Assessment, 41, 461-473.

Coan, R.W. (1978). Critique of the Myers-Briggs Type Indicator. In O.K. Buros (Ed.), The Eighth Mental Measurement Yearbook (pp. 973-975). Highland Park, NJ: Gryphon.

Corman, L. (1986). Cognitive style, personality type, and learning ability as factors in predicting the success of the beginning programming students. SIGCSE Bulletin, 18(4), 80-89.

Devito, A.J. (1985). Review of the Myers-Briggs Type Indicator. In J. Mitchell (Ed.), The Ninth Mental Measurement Yearbook (pp. 1030-1032). Highland Park, NJ: Gryphon.

Evans, G., & Simkin, M. (1989). What best predicts computer proficiency. Communications of the ACM, 32, 1322-1327.

Eysenck, H.J. (1964). Principles and methods of personality description, classification, and diagnosis. British Journal of Psychology, 55, 284-294.

Hellriegel, D., & Slocum, J.W. (1975). Managerial problem-solving styles. Business Horizons, 18, 29-37.

Hirsh, S., & Kummerow J. (1989). Life types. New York: Warner Communication.

Huber, G.P. (1983). Cognitive style as a basis for MIS and DSS designs: Much ado about nothing. Management Science, 29, 567-576.

Jung, C. (1923). Psychological types. New York: Harcourt, Brace.

Kagan, D.M., & Douthat, J.M. (1985). Personality and learning FORTRAN. International Journal of Man-Machine Studies, 22, 395-402.

Keen, P.G.W., & Bronsema, G.S. (1981). Cognitive style research: A perspective for integration. Proceedings of the Second International Conference on Information Systems (pp. 21-52).

Keirsey, D., & Bates, M. (1984). Please understand me: Character & temperament types (5th ed.). Del Mar, CA: Prometheus Nemesis.

Mendelsohn, G.E. (1965). Critique of the Myers-Briggs Type Indicator. In O.K. Buros (Ed.), The Sixth Mental Measurement Yearbook (pp. 321-322). Highland Park, NJ: Gryphon.

McCaulley, M.H. (1987). The Myers-Briggs Type Indicator: A Jungian model for problem solving. New Directions for Teaching and Learning, 30, 37-53.

Myers, I.B., & McCaulley, M.H. (1989) Manual: A guide to development and use of the Myers-Briggs Type Indicator. Palo Alto, CA: Consulting Psychologists Press.

Schneiderman, D. (1980). Software psychology: Human factors in computer and information systems. Cambridge, MA: Winthrop Publishers.

Schweiger, D.M. (1983). Measuring managers' minds: A critical reply to Robey and Taggart. Academy of Management Review, 8, 143-151.

Sundberg, N.D. (1965). Critique of the Myers-Briggs Type Indicator. In O.K. Buros (Ed.), The Sixth Mental Measurement Yearbook (pp. 322-326). Highland Park, NJ: Gryphon.

Werth, L. (1986). Predicting student performance in beginning computer science class. SIGCSE Bulletin, 18(1), 138-142.

Wiggins, J.S. (1989). Critique of the Myers-Briggs Type Indicator. In J.C. Conoley & J.J. Kramer (Eds.), The Tenth Mental Measurement Yearbook (pp. 537-538). Highland Park, NJ: Gryphon.

Yokomoto, C.F., & Ware, J.R. (1982). Improving problem solving performance using the MBTI. In L.P. Graysona & J.M. Biedenback (Eds.), Proceedings of the 90th Annual Conference of the American Society for Engineering Education (pp. 163-167). Washington, DC: ASEE.

Zemke, R. (1992). Second thoughts about the MBTI. Training, 29(4), 43-47.

~~~~~~~~

By Catherine Bishop-Clark, Miami University and Daniel D. Wheeler, University of Cincinnati