Skip to content

Self Affirmation

Principle · All principles
Evidence · 14 claims (6 for, 8 mixed) · 16 studies (7 causal, 3 design, 2 quant-synthesis, 2 review, 1 associational, 1 qualitative), q1–q4 · 5 of 16 report an effect size · 10 claims rest on one study

Conditional relationship

This page owns a brief values-affirmation exercise placed before a threatening evaluation. For a learner who has reason to expect that a coming evaluation could confirm a negative judgement of them (a member of a group negatively stereotyped in the domain, a learner who has already failed in it, a learner about to receive critical feedback on work that matters to them), writing for about 10–15 minutes about a personally important value unrelated to the evaluated domain, shortly before the threat and early in the course, is expected to protect later performance in that domain (general-achievement as course grades, immediate-performance on the assessment) compared with the same writing task about a value unimportant to the learner. The proposed mechanism, from Steele's (1988) self-affirmation theory, is that affirming another valued part of the self makes the threat less consuming, so attention and effort go to the task. The exercise does not teach anything about the domain; it can only help a learner show and keep using what instruction gives them.

The relationship is conditional on four things. Threat: effects are expected only where the learner actually experiences threat; learners not under threat are expected to gain nothing. Timing: before the threat, and early enough to change a trajectory, not after a pattern of failure is set. Framing and privacy: the writing is presented as an ordinary class activity, not graded, not read for quality, and not given only to a group singled out as needing help. Environment: the exercise cannot make up for instruction, feedback or an institution that keep sending the threatening message.

A claim in the wiki tests this relationship, and its result is mixed. A brief values-affirmation writing exercise raised African American seventh-graders' grades in two small field experiments, but a large same-district replication found no effect [~S] records two randomized field experiments with seventh-graders comparing a 15-minute important-value essay with an unimportant-value essay: in the first (Cohen et al. 2006) African American students' fall-term grades rose and the racial gap narrowed by about 40%, with no effect for European American students; in the second (Hanselman et al. 2017), the same team ran the identical protocol with a new cohort of 449 Black and Hispanic students in a district where the previous year's scale-up had found a positive effect (d = 0.15), and found d = -0.07 (95% CI -0.19 to 0.05), with none of the theorised moderators explaining the difference. A designer should therefore expect a small, unpredictable effect, possibly none, and treat the exercise as a low-cost option to run and check locally, not as a reliable gain. Effects on test anxiety from writing about worries, on belonging, and on mindset are separate relationships with their own claims (below).

How the converted neighbours sit beside it. Wise Feedback Across Difference holds how critical feedback is framed for a learner who doubts the giver's intent; affirmation before feedback is a different move on the same threat, and the two can be combined. Motivation holds the diagnosis of which motivational component is low; its belonging component is the nearest relative of this page. Positive Self-Talk holds the words a learner uses at a hard moment during a task; affirmation is done before the task and is about the learner's values, not the next step. Check-ins holds the routine of brief affective touchpoints into which an affirmation prompt can be placed.

Default design

The earlier page's guidance, kept as a concrete design a designer can act on and revise. Each step says what stands behind it; steps without a claim are untested proposals. Doses are from the tested protocol where stated, otherwise proposals.

  1. Decide first whether there is a threat to buffer. Name the evaluation and who might experience it as a threat (a gateway course where a group is stereotyped as weak, a first exam after past failure, critical feedback on a first draft). If you cannot name one, do not run the exercise; use the time for instruction. Evidence: in the tested protocol, students not in the stereotyped group showed no effect (self-affirmation claim [~S]).
  2. Give it to everyone in the class, framed as an ordinary writing activity. Every learner writes; nobody is told it is for students who struggle, or for their group. In the tested protocol the class was randomly split between an important-value and an unimportant-value essay, double-blind (self-affirmation claim [~S]); outside research, give everyone the important-value version. That singling learners out backfires is the earlier page's caution, an untested proposal here.
  3. Run the exercise: 2–3 minutes choosing, 10–15 minutes writing. Show a list of values (relationships with family or friends, creativity, independence, religion or spirituality, membership in a group, humour, music or art, sport, kindness, and an "other" line); the learner picks the one or two that matter most and writes about why they matter and a time they mattered. The writing time is from the tested protocol (15 minutes in class); the list and the choosing step are the earlier page's design, untested.
  4. Do not grade, read or comment on the writing. Writing quality and depth are not assessed; the learner may keep the page. If anything is collected (for a research record or a completion check), say so beforehand and look only at whether a value was named. Untested proposal, from the earlier page's mechanism.
  5. Place the first administration early, before the first major evaluation. In the tested protocol it came early in the fall term (self-affirmation claim [~S]). Proposal: within the first two or three weeks of a course, and at least a day before the first graded test or first critical feedback.
  6. Across a term, repeat it before the next two or three high-stakes points, shorter each time. Session 1: list, choice and 15 minutes of writing. Sessions 2–4 (before later exams or major feedback): 5–10 minutes on the same or a newly chosen value, no list unless the learner wants one. Stop after the gateway assessments of the course; do not turn it into a weekly routine. The earlier page cited repeated administration across a semester in introductory biology and chemistry (Binning et al. 2019, no claim page in the wiki); the number of sessions and their lengths are untested proposals, and the self-affirmation claim [~W] notes in its own Discussion that whether repetition helps, harms or dilutes is open.
  7. Change the threat as well as the learner's response to it. Pair the exercise with feedback that states the standard and the learner's capacity to meet it (Wise Feedback Across Difference), and with assessment practices that keep stakes low where they can be. Evidence carried from a belonging intervention: in a 22-institution trial, a brief online belonging exercise helped only where students' groups were afforded opportunities to belong (Belonging Interventions Improve Outcomes [~M]).
  8. Check what happens, locally. Before and after the term, compare the grades and attendance of the groups you expected to be under threat with the previous cohort or a comparable section, and record that it was run, when, and by whom. Untested proposal; the replication failure above is the reason to check rather than assume.

