Study tips11 min read

How to study for finals in college: a 3-week active recall plan, with the Feynman technique and AI quizzing

Plan your reviews, test yourself with research-backed methods, and use AI to quiz you on your own course notes.

Medly TeamNovember 19, 2026
On this page

The best way to study for finals is to test yourself: turn your notes into questions, answer them from memory, and come back to each topic one, three and seven days later. If you start three weeks before your first final, you have enough time to do this properly for four courses. This guide gives you a day-by-day plan, a worked Feynman technique example with a chatbot playing a beginner, and AI prompts that turn your own notes into practice questions.

Why testing yourself works better than rereading

In a 2006 study, psychologists Henry Roediger and Jeffrey Karpicke gave college students short prose passages to learn. One group read a passage in four separate study periods. Another group read it once, then spent the other three periods writing down everything they could remember, with no feedback.

Five minutes later, the rereaders did better. A week later the result flipped. The students who had tested themselves recalled 61% of the passage, compared with 40% for the group that only reread, according to Washington University's summary. The rereaders also felt more confident that they knew the material. That's the trap: rereading makes your notes feel familiar, and familiar feels like learned.

A 2013 review by John Dunlosky and four colleagues rated 10 common study techniques by the strength of the research behind them. Only two got the top rating.

RatingTechniques
HighPractice testing, distributed practice
ModerateElaborative interrogation, self-explanation, interleaved practice
LowRereading, highlighting, summarization, keyword mnemonic, imagery for text

Source: Dunlosky et al. (2013), summarized by the Association for Psychological Science. Last checked: November 18, 2026.

Distributed practice means spreading your study out over time. A 2008 study of more than 1,350 people found that the best gap between study sessions gets longer the longer you need to remember something. For a final one to three weeks away, that means gaps of a few days. So the plan below brings each topic back 1, 3 and 7 days after you first study it.

How to study for finals in 3 weeks: the day-by-day plan

The plan covers four example finals, taken in this order: Biology, Statistics, Psychology and History. Split each course into two halves ("Bio 1" means the first half of Biology). On a first-pass day, you turn that half into questions and test yourself. Each review tests you on it again. "+3" means three days after its first pass.

DayDateMain taskReviews
1Mon Nov 23First pass: Bio 1None
2Tue Nov 24First pass: Stats 1Bio 1 (+1)
3Wed Nov 25First pass: Psych 1Stats 1 (+1)
4Thu Nov 26Thanksgiving: reviews onlyBio 1 (+3), Psych 1 (+1)
5Fri Nov 27First pass: History 1Stats 1 (+3)
6Sat Nov 28First pass: Bio 2Psych 1 (+3), History 1 (+1)
7Sun Nov 29Catch upBio 2 (+1)
8Mon Nov 30First pass: Stats 2Bio 1 (+7), History 1 (+3)
9Tue Dec 1First pass: Psych 2Stats 1 (+7), Bio 2 (+3), Stats 2 (+1)
10Wed Dec 2First pass: History 2Psych 1 (+7), Psych 2 (+1)
11Thu Dec 3Feynman: Bio and StatsStats 2 (+3), History 2 (+1)
12Fri Dec 4Feynman: Psych and HistoryHistory 1 (+7), Psych 2 (+3)
13Sat Dec 5Catch upBio 2 (+7), History 2 (+3)
14Sun Dec 6RestNone
15Mon Dec 7Practice exam: BioStats 2 (+7)
16Tue Dec 8Practice exam: StatsPsych 2 (+7)
17Wed Dec 9Practice exam: PsychHistory 2 (+7)
18Thu Dec 10Practice exam: HistoryNone
19Fri Dec 11Retest misses: Bio, StatsNone
20Sat Dec 12Retest misses: Psych, HistoryNone
21Sun Dec 13Bio misses only, early nightNone

Dates assume your first final is on Monday, December 14, 2026. For other dates, count back 21 days from your first final. Last checked: November 18, 2026.

Swap in your own courses in exam order. You can download the 3-week plan and edit it in a spreadsheet. If lectures are still running, add the new material to that course's second half. Day 4 is Thanksgiving, so it has reviews only. Days 7 and 13 give you room to catch up when a day slips.

