Skip to content

Check-ins

Principle · All principles
Evidence · 10 claims (2 for, 7 mixed, 1 against) · 20 studies (9 causal, 3 quant-synthesis, 3 review, 2 associational, 2 qualitative, 1 theoretical), q2–q4 · 2 of 20 report an effect size · 3 claims rest on one study

Conditional relationship

A check-in is a short, recurring prompt that asks a learner to report something about their own state: confidence, confusion, progress, workload, or what support they need. For learners of any expertise or age, in classroom, online-instructor-led or workplace settings, a check-in is expected to help only through what follows it. A report can change the next activity (clarify, regroup, slow down, offer support), and that change may then affect later performance, persistence or the learner's sense of being attended to. A check-in that changes nothing has no route to an effect in this model.

The relationship this page owns is narrower than the one on the formative assessment principle, and differs from it in one way. Formative assessment interprets an aligned task response (an explanation, an application) and uses it to choose a next action. A check-in mostly collects a self-report about the learner's state, often affective or social as well as cognitive, on a routine schedule. So it is one cheap input to the formative loop, and a weaker one wherever the question is what the learner can do: a learner's judgment of their own learning is an observation of that judgment, not of their learning. Where a check-in asks for a short task response, such as an exit question, the formative assessment model applies directly. The check-in element describes the component itself.

No claim in this wiki tests a check-in routine as such. Nothing here compares learners who were checked in with against learners who were not, on any outcome or at any horizon. The model below is assembled from evidence on adjacent relationships: support adjusted to the learner's response, the accuracy of learners' self-judgments, and self-monitoring as a mechanism.

Fitting the design to a situation

Ask for three facts when the brief omits them: what the check-in asks about (understanding, confidence, workload, wellbeing), whether answers are public or private, and what the teacher will do with each kind of answer.

If the brief says… Then change… Basis
The check-in asks learners how well they understood Follow the rating with one short unassisted task on the same content; act on the task, not on the rating alone Judgments of learning are often inaccurate [-M]; Prior knowledge needed for accurate self-assessment [~M]
Learners are new to the subject Expect low accuracy in their self-reports; show a modelled example of judging one's own work before relying on the check-in Self-assessment accuracy is low without training [~M]
A learner reports high confidence and then answers wrongly Correct the answer straight away and explain it; confident errors corrected this way are well remembered High-confidence errors improve retention after correction [+M]
The teacher can respond to individuals (tutoring, small groups) Change the help given in the next activity according to the report and the task response, and record what was given Contingent scaffolding improves learning [~M]
The check-in is about wellbeing or social-emotional state Use a brief private or anonymous form, read daily, and follow up individually with learners who report difficulty Teachers used Forms for well-being check-ins and anonymous voice [~W]; Students' self-assessments of well-being give useful information [~W]
First-year or transition learners, where belonging is a concern Treat a check-in as a way to find out who needs support, not as the support itself; the tested interventions are belonging reframing, not check-ins Belonging interventions improve outcomes [~W]
online-self-paced, with no instructor Turn the check-in into a self-monitoring prompt against a stated goal, with a suggested next step for each answer Self-monitoring improves self-regulation [+W]
Answers will be visible to the class Switch to a private channel; public reports of confusion are likely to be withheld untested proposal
Learners say "fine" every time Act visibly on one report and see whether later reports become more specific untested proposal

Observation, state and explanation

What a check-in yields is a report, made under stated conditions: the prompt wording, whether the response was public or private, the stakes, when it was asked relative to study, and what the learner expected to happen next. The state behind the report (understanding, overload, disengagement, low belonging) is inferred, and several states produce the same report. Silence and "fine" are responses too, and they are the hardest to read.

