On this page
Most students check answers the same way: solve a problem, flip to the answer key, move on. It feels efficient, and it feels right to catch mistakes quickly. Research on the timing of feedback gives a more interesting answer. Sometimes waiting before you check helps you remember more. This post explains what the studies found and gives you a simple rule for your next practice set.
What does research say about feedback timing?
The honest summary is "it depends", but the dependence is not random. A meta-analysis by Kulik and Kulik (1988) looked at 53 studies. Applied studies with real classroom quizzes usually found immediate feedback more effective. Experimental studies on learning test content usually found the opposite: delayed feedback worked better. Much of the difference came down to how the studies were set up.
Later lab studies tried to work out why delay can help. Butler, Karpicke and Roediger (2007) had people read passages, take a multiple-choice test and get feedback either immediately or after a delay. On a later recall test, delayed feedback led to better performance. The authors put it down to spacing: seeing the correct answer again after a gap is like a second, spaced study session.
Does this hold for school students?
Metcalfe, Kornell and Finn (2009) tested Grade 6 children learning school vocabulary. Delayed feedback produced better final test scores than immediate feedback, and both beat no feedback. With college students, the advantage disappeared once the researchers controlled for the time between the last feedback and the test. So feedback of either kind is far better than none, and delay sometimes adds a bit more.
The most relevant study for problem solving comes from a real course. Mullet and colleagues (2014) changed when students in an upper-level engineering course got feedback on weekly homework: immediately after the deadline, or one week later. Students who got delayed feedback did better on later exams that had new problems on the same concepts. Yet students said they benefited most from immediate feedback. What felt helpful and what actually helped were different.
When should you check immediately?
Delay is not always right. A review by Shute (2008) concluded that good formative feedback should be specific and timely, and that its effect depends on the learner and the task. Here is a practical reading of that for physics, chemistry and maths practice:
Check the first one or two problems straight away when you are learning a new method. If you have a sign convention wrong in mirror problems, you will repeat the same error twenty times. Catch it early.
Check immediately when you are completely stuck. Staring at a blank page for an hour teaches little. Look at the first step only, then continue on your own.
Delay checking for practice on methods you have already learned. Finish the whole set, put it away, and check the next day. The gap gives you a spaced review of every solution.
What does a good delayed check look like?
Delayed feedback only helps if you actually use it. When you check the next day:
Re-solve before you look. Cover your old working and try each problem again quickly. Then compare with both your first attempt and the answer key.
Classify every error. Was it a concept error, a set-up error (wrong force, wrong sign) or an arithmetic slip? Our guide to analysing a mock test uses the same categories.
Write the correction in a sentence. "I forgot that friction acts opposite to relative motion" is more useful than a red tick.
Redo wrong problems from a blank page two or three days later.
A worked example of the rule
Say you have finished the chapter on current electricity and want to practise 15 problems on resistors in series and parallel.
Do problem 1 and check immediately. You find you added 2 Ω and 3 Ω in parallel as 5 Ω instead of using 1/R = 1/2 + 1/3, which gives R = 6/5 = 1.2 Ω. Good thing you caught it now.
Do problem 2 with the corrected method and check it. Correct.
Do problems 3 to 15 without checking. Put the sheet away.
Next evening, re-solve the ones you were unsure about, then check all 13 against the key and log the errors.
This takes no more time than checking after each problem, but it gives you a spaced review of the whole set.
Where can you get feedback on your method, not just the answer?
An answer key tells you whether your number is right, not where your reasoning went wrong. When you are checking a set the next day and cannot see why your method failed, the Fermi Problem Library has worked problems with a step-by-step solver that explains the reasoning rather than printing an answer, so you can compare each step with your own.
For your next practice set, check only the first two problems straight away and leave the rest until tomorrow.