What to do in each study session

First pass: about 2 hours

  1. Skim the slides and notes for that half of the course and list its topics.
  2. Write 15 to 20 questions on them, or use the AI prompt further down.
  3. Close everything and answer each question on paper, from memory.
  4. Check your answers against your notes. Star every question you missed or only half knew.

Review: 20 to 30 minutes

Answer the starred questions again, plus five you got right, without notes. Remove the star from any question you get right twice in a row. Whatever is still starred after the third review goes on your list for the final week.

Practice exam: one per course in the final week

Use a past final if your professor posts one. If not, mix questions from both halves of the course. Time yourself and keep your notes closed. On Days 19 to 21, retest only what you got wrong, and do it again the evening before each final.

Nursing students can use the same loop with application questions.

How to use the Feynman technique for finals, with a worked example

The Feynman technique is named after the physicist Richard Feynman. It checks whether you understand an idea or have only memorized how it's worded. You explain the idea in plain words to a beginner, notice where you get stuck, go back to your notes, and explain it again more simply. The plan uses it on Days 11 and 12, for the topic in each course you feel least sure of.

A chatbot makes a useful beginner because it will keep asking questions a polite friend might let slide. Paste this in first:

I'm studying for a college final. I'm going to explain one idea from my course.
Act as a curious beginner who has never taken this course. Don't explain the idea
to me and don't give me the answer. After my explanation, ask me 3 to 5 short
questions a beginner would ask: about any word I didn't define, any step I skipped,
and anything that seemed to contradict itself. Then ask me to explain it again,
more simply. When I type DONE, quote any sentence of mine that is wrong or misleading.

Worked example: explaining a p-value in intro statistics

  1. First explanation: "A p-value is the chance that your results happened by chance. If p is below 0.05, the null hypothesis is probably false, so there's a 95% chance your hypothesis is right."
  2. The beginner's questions: "What's a null hypothesis? You didn't say." "The chance of what, exactly?" "Where does 0.05 come from?" "If p is 0.03, is there a 97% chance you're right?"
  3. The gap: you can't answer the last question clearly, so you go back to your notes. A p-value is calculated by assuming the null hypothesis is true. That means it can't be the probability that the null is true, which is the mistake in your first explanation.
  4. Second explanation: "Start by assuming nothing interesting is going on, for example that a coin is fair. That assumption is the null hypothesis. The p-value asks: if it were true, how often would chance alone give a result at least as extreme as ours? Flip a fair coin 10 times and the chance of 9 or more heads is 11 out of 1,024, about 0.011. That's below the 0.05 cutoff that researchers set before they start, so we'd reject the idea that the coin is fair. It doesn't mean there's a 98.9% chance the coin is biased."
  5. Turn the gap into questions: add "What does a p-value assume?" and "What can't a p-value tell you?" to your question list for the next review.

When you type DONE, the chatbot points out anything that is still wrong. Check what it says against your lecture notes or textbook, because chatbots can be confidently wrong. If your gap is about which statistics answer which question, revise descriptive and inferential statistics first.

Use AI to make questions, not answers: a prompt for each course

If you ask a chatbot to summarize your notes, you get more to reread. If you ask it for questions, you get something to recall, which is what the research above supports. Check your syllabus first. Many courses set their own AI policy, and that's the rule you follow.

Start with this prompt, then paste in one half of a course's notes:

Below are my notes for [course], [weeks or topics]. Write 15 questions that test
whether I can recall and use this material without looking at my notes:
- 8 short recall questions
- 4 "explain why" questions
- 3 questions that apply an idea to a new example
Use only what is in these notes. Number the questions. Put all the answers at the
end, after a line that says ANSWERS, so I can try the questions first.

Then add one line that suits the kind of final you're taking:

  • Problem-solving courses (statistics, chemistry, physics, accounting): "Add 5 problems like the examples in my notes, with new numbers and full worked solutions."
  • Concept courses (biology, psychology): "Add 3 questions that describe a new scenario and ask which idea from my notes explains it."
  • Essay courses (history, English): "Add 4 possible essay prompts. For each one, I'll write a thesis and three pieces of evidence from memory."

Check every AI answer against your notes before you trust it, and delete any question on something your course didn't cover. You can also download the prompts to keep beside you.

Medly can do this step for you. Upload the lecture slides, readings and notes for each class, and it builds practice tests from your own lectures in the style of your exams. If a slide doesn't make sense, highlight it and ask Medly to explain.