Response Plausible explanations Observation that could change the interpretation
High confidence, then errors on the task Fluency or familiarity mistaken for mastery; a confident misconception; the prompt asked about a different capability from the task Ask for a short unassisted application of the same capability, preferably after a delay. Compare which items carry confident errors with what was taught most fluently.
Low confidence, then correct performance Accurate awareness of a fragile skill; general anxiety or low self-efficacy; a norm that discourages claiming competence Repeat the task on a new case. Ask privately what the rating meant. Check whether the pattern holds across topics or is specific to this one.
"Fine" or no response, every time Genuinely no difficulty; the channel feels unsafe or public; past reports changed nothing; disengagement Offer a private or anonymous channel and see whether reports change. Act visibly on one report and see whether later reports become more specific. Compare with task performance and attendance.
A request for help that the next session's performance does not reflect Help was given and worked; the learner improved without it; the request was about something other than the task (time, wellbeing) Record what support was actually given and when. Ask what the request was for before attributing a later gain to it.
Reports of overload or stress Task demand; demands from outside the course; low belonging during a transition Ask about time available and other commitments. Compare reports before and after a change in task demand.

These rows are proposals. They have not been tested with learners, and a single probe shifts confidence among the explanations rather than settling one. The probe itself, and acting on the report, can change the state being measured.

Evidence and qualifications

  • Contingent scaffolding improves learning more than fixed or absent support. [~M] This is the evidence for the follow-through. In one-to-one long-division tutoring of fourth and fifth graders (n=40, 8 per condition), fully contingent support beat fixed, moderate, high and no support on both immediate and 1-month tests. In an eighth-grade circulatory-system study, more interactive tutoring matched less interactive tutoring on immediate learning and did better on transfer. A meta-analysis of dynamic assessment found contingent scaffolding beat coaching but did not beat explicit strategy training. The claim page stresses that most effectiveness evidence is one-to-one, that a review found only 8 effectiveness studies, and that contingency across a whole class is hard to achieve. It does not test a self-report check-in as the trigger for adjustment; the support in these studies responded to task responses. Its entries have not been confirmed against their sources.
  • Judgments of learning are often inaccurate [-M] This limits confidence check-ins. In two laboratory experiments, undergraduates who watched a fluent instructor predicted more learning but recalled no more than those who watched a disfluent one, and words in a larger font got higher judgments of learning with no better recall. The bias survived repeated study-test cycles and an explicit warning. The claim page's discussion adds that judgments made right after study tend to be overconfident and delayed ones more accurate. The materials were word lists and a short science video, not classroom courses, and neither study measured what an instructor did with the ratings. Not yet checked against its sources.
  • Self Assessment Accuracy Is Low Without Training [~M] This qualifies the limit above, and says what can change it. In a randomized 2×2 experiment, 80 Dutch pre-university students (mean age about 15) who studied modelling examples of self-assessment rated their genetics problem-solving more accurately than students who did not (ηp²=.10, as printed). A meta-analysis of 51 higher-education studies found the match between self-assessed and teacher marks varied: it was closer in better-designed studies, in advanced courses and in science. So how accurate a check-in's self-report is depends on criteria, training and experience. Neither entry tests a check-in routine, and the entries have not been confirmed against their sources.
  • Self-regulated learning models treat self-monitoring as a core process through which learners notice gaps and adjust strategy [+W] This is the mechanism a check-in would recruit on the learner's side. A review and a theoretical synthesis place monitoring against a goal or criterion at the centre of self-regulated learning, and argue that feedback helps when learners compare current performance with a target. Neither reports a comparison, an outcome or an effect size, so this is a mechanism hypothesis, not a tested effect of check-ins. The marker is lowered from the page's former [+M] for that reason. The claim's discussion notes that self-monitoring is strongest with explicit criteria and a realistic chance to act. Its entries have not been confirmed against their sources.
  • Belonging Interventions Improve Outcomes [~W] This is a neighbour, not a test, and it qualifies the relational aim. Brief belonging-reframing interventions, not check-ins, raised first-year GPA for African American students in one randomized trial (n=92) and first-year full-time completion in a 22-institution randomized experiment (n=26,911). In the larger study the effect appeared only where students' groups were given opportunities to belong. That matches the old page's caution that check-ins do not create belonging by themselves. It does not show that a check-in raises belonging. Both entries were checked by the judge against abstracts.

