How does providing feedback influence quiz performance?

Question

Working on the design of an online quiz tool, I'm considering what would be the best way to give feedback to the users. It's about multiple choice tests used for knowledge testing of health care workers. Each test has a few dozen questions divided into about four themes.

Research suggests feedback is necessary to prevent negative learning effects from looking at wrong answers (Butler & Roediger). Another study shows that providing positive feedback has more positive affective outcomes than does negative feedback (Martocchio & Webster).

Currently I consider three options:

1. Show the right answer immediately after the user submitted an answer. Pros: gut feeling says it reduces negative learning effects, fun game-like interface opportunities. Cons: Slows down users who just want to go through all questions as fast as possible.
2. At the end of each theme, provide an overview of questions, right answers and the user's performance. Feels like a good balance between speed and feedback.
3. Like 2, but at the very end of the test for all questions. Enables users to finish the test as quick as possible. Don't know if they care about that though.

I've been looking for research data, but only found the studies mentioned above. Is there any data from your experience/industry that suggests one option may be better than the other?

With 'better' I mean that the users feel better while doing the test and/or get better scores.

Answer 1

To the best of my knowledge, option 1 will yield the best learning.

The immediacy principle in operant conditioning is probably one of the early theories to support this.

I've seen this effectively implemented in a few high-profile training sessions where the teachers always insisted on providing immediate feedback, even if it means they stop your presentation every 2 seconds.

The importance of immediate feedback can also be explained from a neuroscience perspective - connections between neurons strengthen (ie, learning/memory) better when related pieces of information are presented in close proximity, else the memory either decay or being overridden by new knowledge.

Answer 2

Before considering any option, you have to also consider the business requirements of the quiz.

If the business requirements say that users must pass 80% of the questions, then doing Option 1 would have users anticipate that they could fail the quiz prior to finishing the remainder of the questions. This could result in a potential emotion of reluctance to finish the quiz if the user feels that he/she has incorrectly answered a number of questions passed the threshold (Ex: User cannot incorrectly answer more than 3 questions before going into a fail state).

Option 1 is best when the quiz is used as a gauge of user knowledge of a particular domain (healthcare) as opposed to evaluating their knowledge.

The other options maybe more suitable as it allows users to gain knowledge through "retrieval practice". Additionally (and obviously), it's beneficial if users remember more of the questions being asked on the quiz - irregardless of the answers entered in.

Choosing between #2 and #3 will likely have to come down to solid interface design, user feedback and a good lab since there are a variety of factors involved like the number of questions per theme and the number of themes.