Teacher Development and Mentoring
Teacher Action Research Cycles: A Practical Guide
Teacher action research is a planned inquiry into a bounded question about the teacher’s own practice. A cycle links a question, possible explanations, baseline, defined change, time period, evidence, analysis, limits, and next decision. It can improve the basis for a local instructional decision; the design determines what may be concluded.
First identify what kind of activity you are doing
Names such as action research
do not determine legal or institutional status. Purpose, methods, data, participants, risk, and dissemination plans matter. Use the local approval route before collecting or sharing information.
| Activity | Typical purpose and record | Required check |
|---|---|---|
| Professional reflection | A teacher records observations about a lesson to plan ordinary instruction; no systematic student data set or sharing is assumed. | Follow school rules for notes and student information. Reflection is not exempt from confidentiality duties. |
| Internal instructional improvement | Authorized staff systematically review minimum necessary classroom evidence to improve a local process or decision. | Confirm supervisor, curriculum, assessment, privacy, record-retention, and other local requirements before collecting data. |
| Public dissemination | Findings, examples, records, presentations, posts, or manuscripts are shared beyond the authorized internal audience. | Seek review before planning the sharing. De-identification alone may not settle consent, ownership, policy, or human-subjects questions. |
| Formal research | A systematic investigation may be designed to develop or contribute to generalizable knowledge, subject to the applicable definition and jurisdiction. | Use the organization’s research or ethics office. Only the authorized office should determine whether review, exemption, consent, or other requirements apply. |
Route privacy and approval before the cycle
- Write the purpose, participants, data, and intended audience. Include whether any recording, survey, identifiable work, sensitive topic, comparison group, or public sharing is planned.
- Ask the designated local contact. This may be a principal, district research office, institutional review board or research ethics committee, privacy officer, records officer, or program lead. Do not decide your own exemption.
- Minimize data. Collect only what the question requires. Prefer ordinary work and aggregate or coded records when approved; do not collect names merely for convenience.
- Document access and retention. State where records will be stored, who may see them, when they will be deleted, and how approved sharing will protect students.
- Confirm consent, assent, accessibility, and opt-out requirements. Ordinary grading authority does not automatically authorize research participation or public use of work.
In the United States, the U.S. Department of Education’s FERPA resources address education records, while the Office for Human Research Protections decision charts support institutional determinations under the federal human-subjects rules. Other countries and organizations use different requirements.
Build the full cycle
| Element | Planning question | Common problem |
|---|---|---|
| Question | For which learners, task, condition, and observable outcome? | How can I improve engagement?is too broad to measure or act on. |
| Alternative explanations | What else could explain the current pattern or any later change? | Assuming one cause before baseline evidence is reviewed. |
| Baseline | What existing, comparable evidence describes the starting condition? | Using a memory, impression, or non-comparable task as the baseline. |
| Change and time | What will change, what remains stable, and for how long? | Changing several routines at once or ending only when results look favorable. |
| Evidence and minimization | What minimum evidence answers the question, and who may access it? | Collecting names, surveys, recordings, or extensive records without a defined need and approval. |
| Missing data | What will count as missing, and how will absence, incomplete work, or schedule changes be recorded? | Treating missing evidence as zero or silently removing it. |
| Analysis | What comparison or coding rule will be used before seeing the result? | Selecting only favorable work or changing the success rule afterward. |
| Result and limit | What pattern appeared, how large and consistent was it, and what constrains interpretation? | Calling any increase proof that the change caused improvement. |
| Next cycle | Retain, revise, stop, or test an alternative? What uncertainty remains? | Repeating automatically or hiding a null or negative result. |
Copyable blank cycle planner
Cycle title and dates:
Local approval contact and decision:
Internal audience and any planned dissemination:
1. Bounded question (learners, task, condition, outcome):
2. Why this question matters for an instructional decision:
3. Plausible explanations before the change:
4. Baseline task, dates, and evidence:
5. Defined instructional change:
6. What will remain stable:
7. Duration and decision date:
8. Minimum evidence and measure:
9. Students or work included:
10. Access, accommodations, and participation conditions:
11. Data fields collected; identifiers excluded or coded:
12. Storage, authorized access, retention, and deletion:
13. Missing-data rule:
14. Analysis and success rule set in advance:
15. Departures from plan or contextual events:
After the cycle
16. Result, including size and consistency of the pattern:
17. Missing data and whose evidence may be underrepresented:
18. Alternative explanations still plausible:
19. Limitation on the conclusion:
20. Null or negative result, if any:
21. Decision: retain, revise, stop, or seek more evidence:
22. Next-cycle question and one planned change:Worked completed planner
This constructed example demonstrates how the fields connect. Its numbers are illustrative, not reported student results and not evidence that the change works elsewhere.