When a condition is not met: if there is no identifiable threat, skip the exercise; if the learners' main difficulty is anxiety about a test rather than about what a result says about them, the closer move is writing about the worries themselves (row below); if the course or institution keeps signalling that the learner does not belong, work on that first, because the exercise is not expected to hold against it.

Fitting the design to a situation

The three facts that most change the decision:

  • Whether, and for whom, the evaluation is a threat. The tested effect appeared only for students in a group stereotyped in the domain, and not reliably even there. If unstated, ask: which learners might read a poor result here as confirming something about who they are, and what tells you so (past results, who drops out, what learners say)?
  • When the threatening moments are. The exercise is placed before them. If unstated, ask: when is the first graded test, the first critical feedback, the gateway assessment, and how far apart are they?
  • Who will see the writing. The mechanism assumes the writing is private and unjudged. If unstated, ask: will anyone read, grade, store or code what learners write, and do the learners know?
If the brief says… Then change… Basis
Learners are not in a group stereotyped in this domain, and stakes are low Do not run it; spend the time on instruction or feedback A brief values-affirmation writing exercise raised African American seventh-graders' grades in two small field experiments, but a large same-district replication found no effect [~S] (European American students showed no effect in either Cohen study)
Learners are adolescent, in a classroom with one teacher (middle or secondary school) Teacher hands out the sheet at the start of a lesson early in term; 15 minutes of silent writing; collect nothing, or collect sealed and unread self-affirmation claim [~S] (seventh-graders, in class; one positive and one null replication)
Learners are adult, postsecondary (university gateway course) Put the exercise in the first week, online or in the first lecture, before the first quiz; repeat before the midterm and final untested for affirmation with adults in the wiki; the earlier page cites Miyake et al. 2010 (college physics, no claim page). A belonging exercise at college entry is the nearest tested move (Belonging Interventions Improve Outcomes [~M], carried from belonging to affirmation)
Learners are child (under 12) Use a spoken or drawn version: the child picks a picture card of something that matters to them and tells or draws why; 5–10 minutes untested; the tested learners were seventh-graders
Learners are adults in a work setting (workplace-clinical: onboarding, certification, an appraisal) Offer it as a private reflection before the assessment or feedback, never as part of the appraisal record; let the learner choose whether to do it untested proposal
Learners have low literacy, or work in a second language Let them choose from a short illustrated list and write three sentences, or speak into a private voice note, or write in their first language untested proposal
The brief's main problem is test anxiety (worry about the exam itself) rather than what a result says about the learner Replace or follow the affirmation with 10 minutes writing about their worries immediately before the exam, for learners who report high test anxiety; claim no general effect Brief expressive writing before an exam can raise test-anxious students' scores, but across the wider literature its effect on test performance is negligible [~S] (ninth-graders, d = 0.57 for highly anxious students in one study; a later meta-analysis of 21 studies found a negligible effect)
A single session only (one workshop, one exam day) Place it immediately before the evaluation in that session (10 minutes), and claim no lasting effect self-affirmation claim [~S] and Brief expressive writing before an exam can raise test-anxious students' scores, but across the wider literature its effect on test performance is negligible [~S]: one-off writing exercises have small and fragile effects on performance
A fixed short time budget (for example 10 minutes before a test) 2 minutes choosing a value, 8 minutes writing; no discussion afterwards untested proposal; the tested protocol used 15 minutes
online-self-paced, no teacher present Put the exercise in the course onboarding, before the first graded quiz; require only that a value is chosen and some text entered, never a minimum length or a score untested proposal; the earlier page's MOOC onboarding example
Mixed modes: some learners remote, some in the room, or phone-only Give everyone the same prompt at the same point: paper in the room, a form online; on a phone, a tap to choose the value and a few typed lines or a voice note untested proposal
One teacher for many learners, or scarce facilitator time (a volunteer, one tutor checking in briefly) Use a printed or online sheet that runs itself; the facilitator only reads out the instructions and keeps time; there is nothing to mark untested proposal
The writing is private or sensitive (values, religion, family) The learner keeps the sheet or the file; nobody reads it unless the learner chooses to share; if a record is needed, record only "completed" untested proposal, from the earlier page's privacy requirement
The brief names its own routine (a check-in, a reflective journal, an onboarding "why are you here" prompt) Keep that routine; change only the prompt (a value that matters to you, and why) and its timing (before the threatening evaluation) untested proposal
The brief supplies its own list of values or categories (a school's stated values, a community's own terms) Use that list, with an "other" line, rather than substituting the standard one untested proposal
Delivery is by a chatbot or AI tutor Let the learner write; the tool gives the prompt and nothing else, with no praise or evaluation of the answer The agent's praise was indiscriminate affirmation: validation followed about half of participant turns regardless of what the participant said [~W] (a reflection agent, not a values exercise; carried as a caution that AI praise is not self-affirmation)
The threat arises in group work, where learners from a stereotyped group withdraw Run the exercise before groups form, and set roles so each member must contribute Stereotype threat in collaborative contexts can interfere with working memory and reduce meaningful contribution [+W] (a primer's account; the remedy is untested)
The course or institution keeps sending a threatening message (ranking, deficit talk, no one like the learner succeeding) Work on the environment first (feedback framing, low-stakes assessment, visible support); run the exercise only alongside it Belonging Interventions Improve Outcomes [~M] (a belonging exercise helped only where groups were afforded opportunities to belong; carried to affirmation)
A performance or product for a real audience (a recital, a presentation, a portfolio review) Run a short version (5 minutes) shortly before the performance; keep the performance itself whole and unchanged untested proposal
The learner is about to receive critical feedback on a draft Run the exercise before returning the feedback, and frame the feedback with the standard and an assurance of capacity untested proposal; feedback framing is modelled in Wise Feedback Across Difference

