Sanrey, Bressoux, Lima and Pansu (2021) introduced an individual-level, variance-based index of the halo effect in teachers' judgements — the tendency of a global impression of a pupil to compress teachers' item-by-item predictions of that pupil's performance — and showed in two field studies that teachers judge more homogeneously than pupils actually perform, and that teachers who are more certain of their judgements show a stronger halo effect. That finding, however, is correlational, and the paper closes with a set of empirically actionable open questions that no later study has taken up: whether the certainty–halo link is causal and malleable (can instruction about halo bias, more analytical confirmation of specific skills, or deliberately lowered initial certainty reduce the variance-based halo index?), which teacher characteristics moderate the link, whether motivational–dispositional constructs such as need for cognition and need for cognitive closure predict halo susceptibility, whether halo bias varies systematically with pupil characteristics such as gender, socioeconomic status and ethnicity, and what the psychometric properties and boundary conditions of the new index itself are. This problem asks for the coordinated empirical programme that answers these questions, taking validation of the index as the measurement foundation for the substantive work on its antecedents and malleability.

Background

When teachers evaluate pupils, a judgement formed in one domain can spill over into judgements of other, conceptually distinct domains: a pupil seen as strong in reading may be judged as strong in spelling, mathematics or behaviour beyond what the evidence warrants. This is the halo effect, first documented in rating data by Thorndike (1920) and analysed comprehensively by Cooper (1981), who distinguished 'true halo' (genuinely correlated attributes) from 'illusory halo' (the rater's global impression inflating inter-attribute correlations). In education, such biases matter because teachers' judgements feed grading, tracking, and referrals; Dompnier, Pansu and Bressoux (2006) showed halo effects in French teachers' cross-subject scholastic judgements, and subsequent field studies confirmed that teachers' ratings of achievement are consistently more homogeneous than pupils' actual achievement profiles.

Studying halo at the level of the individual teacher has been hindered by measurement: the classical indicators — inter-domain correlations of judged attributes, factor analyses showing too few factors, and comparisons of judged with actual variance — are sample-level statistics that cannot say which teacher is more halo-prone. Sanrey et al. (2021) addressed this with a new index built from dichotomous (correct/incorrect) test items. For each pupil, the between-item dispersion of a response profile with kk successes among NN items is computed with the binomial-type formula

s2=k(N−k)N2,s^{2}=\frac{k(N-k)}{N^{2}},

applied both to the pupil's actual answers and to the teacher's item-by-item predictions of those answers; systematically lower variance in the predicted than in the actual profile indicates halo, and averaging the pupil-level differences yields a teacher-level halo score. In two French field samples (Study 1: 25 teachers, 199 third-year pupils; Study 2: 20 teachers, 180 first-year pupils) this index reproduced the signal of the traditional group-level indicators, and in Study 2 — where teachers rated their certainty after every prediction on a 4-point scale — judgement certainty predicted a stronger halo effect even when controlling for the pupil's actual between-item variance and the teacher's accuracy. The interpretation draws on dual-process ideas: a certain judge has little motive to search for additional, discriminating evidence, so certainty fosters heuristic, low-effort processing that homogenizes the judgement.

The evidence is nonetheless correlational: certain teachers may differ from uncertain ones in expertise, familiarity with the pupil, or cognitive style, and the index itself has been examined only on two specific reading assessments with fixed item counts, difficulty distributions, and dichotomous formats. The authors therefore close with open questions, which are the subject of this problem: moderation by teacher characteristics such as expertise; the causal status and malleability of the effect (would an intervention instructing teachers about halo bias, encouraging analytical confirmation of specific skills, or deliberately lowering initial certainty reduce the variance-based halo index?); the generality of the certainty–halo link across motivational and dispositional constructs — need for cognition (the stable tendency to engage in and enjoy effortful thinking) and need for cognitive closure (the desire for a definite answer, which motivates heuristic 'seize and freeze' processing) being the proposed candidates; possible systematic variation of halo with pupil characteristics such as gender, socioeconomic status, and ethnicity, which the studies did not examine; and a formal psychometric evaluation of the index's sensitivity and boundary conditions (behaviour under differing item numbers, varying base-rate difficulty, and non-dichotomous item formats). These components are linked by one scientific objective — characterizing the antecedents, causal structure, and scope of halo bias in teachers' judgements on the individual-level index — with the psychometric evaluation of the index supplying the measurement foundation on which the causal and moderator questions rest.

