A Brief History of the Thomas-Kilmann Conflict Mode Instrument

by Ralph H. Kilmann

FIRST CONTACT

In the fall of 1970, I entered the doctoral program in UCLA’s Graduate School of Management. I chose to major in the behavioral sciences. During my first year in the program, with the ten other entering students in the same major, I was required to take a twenty-four-credit seminar, “Foundations in the Behavioral Sciences,” which was team-taught by five faculty members. That’s how I met Kenneth W. Thomas. Aside from teaching at UCLA, he was also completing his Ph.D. degree from Purdue University.

That twenty-four-credit seminar met nine hours every week, from October 1970 to April 1971. Week by week, I developed a keen interest in designing quantitative assessments of human behavior, which required me to learn more about the validity and reliability of assessments. Based on this knowledge, I then designed a projective measure of interpersonal values, which ultimately became the topic of my doctoral thesis (Kilmann, 1972).

While we, the students, diligently pursued our research studies, the five instructors of the seminar regularly updated us on their own projects, which covered a range of topics. But I was particularly fascinated with Ken’s interests. Besides finishing up his dissertation on conflict modes and interdepartmental conflict (Thomas, 1971), he was also writing a chapter, “Conflict and Conflict Management,” for an important handbook in the field (Thomas, 1976).

DISSATISFACTION WITH EXISTING INSTRUMENTS

In our discussions (both inside and outside of class), Ken regularly expressed dissatisfaction with the way conflict-handling behavior was assessed in research studies—based on his doctoral thesis. Blake and Mouton (1964) had developed the original instrument to measure the five modes: Their assessment consisted of five statements, each describing one of the modes. In the original version, respondents were asked to simply pick the one statement that best described their usual behavior. In subsequent uses of the instrument, respondents were asked to rank order the five statements (from “1” for highest to “5” for lowest), according to how well each one captured how they usually approach conflict situations. But from the way the five statements were worded, it was obvious that “collaborating” was the ideal mode, while “avoiding” was the least desirable one. Sure enough, that’s exactly how managers rated themselves, with over 90% ranking themselves highest on collaborating and lowest on avoiding. Their subordinates, of course, experienced those same managers very differently.

By the time the doctoral seminar concluded in April 1971, I’d already made substantial progress on learning about instrument methodology (including the whole host of response biases, including social desirability) and then applying that knowledge to my assessment of interpersonal values. From Ken, I had learned a lot about conflict theory and more about the social desirability response bias. While I don’t recall if there was a single moment, an epiphany of sorts, but at some point Ken and I must’ve said out loud: “Let’s develop a new instrument to measure the five conflict-handling modes.”

THE WORK ON THE TKI BEGINS

Since Ken’s chapter for the handbook had reinterpreted Blake and Mouton’s framework—from the independent dimensions of “concern for task” and “concern for people” to the independent dimensions of assertiveness and cooperativeness, respectively—we naturally proceeded with his revised model for defining the five modes: competing, collaborating, compromising, avoiding, and accommodating. We also made good use of the chapter I was preparing for my dissertation, which summarized and integrated the major works on instrument methodology and response biases.

From the start, Ken and I knew that the biggest challenge for designing a better instrument to measure conflict-handling behavior was to overcome the social desirability bias: people’s tendencies to present the most socially acceptable image of themselves, rather than present themselves as they really are. That bias is what plagued the Blake and Mouton instrument, as well as the two other conflict-handling instruments: Lawrence and Lorsch (1967) and Hall (1969).

Top of Page

Addressing the Social Desirability Response Bias

How did we minimize the social desirability bias for our new conflict instrument? In a nutshell, Ken and I created thirty pairs of statements. For each pair, the respondent has to choose either the A or B item (for example, one item depicts collaborating while the other item describes avoiding). But each pair of statements was specifically designed—through a multi-stage research process—to be equal in social desirability (for example, “I attempt to get all concerns and issues immediately out in the open” versus “I try to postpone the issue until I have had some time to think it over”). Since each pair of items on the TKI is equal in social desirability (and thus each item in a pair will make a person feel equally desirable to himself and others), a person has to choose each item based solely on its content (whether he usually relies on avoiding over competing, and so forth.). Ken and I thus made it very difficult for people to choose items to make themselves look good; instead, they have little choice but to indicate how they actually behave in conflict situations.

There is a nice byproduct of this procedure for dramatically reducing the social desirability response bias: The forced-choice design conveniently measures the relative frequency of respondents choosing one mode over another. It was apparent to Ken and me that people don’t use several modes at the same time (which would be measured, ironically, if there wasn’t a forced-choice or similar format on a conflict assessment). Instead, people habitually use one mode more than another and that is precisely what we sought to measure with our Conflict Mode Instrument (which is nicely achieved within a forced-choice format).

