Analytics: what the responses say so far
Everything on this page is computed in the browser from the responses it can see: the session log stored in this browser, the responses collected by serve.py when the page is served from the study laptop, and any JSON export you load below. Rates are descriptive. With a pilot-sized sample they show whether the instrument records what it should, not whether an effect exists.
Overview
Appropriate reliance by explanation condition
Follow rate on sound advice, override rate on flawed advice (Adjust, Reject or Ask a human adviser), and their combination. Over-reliance is following flawed advice, under-reliance is not following sound advice.
Trust, time and secondary measures
Means of the 1 to 7 ratings and the median decision time in seconds, per condition, split by sound and flawed advice.