Eight of the twenty rows cite a claim. Three (the first two and the single-session row) rest at least partly on the claim that tests values affirmation itself, and that claim's two studies disagree; the others carry a neighbouring finding (belonging, expressive writing about worries, an AI reflection agent, a primer on stereotype threat in groups) to this setting and say so. The other twelve are untested proposals.

Observation, state and explanation

What can be observed: whether the learner completed the exercise and named a value; when it was done relative to the evaluation; the learner's grades or scores on the evaluations that followed; attendance, submissions and dropping out; and what the learner says about the course. "Under threat", "affirmed" and "buffered" are inferred states. A study's group-average effect does not say whether this learner was helped.

Response Plausible explanations Observation that could change the interpretation
A learner leaves the sheet blank or writes one line Did not see the point; the values list does not fit them; distrust of who will read it; low literacy or language Ask privately whether the list had something that mattered to them; offer the spoken or first-language version; repeat the privacy assurance.
The group expected to be under threat does no better than last year's cohort No effect in this setting (as in the large replication); the threat was not present; the threat came from the environment and is still there; the comparison cohort differed Compare with a parallel section in the same term; ask learners how threatening the course felt; look at what else changed.
Learners ask "why are we doing this — is this because of our results?" The exercise read as a remedial message to a group Check how it was introduced and to whom; next time give it to everyone as an ordinary activity.
Grades of the target group improve early but fall again later in term Effect fading; later threats were not preceded by a repeat; new difficulty in the material Look at the timing of repeats against the timing of the drop; check whether the drop is shared by all learners.
A learner writes at length about a domain-related value ("being good at maths") Misread the prompt; for this learner the domain is the value at stake Restate that the value should be from outside the subject; do not correct it as an error.

These rows are proposals, untested with learners as diagnostics. The averages in the claims do not forecast an individual's response, so keep the explanations that remain open.

Evidence and qualifications

The learners, outcomes and comparisons differ across these claims (seventh-graders' grades, college completion, exam scores of anxious ninth-graders), so do not rank them by effect label. No claim in the wiki records values affirmation with adults outside school, with children under 12, before feedback rather than before a term, or with repeated administrations compared with one.

The evidence by question

Further evidence, not yet read against this model

Claims that bear on the page more distantly, each with what it is and how far it bears.

Objective and learner-valued goal

The designer's objective is usually that learners in a threatening course perform and persist as well as their preparation allows (grades in the gateway course, staying in the programme). The learner's goal may be different: to pass this course without being seen to struggle, to prove someone wrong, to belong in the field, or simply not to spend time on something unrelated to the subject. Ask, or watch for it: a learner who sees the exercise as a distraction from revision, or as a sign that they are expected to fail, has a goal that diverges from the design, and that divergence is itself an observation about the threat.

