Before you start

Pick one belief. Not the biggest, and not the one you think you ought to work on, but one that is tied to strong feeling, gets in the way of something you want, and matters to you. If several qualify, start with the one most likely to give you an early success. Beliefs held for years do not move in one go, and the first experiment is mostly there to show you the method works.

A few kinds of belief are better tested with a professional than alone: anything about a traumatic memory, a genuine medical risk, or thoughts of harming yourself. If you are already seeing a therapist, plan the experiment with them and follow their advice over anything here.

Step 1. Write the belief as a prediction

Most worries arrive as feelings or as vague verdicts: I am boring, something will go wrong, I cannot handle it. None of those can be tested, because none of them say what would happen. Rewrite the worry in the form if I do this, then that will happen.

Then rate how much you believe it right now, 0 to 100. Go with your gut. You will rate it again afterwards, and the gap between the two numbers is the point.

Step 2. Choose the situation

You want the best test of the belief that you are willing to try. Not the scariest situation, and not one so easy that a good result proves nothing. Real situations beat imagined ones, and many take only a few minutes. Say what you will do, where, and with whom, and if you can, give it a date and time. People who set a time follow through far more often than people who plan to do it "this week."

If the best test is out of reach for now, build a step ladder and start on a rung you are willing to climb today.

Step 3. Say what you expect to see

Describe the worst you think will happen if you go in without your usual precautions. Make it something you could see, hear, or count. Then write down how you would know it had come true. "I will feel awful" does not count, because you almost certainly will feel awful, and that tells you nothing about the belief. "Someone laughs, or comments on it, or two people look at each other" is a result you can check.

Rate how likely the prediction feels, 0 to 100. This is a separate number from the belief rating. The belief is the general rule; the prediction is what it says about this one occasion.

Step 4. Decide what to leave out

Safety behaviors are the things you do to stop the feared outcome: rehearsing every sentence, gripping the cup so your hands cannot shake, sitting near the door, checking your pulse, asking someone to come with you. They feel sensible. The problem is that when the feared thing does not happen, you credit the precaution, not the fact that the fear was wrong. The belief survives untouched.

So you name the safety behaviors in advance and choose which ones to drop this time. Not necessarily all of them. One or two you are willing to do without is a real experiment; all of them at once is often too much and gets abandoned.

Decide, too, where your attention will go. Fear pulls it inward, onto your body and how you sound. Plan to keep it outward: on faces, on what people actually do, on the room. More on safety behaviors and why they matter.

Step 5. Do it, and watch

Hands off the wall. Attention outward. Stay until the prediction has actually been tested, not until the urge to leave peaks, because leaving at the peak teaches you that escape was necessary. Feelings rise and then fall on their own, and staying is how you find that out.

If you can, jot a few words as it happens or straight after: what people did, what was said, what you noticed. Memory edits in favor of the fear within hours.

Step 6. Record the facts, then re-rate

Write down what happened as a camera would have recorded it. What did you do? What did other people actually do or say? Feeling terrible is real, but it is not the outcome. Then answer the plain question: did the prediction come true? Yes, partly, or no.

Re-rate the prediction, knowing what you know now. Re-rate the belief. A drop of ten or twenty points is progress. Beliefs change through accumulated evidence, which is why the next step exists.

Step 7. Draw the lesson and build on it

This is where the value lives, and it is the step most people skip with "it went fine." Go further. What does the outcome say about the belief in general? About how you come across? About how safe you actually are, or what these feelings mean? Write one sentence you could reread the next time the worry shows up.

Then listen for the doubt. "Yes, but they were being polite." "Yes, but it was a small group." That doubt is your next experiment. Repeat the same test to rule out luck, scale it up, or aim straight at the new doubt.

When the feared thing happens

Sometimes it does. Someone laughs. Your voice does shake. The experiment has not failed. You have arrived at the second half of the fear, which is usually the more important half: was it as bad as you expected, how did people actually react, and how long did it matter? The real cost tends to be far smaller than the imagined one, and finding that out changes the belief more than a smooth run would have.

Three ways to do it

On paper. Print the worksheet. It has the same seven steps, with room to write.

In your browser. The online worksheet saves to your own device and can be printed or saved as a PDF. Nothing is sent anywhere.

In the app. Behavioral Experiment for iPhone and iPad keeps your beliefs and experiments together, shows the rating over time, reminds you when one is due, and includes worked examples and the articles from the Learn section, in English and Spanish.

---

This guide follows the seven steps described by Warnock-Parkes and colleagues (2025) and the earlier Oxford Guide (Bennett-Levy et al., 2004). See the research page for the references, and Learn for the ideas behind each step. Educational content, not medical advice or a substitute for professional care.