Making a Big Decision

In 1973, Ken and I completed our research article about the development and validation of our conflict instrument, which we planned to submit to an academic journal. But we faced a dilemma: Should we include the TKI in the appendix of our article (which would limit its exposure to other researchers and professors)? Or should we try to get the TKI published as a stand-alone assessment tool (which could bring the TKI to a much larger audience in business, education, and government)?

At one point in this decision-making process, Ken, as a newly appointed assistant professor, decided to ask a distinguished professor in marketing about our predicament. Dr. Hal Kassarjian gave Ken this advice: “Don’t be ridiculous. Only researchers are likely to use your thirty-item questionnaire. Put it in the appendix of your article and be done with it.”

But a few months later, Dr. Warren Schmidt, the Director of Executive Development at UCLA, asked Ken to help him develop some course materials for a training program on conflict management. The work was being sponsored by a publishing company in New York: Xicom, Inc. During one of Ken's visits to Xicom, he brought along a copy of the TKI and asked if they might be interested in publishing it as a management training tool. Because of Xicom's positive response, Ken and I did not include our instrument in the appendix of our forthcoming article (Kilmann and Thomas, 1977). We instead signed a formal contract with our new-found publisher.

Xicom Publishes the TKI

In 1974, Xicom published the TKI as a sixteen-page booklet. Along with the thirty pairs of items and a self-scoring page, respondents could also graph their scores in comparison with a normative sample of 339 middle and upper managers in government and business. In this way, respondents could learn if any of their scores were relatively high (in the top 25%) or low (in the bottom 25%), thereby suggesting which modes they might be using too much or too little, as compared to our normative sample. The TKI booklet also helps respondents interpret their scores by (1) suggesting when each mode is most useful in a work situation and (2) asking a series of developmental questions for respondents who scored high or low on a particular mode.

Top of Page

ADDRESSING CRITICISMS: AN ONGOING PROCESS

By the late 1970s, we started receiving a few complaints from customers about the TKI’s exclusive use of male pronouns. At first, since only five of the thirty pairs of statements included a “he,” “his,” or “him” in one of the A/B choices, we weren’t particularly concerned. But by the early 1980s, the frequency of complaints about “the masculine bias” gradually increased. Consequently, by the mid-1980s, the use of the male pronouns on five items was changed to he/she, his/hers, or him/her, which easily resolved the gender issue.

In the late 1980s, another criticism developed, which concerned the small (and demographically limited) sample that had established our conflict-handling norms. Recall: We sampled only 339 middle and upper-level managers in business and government. Although not reported at the time, Ken and I knew that our sample from the 1970s was predominately white male and highly educated, which was, of course, very different from the increasingly diverse U.S. workforce. But at the time these complaints were surfacing, we were already busy with other research projects: He was studying empowerment and intrinsic motivation, while I was exploring corporate culture and transformation. So we decided to live with the limited norm sample, even though we continued to use the TKI in our university teaching and consulting work.

CPP Acquires Xicom

In 1998, Xicom, Inc. was acquired by CPP, Inc. in Mountain View, California, the publisher who now distributes the TKI throughout the world, along with its other well-known instruments, including the Myers-Briggs Type Indicator, the Strong Interest Inventory, and the FIRO-B.

Soon after the acquisition, the TKI booklet was printed in full color and translated into Spanish, French, and Portuguese. But we continued to get complaints about the limited sample that was still being used to categorize whether a person scores high, medium, or low on each conflict mode. With the new developments of the World Wide Web, CPP put the TKI online in 2002, which provided computer-generated scores and instantly graphed the results on screen. Conveniently, this online service, SkillsOne, enabled CPP to gather extensive data from a diversified sample of respondents—which set the stage for finally updating the TKI norms from the 1970s to the 2000s. 

Using SkillsOne to Update the Normative Sample

After 60,000 people had taken the TKI through SkillsOne (which, by design, also collected a wide array of demographic information), CPP’s research department created a randomized sample of 4,000 men and 4,000 women—stratified by race/ethnicity, age, organizational level, and geographical region in order to mirror the U.S. workforce. The full report was prepared by Nancy A. Schaubhut and then published by CPP: Technical Brief for the Thomas-Kilmann Conflict Mode Instrument: Description of the Updated Normative Sample and Implications for Use (2007).

Although Ken, I, and the people at CPP were fully prepared to offer different respondents a different normative profile (based on gender, race/ethnicity, age, level, and region) in order to categorize their scores as high, medium, or low, that didn’t turn out to be necessary. CPP’s extensive research study resulted in two big surprises: (1) there were no significant differences, practically speaking, in the conflict norms across all demographic distinctions and (2) there were negligible changes in the norms from the 1970s to the 2000s.

