How to Choose the Right CBT Homework for Social Anxiety

Social anxiety responds to behavioural experiments more reliably than to cognitive work alone. The treatment moves when your client has data showing their predictions about social events were wrong, in their own words, from their own week.

The wrong homework for social anxiety is a generic thought record asking them to identify cognitive distortions. They’ll fill it in on Sunday evening from memory, list “mind-reading” and “catastrophising” because those words are familiar, and the next session you’ll both be looking at a form that doesn’t move anything.

The right homework is a paired prediction-and-outcome form, built around a specific social event already on their calendar. The event is the lever. Don’t ask them to find anxiety in their week. Pick something they were going to do anyway: a work meeting, a drink with a colleague, a phone call to their landlord, a family dinner. Build the form around that.

The prediction section runs before the event. Five fields. What’s the event, what are you most afraid will happen, what’s the likelihood of that happening 0-100, what’s the likely cost if it does happen 0-100, and how distressing do you predict the event will be 0-10.

The outcome section runs immediately after. What actually happened. Did the feared thing happen yes/no. If yes, what was the actual cost 0-100. How distressed were you in the moment 0-10. One short note on what surprised you.

The two sections together produce the data. The fear was 8, the actual was 4. The predicted likelihood of being judged was 75, the actual evidence of judgement was 0. The cost they’d predicted was catastrophic, the cost they observed was negligible. Across five or six experiments on different events, your client builds a personal database of times their prediction system was off, in their own words.

On paper, this falls apart. The prediction gets written Sunday before the event. The outcome write-up doesn’t happen because the event ended at 10pm and your client wasn’t going to dig out a folder when they got home. The data you get back is incomplete.

In my-cbt, you build the prediction-and-outcome form as a two-section worksheet. Assign the prediction due before the event. Assign the outcome due an hour after. Each section is a separate small form on the phone. The client opens both forms from their portal at the right time. They tap a few sliders, type or dictate a sentence, hit submit. Both submissions are timestamped, both live in the case file.

After five experiments, the homework review is the conversation. You point at the gap between predicted and actual on each one. Your client sees their own data showing the predictions were too dark. They no longer need you to convince them of anything. The case file has done it.