Hi Qing, i'm wondering how is the data structured for a one-sample t-test.

Let's assume that from one exposure a user can see 10 options from ranker A and 10 from ranker B. Over the course of a week, a user might go thro3gh let's say 3 exposures that have a checkout metric

treatment_ranker control_ranker

exposure 1: 1 0

exposure 2: 0 1

exposure 3: 1 0

at the entity unit level (i.e, consumer_id) do we assign a value of 2 to treatment ranker and 1 to control ranker? Then the difference comes to 1?