Many studies have shown that learners are not particularly good at estimating how well they have learned something. In a 2005 paper:
Koriat and Bork found that: ”JOLs (Judgments of Learning) can instill a sense of competence during learning that proves unwarranted during testing.”
But other studies (Nelson and Dunlosky, 1991) have shown that JOLs are more accurate if you ask the students about their learning after a delay.
So, what does this mean for a trainer? If you are using Level 1 surveys to elicit feedback from a course you are going to get more accurate data if you do it after a delay and not immediately after course completion. The delay does not have to be long, perhaps just an hour or so – enough time to clear out working memory.