Bell Ringer and Exit Ticket Statistics: What the Research Shows
Bell ringers and exit tickets are classroom habits with a lot of folk wisdom and not much direct research. This page collects 28 statistics from the research that sits underneath them: how much class time is lost, what quizzing and feedback do for learning, and what short writing prompts do. Every figure links to the page where it was read, and the caveats are listed alongside the numbers. Last checked October 24, 2026. Compiled by Clay Shumate, a secondary classroom teacher.
Key takeaways
- A classroom-only review of 222 studies and 48,478 students found quizzing raised achievement by an average effect of g = 0.499 (Psychological Bulletin, 2021).
- Quizzes with corrective feedback produced about twice the benefit of quizzes without feedback: g = 0.73 versus g = 0.39 (Rowland, 2014).
- When students got half the items right or fewer on the first test and received no feedback, the testing benefit was about zero (Rowland, 2014).
- Classroom interruptions in five observed high schools cost about 3.3 minutes per class hour, roughly 18 minutes a day (Brown University, 2017).
- The U.S. Department of Education’s practice guide rates using quizzes to re-expose students to key content as strong evidence, but rates using quizzes to find content that needs further study only minimal.
- Direct research on bell ringers and exit tickets is thin. The two items found are a 40-student thesis and a study of student affect, not learning.
Class time and interruptions
The case for a tight start to class is about protecting minutes. The interruption figures come from volunteer high schools observed in one spring, so treat the annual projections as an estimate.
| Statistic | Data year | Source |
|---|---|---|
| Observers in 5 high schools (63 class periods) logged 15.3 interruptions per school day, about 2.8 per class hour. Spring 2017 only, volunteer schools. | 2017 | Brown University (Kraft & Monti-Nussbaum) |
| The same observations showed about 3.3 minutes lost per class hour, roughly 18 minutes a day. That projects to about 54.5 hours (about 9.9 school days) a year, or about 6.7 days if tardy students are excluded. | 2017 | Brown University (Kraft & Monti-Nussbaum) |
| Each interruption cost about 78 seconds on average, and phone calls cost nearly 2 minutes. Over half caused further disruption. Teachers’ own estimates were higher, at about 6.9 minutes lost per 60-minute class. | 2017-18 | Brown University (Kraft & Monti-Nussbaum) |
| The average U.S. public school day was 6.7 hours, up from 6.6 in 2007-08. The school year averaged 179 days and 1,203 hours. This is older data. | 2011-12 | NCES (Digest of Education Statistics) |
Retrieval practice and low-stakes quizzing
A bell ringer that asks students to recall yesterday’s material is a form of retrieval practice. The evidence for that is strong. It also has limits: most early studies were in labs or with college students, and the benefit shrinks when students cannot answer and get no feedback.
| Statistic | Data year | Source |
|---|---|---|
| A classroom-only review covered 222 independent studies with 48,478 students. Quizzing raised achievement by g = 0.499, a medium effect, showing the benefit holds beyond the laboratory. | 2021 | Psychological Bulletin (Yang et al.) |
| A meta-analysis of the testing effect pooled 159 effect sizes from 61 studies. Testing beat restudying by an average Hedges’ g of 0.50. This is mostly lab and college work, with heavy study-to-study variation. | 2014 | Psychological Bulletin (Rowland), hosted copy |
| In the same meta-analysis, tests with corrective feedback gave g = 0.73 (52 effects). Tests without feedback gave g = 0.39 (107 effects). Feedback after a quiz roughly doubles the benefit. | 2014 | Psychological Bulletin (Rowland), hosted copy |
| Null finding: when students got 50% or fewer items right on the first test and received no feedback, the testing benefit was about zero. This suggests easier recall questions work better than very hard ones when feedback is missing. | 2014 | Psychological Bulletin (Rowland), hosted copy |
| In a classic lab study of 120 undergraduates, students who were tested on a passage recalled 56% after one week, while those who restudied recalled 42%. Prose passages and college students, so not secondary. | 2006 | Psychological Science (Roediger & Karpicke) |
| Mixed finding from the same research: restudying beat testing on a 5-minute delay. Testing won only at 2 days and 1 week. Students also felt more confident after restudying. | 2006 | PubMed (Psychological Science) |
| A spacing review synthesized 184 articles, 839 assessments, and 317 experiments. The best gap between study sessions grew as the delay before the final test grew. | 2006 | Psychological Bulletin (Cepeda et al.), via PubMed |
| A review of ten study techniques gave only two the top “high utility” rating: practice testing and distributed practice. Five techniques were rated low, including rereading, highlighting, and summarization. Press release summarizing the review. | 2013 | Association for Psychological Science (on Dunlosky et al.) |
| The practice guide rates “use quizzes to re-expose students to key content” as backed by strong evidence. | 2007 guide | IES / What Works Clearinghouse |
| The same guide rates spacing learning out over time as moderate evidence, and prompting students to ask and answer deep explanatory questions as strong. | 2007 guide | IES / What Works Clearinghouse |
| Caveat: the guide rates “use tests and quizzes to find content that needs further study” only minimal. Using exit tickets to diagnose gaps has the thinnest backing of the quiz recommendations. | 2007 guide | IES / What Works Clearinghouse |
Feedback and formative assessment
Checking what students know is only useful if something happens next. The research on feedback is positive on average and very uneven from study to study.