Problem Statement

The problem is to determine the antecedents, causal structure, and scope of the halo effect in teachers' judgements as measured by the variance-based halo index of Sanrey et al. (2021), through a coordinated empirical programme whose components jointly serve this single objective:

  1. Causal status and malleability of the certainty–halo link: test, with an experimental or intervention design (e.g. instructing teachers about halo bias, inducing more analytical confirmation of specific skills before judgement, or deliberately lowering initial judgement certainty), whether judgement certainty causally contributes to halo susceptibility — that is, whether such interventions reduce the variance-based halo index relative to an adequate control condition, beyond what regression adjustment for accuracy and pupils' actual heterogeneity can show.

  2. Moderators: identify which teacher characteristics — teacher expertise in particular, but also experience and related professional variables — moderate the relationship between judgement certainty and halo susceptibility on the index.

  3. Dispositional predictors: test whether motivational and dispositional constructs, specifically need for cognition and need for cognitive closure, predict halo susceptibility on the index (strengthening or weakening reliance on heuristic processing), beyond judgement certainty.

  4. Pupil-characteristic variation: determine whether halo bias on the index varies systematically with pupil characteristics — gender, socioeconomic status, and ethnicity — i.e. whether some pupil groups systematically experience more compressed (halo-driven) teacher predictions than others at equal actual achievement.

  5. Measurement foundation: formally evaluate the psychometric properties, sensitivity, and boundary conditions of the variance-based halo index itself — its behaviour under differing numbers of items, varying base-rate difficulty, and non-dichotomous item formats — since components 1–4 presuppose a validated individual-level halo measure.

References

  1. Sanrey, C., Bressoux, P., Lima, L., & Pansu, P. (2021). A new method for studying the halo effect in teachers' judgement and its antecedents: Bringing out the role of certainty. British Journal of Educational Psychology, 91(2), 658–675. https://doi.org/10.1111/bjep.12385
  2. Thorndike, E. L. (1920). A constant error in psychological ratings. Journal of Applied Psychology, 4, 25–29. https://doi.org/10.1037/h0071663
  3. Cooper, W. H. (1981). Ubiquitous halo. Psychological Bulletin, 90(2), 218–244. https://doi.org/10.1037/0033-2909.90.2.218
  4. Dompnier, B., Pansu, P., & Bressoux, P. (2006). An integrative model of scholastic judgments: Pupils' characteristics, class context, halo effect and internal attributions. European Journal of Psychology of Education, 21(2), 119–133. https://doi.org/10.1007/BF03173572
  5. Schmidt, F. T. C., Kaiser, A., & Retelsdorf, J. (2023). Halo effects in grading: An experimental approach. Educational Psychology, 43(2–3), 246–262. https://doi.org/10.1080/01443410.2023.2194593
  6. Brunner, S. N., & Karlen, Y. (2026). Teacher judgments of student self-regulated learning – Examining the effect of student and teacher characteristics. Metacognition and Learning, 21(1). https://doi.org/10.1007/s11409-026-09469-x
  7. Taylor, M., McLean, L., & LoCasale-Crouch, J. (2026). Head Start teachers' perceptual accuracy of children's approaches to learning: Relations with teacher experiences and personal characteristics. Acta Psychologica, 266, 106822. https://doi.org/10.1016/j.actpsy.2026.106822
  8. Behrmann, L. (2019). The Halo Effect as a Teaching Tool for Fostering Research-Based Learning. European Journal of Educational Research, 8(2), 433–441. https://doi.org/10.12973/eu-jer.8.2.433
  9. Wetzel, C. G., Wilson, T. D., & Kort, J. (1981). The halo effect revisited: Forewarned is not forearmed. Journal of Experimental Social Psychology, 17(4), 427–439. https://doi.org/10.1016/0022-1031(81)90049-4
  10. Siffert, C., Romanet, F., Desmazières, M., Drault, P., & Gourjon, G. (2025). Cognitive biases in osteopathic diagnosis: a mixed study among French osteopaths. Diagnosis, 12(3), 382–395. https://doi.org/10.1515/dx-2024-0144
  11. Don, H. J., Kim, L. E., & Double, K. S. (2025). When good results mislead: How positive student outcomes lead teachers to overestimate the effectiveness of their assistance. International Journal of Educational Research, 133, 102719. https://doi.org/10.1016/j.ijer.2025.102719