Critical Thinking: How to Be Right More Often

DAY 3 | Testing the assumptions and evidence under a confident conclusion

Behind every confident conclusion sits a chain of small inferences that nobody checked, and if one link is weak the whole conclusion is too. So far this week we have worked on catching bias in how you judge, and on the frame you put around a problem. Today we go to the join between them: the moment a pile of raw facts turns into a belief you will act on. That turn happens so fast it feels like observation, as if you simply saw the truth. You did not. You selected some of the data, added meaning to it, and leapt to a conclusion, all in a fraction of a second, and now you are defending the view from the top of that leap as though it were the ground floor. Critical thinking is largely the discipline of climbing back down and checking each step you skipped.

This work is uncomfortable, because it is aimed mostly at yourself. It is easy to question other people's assumptions. It is hard to notice the ones you are standing on without realising, the beliefs that feel not like assumptions at all but simply like the way things are. Most of the skill is learning to make those invisible beliefs visible, so they can be tested before they quietly steer months of work in the wrong direction.

The ladder of inference

The organisational thinker Chris Argyris gave this fast leap a shape he called the ladder of inference, and it is the single most useful thinking model I know. At the bottom of the ladder is the full pool of observable data, everything that actually happened. In an instant, you select a thin slice of it, add meaning based on your past experience, draw conclusions, and act on beliefs that feel like direct observation but are several assumption-laden rungs above what you really saw. You climbed so fast you never noticed climbing.

Customers are leaving because our prices are too high is a view from the top of the ladder. Somewhere at the bottom was some real data, perhaps a handful of customers who mentioned price, and from that thin slice, a confident conclusion was built and then treated as fact. The discipline is to climb back down and ask at each rung: what did we actually observe, as opposed to conclude? What data did we select, and what did we ignore? Are there other meanings the same facts would support just as well? Climbing down the ladder turns a comfortable conviction back into a question you can test, and asking that one question, how do we know this, at the right moment has saved me from more expensive mistakes than any other habit I have.

Surface the assumptions, then test the load-bearing one

Once you accept that every conclusion rests on assumptions, the work is to drag them into the light and check the ones that matter. Not all assumptions are worth the effort. The skill is finding the load-bearing one, the belief that, if it turned out to be false, would collapse the whole conclusion. That is the one to test first, because it is the cheapest possible way to avoid a big mistake.

• List what has to be true for your conclusion to hold. Write the assumptions out plainly, including the ones that feel too obvious to say.

• For each, separate what you have actually observed from what you have inferred or been told. Mark which is which. The mix is usually humbling.

• Find the load-bearing assumption, the one whose failure would change everything, and design the cheapest test that could prove it wrong.

• Then go and look. The exit survey, the real data, the actual user. One honest hour of looking beats a week of confident argument built on none.

Two habits turn this from an attitude into a practice.

The first is steelmanning, the opposite of the straw man. Instead of arguing against the weakest version of a view you disagree with, you build the strongest possible version of it, the one its smartest supporter would make, and then see whether your own position still stands. If you cannot state the opposing case well enough that its holder would nod, you do not yet understand the decision; you only understand your side of it.

The second is actively hunting for the evidence that would prove you wrong. Most of us, once we have a view, go looking for support, and support is always easy to find, which is why finding it tells you almost nothing. The more powerful move is to ask what evidence would show my conclusion is false, and then go looking for that on purpose. A belief that survives a genuine attempt to break it is worth trusting. A belief you have only ever tried to confirm is just a hope you have grown attached to. This is uncomfortable precisely because it means seeking out the thing you least want to find, and that discomfort is the clearest sign you are doing it properly.

Correlation is not causation

One evidence trap deserves its own warning because it fools careful people constantly: the leap from two things moving together to one causing the other. Sales dipped the same month a competitor launched, so the competitor caused the dip. Perhaps. Or the season did, or a quiet price change did, or the two are unrelated, and it is a coincidence. Things move together for many reasons, and only one of them is cause. The tell is a sentence with 'because' in it that nobody has actually tested; that small word does a lot of quiet work, and it is worth stopping every time you hear yourself use it. Before you build a plan on because, ask whether the two things could move together without one driving the other, whether something else could be driving both, and whether the timing might simply be chance.

The price theory from today's post is a version of this. Prices were high, and customers left, so price drove the leaving. The two moved together, and the story felt complete. Only when someone looked past the correlation to the actual reason, the broken first week, did the real cause appear, sitting quietly behind a coincidence everyone had mistaken for an explanation. Whole strategies get built on a correlation that someone promoted to a cause, and they fail because the lever being pulled was never connected to the thing it was meant to move.

The mistakes that pass for analysis

Some of the most confident thinking is the least critical, and it fails in recognisable ways.

• Gathering only the evidence that supports the conclusion you started with, and mistaking the size of the pile for its strength.

• Treating a strongly held belief as a fact, so it never gets written down as an assumption and never gets tested.

• Arguing against the weakest version of the opposing view, winning easily, and learning nothing.

• Reasoning from memory and opinion when the actual data is one honest hour away.

• Promoting a correlation to a cause because the timing lined up and the story felt satisfying.

The cost of skipping this

The cost of not testing your assumptions is rarely dramatic, which is exactly why it is dangerous. It is a slow tax paid in quiet ways: the recommendation built on a cause nobody verified, the requirement everyone agreed to because it was never named as an assumption and so never questioned, the months of work that followed a conclusion someone reached at the top of the ladder in half a second. None of these looks like thinking failures when they happen. They look like bad luck, or shifting requirements, or somebody else's mistake. They were none of those. They were a single question: 'how do we know this', that nobody asked out loud while it was still cheap to ask. The analysts people come to trust are the ones who ask it early and without drama, and who would rather be briefly uncomfortable now than confidently wrong later.

Go out and be successful. Oluwatosin Ogunkoya | Flotog BA Insights  |  www.flotogbainsights.com

Tomorrow: Tools That Sharpen Judgment

A small kit of reasoning moves, from the pre-mortem to the outside view, for high-stakes calls.