| Statistic | Data year | Source |
|---|---|---|
| The feedback strand shows about 6 months of extra progress on average, from 155 studies, with a high evidence rating. Secondary pupils gain about 5 months. Digital feedback averages about 4 and verbal about 7. | Current page (2026) | Education Endowment Foundation |
| A feedback meta-analysis covered 435 studies, 994 effects, and over 61,000 participants. The average effect was d = 0.48. About 17% of effects were negative, and variation between studies was very high. | 2020 | Frontiers in Psychology (Wisniewski, Zierer & Hattie) |
| A foundational review found that typical formative-assessment experiments had effect sizes of 0.4 to 0.7, larger than most educational interventions. It is a 1998 review, and later work found more variable results. | 1998 | Black & Wiliam, Inside the Black Box (hosted copy) |
Short writing prompts
Many bell ringers are a few sentences of writing. The research on writing to learn is positive but modest, and it is weaker in grades 6-8.
| Statistic | Data year | Source |
|---|---|---|
| A meta-analysis of 56 experiments (grades 1-12) found that writing about content in science, social studies, and math raised learning, with an average effect of 0.30. | 2020 | Educational Psychology Review (Graham, Kiuhara & MacKay) |
| A review of 48 school-based writing-to-learn programs found a small positive effect on achievement. Metacognitive prompts and longer programs helped. Grades 6-8 and longer assignments lowered the effect. | 2004 | Review of Educational Research (Bangert-Drowns et al.) |
| Note-taking had an average effect of 0.48 against reading-only controls (23 comparisons). Summarizing had 0.33 in middle and high school (11 studies). | 2010 | Carnegie Corporation (Graham & Hebert), hosted copy |
Paper or screen
Whether to run a bell ringer on paper or a device is a common question. The gap is small and depends on the text.
| Statistic | Data year | Source |
|---|---|---|
| A meta-analysis of 54 studies with about 171,055 participants found paper reading beat screen reading for comprehension. Hedges’ g was -0.21, which is small. | 2018 | Educational Research Review (Delgado et al.) |
| In the same meta-analysis, the paper advantage was larger under time limits. It held for informational texts but not for narrative-only texts. | 2018 | Educational Research Review (Delgado et al.) |
Teacher time
Any routine has to fit into a teacher’s week. These figures show why a bell ringer or exit ticket that adds grading time rarely lasts.
| Statistic | Data year | Source |
|---|---|---|
| Teachers reported about 53 hours worked per week, versus 44 for comparable working adults, about 9 more. | 2024 | RAND (State of the American Teacher) |
| A national teacher survey gave median weekly times of 5 hours on grading and feedback and 5 hours on planning. Only 46% of in-building time went to teaching. The survey year is not stated on the page. | 2022 (published) | EdWeek Research Center / Merrimack College |
| Among 2,232 U.S. public K-12 teachers, weekly AI users reported saving an average of 5.9 hours a week, and monthly users 2.9. These are self-reported savings, and only 16% used AI for grading at least monthly. | 2025 | Gallup / Walton Family Foundation |
Research on bell ringers and exit tickets themselves
This is the thinnest part of the evidence, and it should be read as such. Neither item below shows that bell ringers or exit tickets raise achievement across schools.
| Statistic | Data year | Source |
|---|---|---|
| A small bell-ringer study had 40 ninth-grade science students in two classes. Higher-order-thinking bell ringers produced greater content gains than simple-recall ones. Engagement showed little change. Teacher-researcher thesis, no statistics reported. | Undated | Montana State University ScholarWorks |
| A study gave 3-item exit tickets to 1,067 grade 5 students in 17 schools. They worked as reliable measures of enjoyment and perceived difficulty. They measured student affect, not learning. | 2022-23 | NSF Public Access Repository (Shehzad, Recker & Clarke-Midura) |
What this does and does not support
This is a reading of the research, not a finding in any of the studies above. Starting class with a few questions on earlier material matches the strongest evidence here, retrieval practice, as long as students can answer most of them and get feedback afterward. Exit tickets are better supported as a way to practice recall than as a way to find gaps, since the practice guide rates that second use only minimal. The research also does not show that any specific format, such as paper over digital or a particular number of questions, is the one that works. For free examples, see the exit ticket library, the bell ringer creator, and the free 1,000 bellringers download.
How these statistics were gathered
Each figure was read on the page linked in its row, and figures that could not be confirmed on the page were left out. Where a link goes to a hosted copy of a paper or to a press release summarizing it, the row says so. Descriptions are written in plain language and may simplify the source’s wording; the linked page is the authority. Effect sizes (g or d) are standardized measures where about 0.2 is small, 0.5 is medium, and 0.8 is large. If you spot an error or a newer study, please let us know through the site.