For example, a first-year engineering student from a group under-represented in engineering may want, above all, to be treated as an engineer, not as someone needing support. A values exercise given to the whole class, early, without comment, serves the designer's objective without contradicting that goal; the same exercise given to "students who may find this course hard" contradicts it. Record which way it was given.

What would revise this model?

The expectation should weaken further if more large, preregistered replications with threatened groups find effects near zero, as the one recorded here did; the page would then treat values affirmation as an option without expected benefit and keep only the environment-side steps. It should narrow if effects appear only under conditions that can be named in advance (a particular threat level, timing, or school climate), which the recorded replication looked for and did not find. If repeated administrations do better than one, step 6 becomes a requirement; if they dilute the effect, it should be dropped. If the 7–9-year benefits reported second-hand are confirmed in their sources with effect sizes, the case for early administration strengthens. Do not protect the model by calling every null result "the learners were not under threat" after the fact: decide who is expected to be under threat before the course starts.

Grades in the targeted course, overall GPA, persistence, test anxiety and long-term attainment are separate outcomes; evidence for one is not evidence for the others.

Claims

  • Check-ins — low-stakes affective touchpoints that, like affirmation, surface and regulate learner emotional states before demanding tasks
  • Assessment for Learning — affirmation pairs naturally with low-stakes, feedback-rich assessment cultures that reduce evaluative threat
  • Purposeful Reflection — the values-writing exercise is a structured reflection activity; both rely on brief writing to reframe learner experience
  • Wise Feedback Across Difference — framing critical feedback for learners who doubt the giver's intent; a different move on the same threat
  • Motivation — diagnosing which motivational component is low; belonging is the nearest relative of this page
  • Positive Self-Talk — the learner's own cues at a hard moment during a task, as distinct from values writing before it

Examples

Illustrative

Values-affirmation writing exercise (Cohen et al. protocol) — At the start of a term or before an exam, learners rank a list of values (relationships with family/friends, creativity, independence, membership in a social group, etc.) and spend 10–15 minutes writing about why their top-ranked value is important to them. Control groups write about why a least-important value might matter to someone else. This is the protocol used in Miyake et al. (2010) and Walton & Cohen (2011).

Repeated affirmation in gateway courses — Binning et al. (2019) administered the values essay multiple times across a semester in introductory biology and chemistry courses, finding sustained GPA benefits for Latinx students — evidence that timing and repetition matter more than a single administration.

Affirmation before feedback — Instructors or online platforms present a short values reflection before returning critical feedback on drafts or problem sets, increasing the likelihood that learners engage with rather than dismiss the feedback.

Onboarding in MOOCs and online courses — Some course onboarding sequences include a brief "why are you here / what matters to you" reflective prompt; while not always framed as affirmation, it serves the same buffering function for learners entering a high-failure-rate environment.

Key Sources

  • Steele, C. M. (1988). The psychology of self-affirmation: Sustaining the integrity of the self. In L. Berkowitz (Ed.), Advances in Experimental Social Psychology (Vol. 21, pp. 261–302). Academic Press. doi:10.1016/s0065-2601(08)60229-4
  • Cohen, G. L., & Sherman, D. K. (2014). The psychology of change: Self-affirmation and social psychological intervention. Annual Review of Psychology, 65, 333–371. doi:10.1146/annurev-psych-010213-115137
  • Miyake, A., Kost-Smith, L. E., Finkelstein, N. D., Pollock, S. J., Cohen, G. L., & Ito, T. A. (2010). Reducing the gender achievement gap in college science: A classroom study of values affirmation. Science, 330(6008), 1234–1237. doi:10.1126/science.1195996
  • Walton, G. M., & Cohen, G. L. (2011). A brief social-belonging intervention improves academic and health outcomes of minority students. Science, 331(6023), 1447–1451. doi:10.1126/science.1198364
  • Binning, K. R., Kaufmann, N., McGreevy, E. M., Fouad-Ghalwat, S., Foster, M. J., Johnson, H. E., ... & Browman, A. S. (2019). Changing institutional contexts through multifaceted social-psychological interventions in one community college. Educational Psychologist, 54(3), 177–197.
Page metadata
Field Value
type principle
id self-affirmation
title Self Affirmation
description For learners who expect an evaluation to confirm a negative judgement of their group or of them, a 10–15 minute private essay on a personally important value, given to the whole class before the threat, is expected to protect later grades compared with writing about an unimportant value; one claim tests this, with two positive small field experiments and a null large replication, so effects are small and unpredictable.
status review
canonical True
generated by claude/unspecified on 2026-10-15
trust unverified — no one has confirmed this page's content

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