The findings Leeteracy is built on, the studies behind them, and how we measure our own work. Each is stated no more strongly than its study supports.
A chart from Roediger and Karpicke, 2006, Experiment 1. Students read a short passage, then either read it again or recalled it once. Tested five minutes later, those who reread it recalled 81% of it, against 75% for those who recalled it. After two days, 54% against 68%. After a week, 42% against 56%.
Each one links to its study. Where a study’s result is narrower than the headline people quote, we use the narrower version.
Students who practiced recalling a text remembered more of it on delayed tests than students who spent the time rereading it.
Roediger & Karpicke (2006). Test-enhanced learning. Psychological Science, 17(3), 249–255. doi.org/10.1111/j.1467-9280.2006.01693.x
In a lab study, college students who practiced applying and analyzing a text scored 72% on questions of that kind after two days, against 46% for students who practiced recalling its facts.
Agarwal (2019). Retrieval practice & Bloom’s taxonomy: Do students need fact knowledge before higher order learning? Journal of Educational Psychology, 111(2), 189–209. doi.org/10.1037/edu0000282
Across ten common study techniques, rereading and highlighting rank among the least effective ways to learn, and practicing retrieval among the most effective.
Dunlosky, Rawson, Marsh, Nathan & Willingham (2013). Psychological Science in the Public Interest, 14(1), 4–58. doi.org/10.1177/1529100612453266
In a university physics course, mixing topics in homework raised median scores on two surprise tests by 50% and 125%, even though students felt they had learned less.
Samani & Pan (2021). npj Science of Learning, 6, 32. doi.org/10.1038/s41539-021-00110-x
Errors made with high confidence are the ones most likely to be corrected once the right answer is shown, which is why Leeteracy asks how sure you are before you answer.
Butterfield & Metcalfe (2001). Journal of Experimental Psychology: Learning, Memory, and Cognition, 27(6), 1491–1494. doi.org/10.1037/0278-7393.27.6.1491
High school students who practiced calibrating with guidelines, especially in groups, predicted their scores more accurately and scored higher on tests.
Bol, Hacker, Walck & Nunnery (2012). Contemporary Educational Psychology, 37(4), 280–287. doi.org/10.1016/j.cedpsych.2012.02.004
In a field experiment with nearly 1,000 high school math students, those who practiced with a ChatGPT-style tutor scored 48% higher on practice problems than classmates using only textbooks, then 17% lower on an exam taken without it. A version built to give hints instead of answers avoided the drop.
Bastani et al. (2025). Generative AI without guardrails can harm learning. PNAS, 122(26). doi.org/10.1073/pnas.2422633122
Since April 1, 2023, the NCLEX has included Next Generation items built on the NCSBN Clinical Judgment Measurement Model: recognizing cues, analyzing them, prioritizing hypotheses, generating solutions, taking action and evaluating outcomes.
We report what we can show, and say what we can’t.
Fall 2026, one pathophysiology course, with Leeteracy optional all term. Grades were matched to usage anonymously, and the numbers below are what that match showed.
Exam 1 points for students who finished 12 or more quizzes, against those who finished 3 or fewer, comparing students with the same Quiz 1 score.
Why +7 and not 11. The bars are raw averages, so they include stronger students doing more. Holding the Quiz 1 score level leaves about 7 points. Working back through missed questions went with about 4 more.
Method. 131 students, grades matched anonymously. A strong link, not proof of cause: a planned comparison next term tests cause. Source: Leeteracy usage data through October 18, 2026.