Regarding the first surprise, we found it amazing that demographic differences had no practical significance on conflict norms (even though there were statistically significant differences due to the very large number of respondents in the SkillsOne sample). This practical finding is all the more striking since the demographic profile of the U.S. workforce has changed so dramatically in the past thirty years—particularly with the large influx of women and minorities into the workplace. The way people handle their differences, however, has stayed pretty much the same (practically, not statistically, speaking), independent of gender, race/ethnicity, age, organizational level, and geographical region.

Regarding the second surprise, out of a total of fifteen norms for the TKI (the three categories of high, medium, and low scores for five conflict-handling modes), seven norms stayed exactly the same, while the other eight norms changed by only one number on the assessment. For example, instead of the top 25% of compromising scores ranging between 9 and 12 on the original norms, the top 25% varies between 10 and 12 on the revised norms. Of the other norm categories that changed, they also shifted by only one number.

Top of Page

Implications of the Updated Norms

In sum, although experience shows that there are striking differences in behavior between male and female, black and white, young and old, non-supervisory personnel and top-level executives, and southerners and mid-westerners—conflict-handling norms are pretty much the same across these demographic distinctions and have remained largely the same for the past thirty years. Little did Ken and I know that our 1970s’ homogeneous sample of 339 managers would indeed result in the same norms as the 2000s’ diversified sample of 8,000 respondents. Perhaps the one behavior that helps define the essence of the U.S. culture is the way that its citizens—across the board—handle conflict situations.

With these minimal changes in norms for the TKI, it was easy to redo the one page in the instrument that profiles a respondent’s results into high, medium, and low scores. And we can now say, with confidence, that the TKI’s normative profile is based on a large, diverse, and up-to-date sample of respondents.

The Cultural Challenge for the Future

But two questions have been increasingly asked about the TKI assessment: Is the instrument unique to the U.S. culture? Can it be effectively applied to other countries and cultures? These issues are even more important to address now that the TKI is available in several languages: Spanish, French, Portuguese, Danish, Dutch, Swedish, and Chinese (both traditional and simplified).

Recently, CPP partnered with OPP in Europe and Fellipelli in Brazil to conduct its first international study of culture and conflict: Workplace Conflict and How Business Can Harness It to Thrive (2008). By collecting data in nine countries (United States, Belgium, Brazil, Denmark, France, Germany, Ireland, the Netherlands, and the United Kingdom), this technical report examines the global causes—and costs—of conflict in the workplace.

Although this international study on conflict management didn’t make use of the TKI to assess specific modes and how they vary from country to country, CPP will undoubtedly expand its online capabilities to other countries. In this way, data can easily be collected to see how conflict norms vary across country and if there are significant demographic differences across subpopulations within the same country. Such research will help determine if the current U.S. norms for managing conflict can remain on all TKI translations or if different norm profiles must be developed for different countries and different subpopulations within those countries.

THE TKI IN PERSPECTIVE

Looking back at the past thirty-eight years, the TKI was initially conceived when Ken and I met each other at UCLA in 1970—he as a young assistant professor and I as an even younger doctoral student. At that time, we had no idea if our instrument would actually be published outside of our research article. And we certainly never dreamed the TKI would become so widely used in many different kinds of organizations. But we have remained exceedingly grateful for the excellent work that Xicom once did, and CPP is now doing, to bring our creation to many others.

Top of Page

BIBLIOGRAPHY

Blake, R. R., and Mouton, J. S. The Managerial Grid (Houston: Gulf Publishing, 1964).

CPP, Inc. "Global Human Capital Report: Workplace Conflict and How Businesses Can Harness It to Survive. (Mountain View, CA: CPP, Inc., 2008).

Hall, J. Conflict Management Survey (Conroe, Texas: Teleometrics International, 1969).

Kilmann, R. H. The Development and Validation of a Projective Measure of Interpersonal Values (University of California at Los Angeles. Unpublished Dissertation, 1972).

Kilmann, R. H., and Thomas, K. W. "Developing a Forced-Choice Measure of Conflict-Handling Behavior: The MODE Instrument," Educational and Psychological Measurement, Vol. 37, No. 2 (1977), 309-325.

Lawrence, P. R., and Lorsch, J. W. Organization and Environment (Boston: Graduate School of Business Administration, Harvard University, 1967).

Schaubhut, N. A. "Technical Brief for the Thomas-Kilmann Conflict Mode Instrument: Description of the Updated Normative Same and Implications for Use" (Mountain View, CA: CPP, Inc., 2007).

Thomas, K. W. Conflict-Handling Modes in Interdepartmental Relations (Purdue University. Unpublished Dissertation, 1971).

Thomas, K. W. “Conflict and Conflict Management,” in M.D. Dunnette (Ed.), Handbook of Industrial and Organizational Psychology (Chicago: Rand-McNally, 1976), 889-935.

Thomas, K. W., and Kilmann, R. H. The Thomas-Kilmann Conflict Mode Instrument. (Mountain View, CA: CPP, Inc., 1974).

Back to Top