Actual TKI
30
forced-choice pairs
5
conflict modes
No formula for who wins
The page I took
36
item pairs
6
conflict modes
One formula predicting who wins
Most of us have quoted an assessment result at some point. A personality type in a bio, a working-style label in an onboarding doc, a number from something we took online once. We quote the finding. We almost never quote the instrument.
I did this with a conflict-style assessment for a graduate leadership weekend, and I only noticed months later, while sorting the files to write about them.
What I was asked to do
The pre-work was three self-assessments before day two. The syllabus listed them plainly, each with a link:
Assessment – During the ALW, you will spend time developing your self awareness of our strengths and assets as a leader. Complete these assessments before Day 2 (Saturday) to prepare for the session.
TKI – Sample Thomas Kilmann Inventory can be found online
I followed the link, answered 36 forced-choice pairs, saved the result pages, and filed them under TKI-Conflict-Mode. Then I wrote and spoke about my TKI results.
What came back
The report named a primary mode and scored six scales:
Cooperation
75% 9 of 12
Constructive confrontation
66% 8 of 12
Compromise
58% 7 of 12
Destructive confrontation
33% 4 of 12
Avoidance
33% 4 of 12
Adaptation
33% 4 of 12
Then it did something I should have stopped at:
A = ½ constructive confrontation + ½ destructive confrontation + cooperation + ½ compromise
B = avoidance + adaptation + ½ compromise
Your Result: A = 153; B = 95, so you have more chances to win the conflict.
I remember reading that as good news.
The check I didn’t run
The Thomas-Kilmann Conflict Mode Instrument is a real, specific thing. Thirty pairs of statements, not thirty-six. Five modes — competing, collaborating, compromising, avoiding, accommodating — not six. Held under copyright by The Myers-Briggs Company and not free; about $22 on paper, about $45 online.
And its central claim, the thing the whole framework exists to argue, is that no mode is the good one. Competing is right when a decision is urgent. Accommodating is right when you are wrong. The instrument describes situational fit. It does not tell you your odds.
The page I was sent to splits competing into “constructive” and “destructive” confrontation, renames accommodating to “adaptation,” hands out an animal, and calculates whether you will win. Every one of those is a departure from the instrument whose name was on the link. Its own footer says the results “can in no way be interpreted as a specialist or doctor’s advice.”
Here is the part that stung when I went back. The site never claimed otherwise. Its description of itself reads:
This test was inspired by the Thomas-Kilmann conflict modes study but has completely different pairs of statements, scoring algorithm, and result scales.
Completely different pairs of statements, scoring algorithm, and result scales. That is a full and accurate disclosure, published by the people who built the thing, in one sentence, before you answer a single question.
Syllabus
“TKI – Sample Thomas Kilmann Inventory”
Questionnaire
“Inspired by” the conflict modes study
My notes
A folder named “TKI-Conflict-Mode”
Nobody misled me. The word “TKI” was not on the test. It was on the link that sent me there, and then it was on my folder, and then it was in my mouth.
What 75% actually means
Six scales, 36 items, two scales per item: each scale appears in twelve pairs, so every percentage is a count out of twelve. Cooperation 75% is nine selections. Constructive confrontation 66% is eight. Compromise 58% is seven. The three scales at 33% are four apiece.
Add them: 9 + 8 + 7 + 4 + 4 + 4 = 36. Exactly the number of items, because each item is one choice, and every choice you give to one scale is a choice you take from another. The six numbers are not six findings about me. They are one distribution of 36 tokens across six buckets.
The real TKI is forced-choice too, so its raw counts have the same property. The difference is what it does next: it converts them into percentiles against a normative sample of 8,000 employed adults, and its own report says those are the part that matters — “the percentage of people in the sample who scored the same as or lower than you on each mode.”
That is the whole job, and it is why the licensed one costs money: someone had to go and measure eight thousand people first. A raw count tells you which mode you picked most often against yourself. Only a percentile supports a sentence like “I am a cooperative person.”
This report
9 of 12
A count of the choices I made against other choices.
Actual TKI report
8,000
The employed adults in its normative comparison sample.
My 75% was a proportion of my own answers, formatted to look like a standing among strangers.
Why the free one exists
A licensed instrument costs real money per person, per administration. A weekend course cannot buy a whole cohort three different assessments for pre-work that exists to warm people up for a Saturday discussion. So the syllabus points at a free page covering similar ground, marks it “sample,” and moves on. The same syllabus sourced Career Anchors from the same site, the same way. The cost constraint is real and I don’t think there is a villain in it.
But the fix has a specific failure built in. The people who built the questionnaire labelled it accurately — inspired by. The syllabus relabelled it TKI – Sample Thomas Kilmann Inventory. One word of hedging was carrying the entire distinction, and hedges do not survive the trip into a student’s notes. What survives is the name. Mine went into a folder called TKI, into a reflection, and into how I described myself out loud.
So the integrity leaked at the relabelling, not at the free test — which told the truth about itself. “A free questionnaire covering similar ground; note it is not the licensed TKI” would have cost nothing and prevented all of it.
What I should have done, in order
- Open the instrument’s own source before the result page. One search for who publishes it and what it costs. If it is free and the real one is not, that is the end of the question.
- Count the scales against the published framework. Six modes where the literature says five is not a refinement. It is a different instrument.
- Look for claims the original refuses to make. A win probability was the tell. The framework’s whole argument is that modes are situational, so any output ranking you against an opponent is coming from somewhere else.
- Work out whether the scores are allowed to move independently. Convert the percentages back into counts and add them up. If they sum to the number of items, the scales are trading against each other and no single one of them means what it appears to mean on its own.
- Name the source, not the framework, in my own notes. “Conflict-mode questionnaire, psycho-tests.com, Nov 2025” would have cost four extra words and prevented all of this.
What the exercise was still worth
Not much needs overcorrecting. The pre-work did its job — it got a room thinking about how they handle disagreement before a session on exactly that, and the Saturday conversation never depended on the scoring being sound. A rough prompt that starts a good discussion is a legitimate tool.
My failure was narrower than the instrument’s: I let a label stand in for knowing what I had taken, then repeated it. In testing work I would never accept a result without knowing what produced it — the build, the environment, the version. I applied none of that to a measurement of myself, because it arrived with an authoritative name on it and said something I was pleased to hear.
That last clause is the uncomfortable one. The report told me I was cooperative and likely to win. I did not go looking for its provenance, and I think the flattery is most of the reason.
The takeaway
An instrument’s name is a claim, and claims get checked. The thirty seconds it takes to find out who publishes something, how many scales it has, and what it charges is the cheapest quality gate available, and I skipped it on the one assessment that was ostensibly about self-awareness.
I still don’t know what the actual TKI would say about me. That is a separate piece of work, and now a more interesting one, because I would be comparing two instruments rather than accepting one.
Thanks for reading. If you have quoted an assessment result recently — a type, a style, a colour, a letter — do you know who published the thing that produced it?
Comments