| Planner field | Illustrative completion |
|---|---|
| Question | During four comparable grade 7 science evidence tasks, does adding a two-minute private written rehearsal before discussion change the proportion of present students who cite a data value in their individual exit response? |
| Alternatives | Differences may reflect task difficulty, attendance, prior knowledge, reading access, teacher prompting, or growing familiarity with the response format rather than rehearsal. |
| Approval and privacy | Illustrative route: principal confirms an internal instructional-review plan; existing exit responses are scored without names in the analysis table; no recording, survey, or external sharing; coded file is stored in the approved drive and deleted after the recorded decision date. An actual teacher must obtain the applicable local determination. |
| Baseline | Two prior comparable prompts without rehearsal. Record for each present student whether the exit response includes at least one accurate value from the supplied table. Keep the two task totals separate. |
| Change and time | Teach the same response routine, then add two minutes for every student to draft privately before optional pair discussion. Use it for the next two comparable tasks across two weeks. |
| Minimum evidence | Present-student count, number of scorable exit responses, number citing an accurate value, task identifier, and a note about schedule or access disruptions. No student names or demographic fields in the analysis file. |
| Missing-data rule | Report absent students and blank or unscorable responses separately. Do not count absence as failure or carry a student’s result from another task. |
| Analysis set in advance | For each task, divide responses with an accurate cited value by scorable responses. Display all four task proportions and counts. Review the work for misread values. Do not combine tasks if prompt demand differs materially. |
| Illustrative result | Baseline tasks: 14/22 and 16/23 scorable responses cited an accurate value. Change tasks: 16/22 and 17/24 did so. The second change-task proportion is close to the second baseline proportion; the pattern is small and inconsistent. |
| Missing data | One baseline response and two change responses were blank; attendance differed by task. The plan did not collect reasons, so it cannot explain the missing responses. |
| Limitation and alternatives | Only four tasks were compared, prompt difficulty was not equated, and the same students became more familiar with the routine over time. The example cannot isolate rehearsal as the cause. |
| Null-result record | The planned pattern, an increase that appeared across both change tasks, did not occur. Record this as a null result, not as hidden improvement. |
| Decision and next cycle | Do not claim the rehearsal changed exit-response evidence. Retain it only if it has a separate instructional purpose and acceptable burden. A next cycle could compare whether explicitly modeling how to select a data value reduces misread citations, using the same privacy route and predefined measure. |
Analyze without losing the cases
Begin with a table that retains task dates, counts, and missing evidence. Inspect student work using the rule set in advance, then calculate a proportion, mean, median, distribution, or coded pattern that fits the question. Small local samples often need the counts beside percentages. If work is paired across time and the approved plan permits coding, inspect whether the same learners changed rather than relying only on class averages.
Look for implementation departures, task differences, access barriers, absences, new supports, schedule interruptions, and evidence that contradicts the expected result. A second evidence source can test an interpretation when it is necessary and approved; collecting more kinds of data is not automatically better.
Write the result in five parts
- Result: State the observed pattern with counts or other defined evidence.
- Comparison: Relate it to the baseline and the success rule set before analysis.
- Missing evidence: State what was absent or unscorable and who may be underrepresented.
- Limitation: Name the main design or implementation constraint.
- Decision: Retain, revise, stop, or run a specifically justified next cycle.
A null result can be useful: it may rule out a proposed explanation under the tested conditions, show that the measure was insensitive, or identify a problem in implementation. State which interpretation is supported and which remains uncertain. Do not recast a null result as success merely because the process prompted reflection.
Close one cycle before starting another
Record the question, decision rule, evidence, result, missing data, limitations, and approval conditions in one place. If the next cycle changes the audience, data, participants, recording method, or dissemination plan, return to the approval route. A series of cycles remains local evidence unless a stronger design and authorized review support a broader claim.
Sources
- ERIC record EJ1134554, Classroom Action Research: a teacher-education example of systematic classroom inquiry and reflection.
- U.S. Office for Human Research Protections, Human Subject Regulations Decision Charts: institutional review routing under U.S. federal rules.
- U.S. Department of Education, Family Educational Rights and Privacy Act resources: education-record privacy.
What changed: This revision adds a complete cycle, approval and privacy routing, blank and worked planners, explicit null-result handling, and a next-cycle decision record.