Statistics finals practice questions: hypothesis testing

These five questions build on the p-value example above. Answer each one from memory before you open the answer.

Question 1: (1 mark)

A study reports p = 0.03, using a significance level of 0.05 that was set in advance. Which statement is correct?

A. There is a 3% chance the null hypothesis is true.

B. If the null hypothesis were true, a result at least this extreme would happen about 3% of the time.

C. There is a 97% chance the alternative hypothesis is true.

D. The effect in the study must be large.

Answer

B. A p-value is calculated by assuming the null hypothesis is true, so it can't be the probability that the null or the alternative is true. It says nothing about how big the effect is.

Question 2: (2 marks)

Define a Type I error and a Type II error in hypothesis testing.

Answer
  1. A Type I error is rejecting the null hypothesis when it is true (a false positive). Its probability equals the significance level.
  2. A Type II error is failing to reject the null hypothesis when it is false (a false negative).

Question 3: (3 marks)

A coin is flipped 10 times and lands heads 8 times. Using a one-sided test at the 0.05 significance level, test whether the coin is biased toward heads. Calculate the p-value and state your conclusion.

Answer
  1. Null hypothesis: the coin is fair, so P(heads) = 0.5. The p-value is P(8 or more heads in 10 flips).
  2. There are 2¹⁰ = 1,024 equally likely outcomes.
  3. Ways to get 8 heads: 10C8 = 45. Ways to get 9 heads: 10. Ways to get 10 heads: 1. Total: 45 + 10 + 1 = 56.
  4. p = 56 ÷ 1,024 = 0.0547, about 0.055.
  5. 0.055 is greater than 0.05, so do not reject the null hypothesis. There isn't enough evidence that the coin is biased toward heads.

The usual mistake is using only P(exactly 8 heads) = 45 ÷ 1,024 = 0.044, which wrongly leads to rejecting the null.

Question 4: (2 marks)

A researcher changes the significance level from 0.05 to 0.01 and keeps everything else the same. What happens to the probability of a Type I error and the probability of a Type II error?

Answer
  1. The probability of a Type I error falls from 0.05 to 0.01, because it equals the significance level.
  2. The probability of a Type II error rises, because stronger evidence is now needed to reject the null, so a real effect is more likely to be missed.

Question 5: (3 marks)

A population has a standard deviation of 10. The null hypothesis is that its mean is 50. A random sample of 100 has a mean of 52. Calculate the z statistic and decide, using a two-sided test at the 0.05 level, whether to reject the null hypothesis.

Answer
  1. Standard error = standard deviation ÷ √n = 10 ÷ √100 = 10 ÷ 10 = 1.
  2. z = (sample mean - null mean) ÷ standard error = (52 - 50) ÷ 1 = 2.
  3. The critical values for a two-sided test at 0.05 are 1.96 and -1.96. 2 is greater than 1.96.
  4. Reject the null hypothesis. The two-sided p-value is about 0.046.

Most lost marks come from dividing by the standard deviation (10) instead of the standard error (1).

To practice more questions like these from your own courses, upload your lecture notes to Medly and it will build a practice test from them. It's free to start, and you don't need a card.

Frequently asked questions

How many hours a day should I study for finals?

No study gives a single right number. This plan takes about 2 to 3 hours on most first-pass days and under an hour on review-only days, alongside your classes.

Is rereading your notes a good way to study for finals?

No. In Roediger and Karpicke's 2006 study, students who reread a passage remembered less a week later than students who tested themselves, and Dunlosky's 2013 review rated rereading low.

How do I study for 4 finals in one week?

Start three weeks before your first final and rotate through the courses, so every course gets a first pass, three spaced reviews and a practice exam. Put the courses in exam order so the first final gets its practice exam first.

Can I use ChatGPT to study for finals?

Yes, if your syllabus allows it. Use it to write questions from your own notes and to play a beginner for the Feynman technique, and check its answers against your lecture notes.

Is 3 weeks enough time to study for finals?

Yes, for about four courses if you have your notes in order. If you have five or more finals, or you're far behind on the reading, start a week earlier and add more first-pass days.

What is the Feynman technique?

It's a way to test your understanding by explaining a topic in plain words, as if to a beginner, then going back to your notes wherever you get stuck. It's named after the physicist Richard Feynman.

Try Medly for free