Keep the learners, setting, what was compared with what, the outcome and its horizon for each of these when transporting them, and do not rank them: they answer different questions. No entry here reports a standardized effect for anything resembling a check-in.

The evidence by question

Further evidence, not yet read against this model

Claims this page cited before the 2026-10-02 rewrite that are not used in the model above. They bear on the relationship, but none has yet been re-read against it; each says how far its own sources have been checked.

Objective and learner-valued goal

The designer's objective for a check-in is usually information: which learners need what, soon enough to act. The learner's valued goal may be quite different: being noticed, getting help without being singled out, or not being asked at all. Ask what the learner wants from the routine and why, record it separately from the designer's objective, and treat agreement as something to find out rather than assume. A learner who values privacy will give "fine" on a public check-in whatever their state.

For example, in an online instructor-led course, a designer may want a weekly 1–5 confidence rating per topic to decide where to reteach. A learner working full-time may value something else: a way to say "I'm behind this week and need the deadline moved" without explaining in front of peers. A confidence scale cannot carry that message, and a low rating from that learner may reflect lost time rather than misunderstanding. A design serving both might pair a private free-text "what would help this week" with a short task item on each topic. The confidence rating would then be read beside the task item, not instead of it. These are local design choices; no claim here establishes them.

What would revise this model?

The model would gain support from a comparison of matched groups with and without a check-in routine, where instructors acted on the reports, showing a difference on a stated outcome at a stated horizon. It should weaken if such comparisons find no difference even when follow-through is documented. It should also weaken if acting on self-reports does no better than acting on short task responses alone, since a check-in would then add cost without information. If trained or delayed self-reports turn out to track performance closely in a given setting, the warning against reading confidence as mastery can be relaxed there. If learners' reports become less candid once they see them acted on, for instance because acting meant regrouping them publicly, the follow-through itself needs redesign.

An immediate change in what learners report, a change in engagement, later performance, and persistence are separate outcomes, and evidence for one does not establish the others. The present evidence does not establish an effective check-in frequency, prompt or format, or any benefit for belonging from check-ins.

Claims

  • Formative Assessment — check-ins are one lightweight way to gather evidence during learning rather than after it.
  • Flexible Grouping — check-in results can inform regrouping by need, confidence, or pacing.
  • Immediate Feedback — the value of a check-in rises when it leads to prompt response.
  • Building Empathy — some check-ins help learners understand peers' current experiences and constraints.

Examples

  • Confidence meters: Learners rate readiness or understanding on a 1-5 scale before or after a task.
  • Opening prompts: A meeting begins with a quick response to "What is one thing you need today to be successful?"
  • Exit check-ins: Learners note one point of confusion, one success, and whether they want follow-up.
  • Private digital pulse checks: In online environments, short forms or chat reactions let learners signal status without speaking publicly.

Key Sources

  • Champion, T. D. (2015). Scheduled check-ins increase student completion of assignments in an online non-majors science course at an HBCU. SoTL Conference.
  • Fischer, J., & Alpert, A. (2019). Promoting intern resilience: Individual chief wellness check-ins. The Journal of Teaching and Learning Resources. doi:10.15766/mep_2374-8265.10848
  • Kluch, S. (2020). Leading remotely: What managers need to keep teams engaged. Gallup.
Page metadata
Field Value
type principle
id check-ins
title Check-ins
description A brief, recurring prompt for learners to report their state can guide the next instructional move, but only when the report is read as uncertain evidence, checked against performance where accuracy matters, and acted on.
status review
canonical True
generated by claude/unspecified on 2026-10-02
trust unverified — no one has confirmed this page's content

Frontmatter is the page's machine-readable record — see the Schema & Guide.