Learners are substantially overconfident when evaluating the quality of their self-generated examples¶

Claim · All claims
Evidence · 2 studies · 2 causal r2–r3 · q2–q3 · i3 large
Currency · newest study 2023 · checked against its sources 2026-10-20 (automated, 2 of 2 entries) · not reviewed by a person

Subclaims¶

q3 i3 University students showed substantial overconfidence (mean bias 22.9%) and inaccuracy (absolute accuracy 32.4%) in evaluating their self-generated examples. → Linda Froese 2022

q2 i? Learners overestimate the quality of their self-generated examples relative to expert ratings, especially for poor examples. → Linda Froese 2023

Evidence¶

Linda Froese 2022¶

Linda Froese, Julian Roelle. (2022). Expert example standards but not idea unit standards help learners accurately evaluate the quality of self-generated examples. Metacognition and Learning. https://doi.org/10.1007/s11409-022-09293-z

q3 · i3 · causal · r3

In the 2×2 factorial experiment (N = 131 university students), all learners generated examples for eight declarative concepts and rated them no/partial/full credit. The article reports "a mean bias of 22.9%" significantly greater than zero, d = 1.20, and overall absolute accuracy of 32.4%.

"Overall, they were overconfident with a mean bias of 22.9% (SD = 19.1%), which was significantly greater than zero, t(130) = 13.75, p < .001, d = 1.20."

Linda Froese 2023¶

Linda Froese, Julian Roelle. (2023). Expert example but not negative example standards help learners accurately evaluate the quality of self-generated examples. Metacognition and Learning. https://doi.org/10.1007/s11409-023-09347-w

q2 · i? · causal · r2

In the present experiment's descriptive results, even learners provided with expert example standards overestimated examples rated 0% by experts by ca. 57%, replicating the substantial overconfidence pattern reported by Froese and Roelle (2022) and Zamary et al. (2016).

"Notably, like in Froese and Roelle (2022), the expert example standards did not yield fully accurate judgments. For instance, the learners still overestimated their poor examples, which were credited with 0% by the expert raters, by ca. 57%."

Discussion¶

Related Claims¶

• Providing expert example standards reduces bias in learners' evaluations of self-generated examples — related
• Learners partly use example length as a cue when evaluating their self-generated examples — related
• Readers in all conditions except sketching were overconfident, with the most overconfidence in the photograph condition and the least in the sketching condition — related
• In one 2×2 experiment with 120 undergraduates, an Assessment-Performance Calibration Activity (APCA) gave the most accurate self-assessment and least overconfidence, and adding a feedback-literacy script did not improve calibration further — related
• Prior Knowledge Needed For Accurate Self Assessment — a broader claim this one bears on
• Full reasoning traces worsened calibration bias and Brier scores relative to both other conditions — related
• Sketching during reading yielded the most accurate comprehension monitoring, with significantly less absolute error than photograph, no-image, and animation conditions — related
• SRSD instruction improved students' accuracy in self-evaluating their texts' quality, although overestimation persisted — related
• Idea unit standards do not enhance judgment accuracy in evaluating self-generated examples — related
• Negative example standards do not reliably improve judgment accuracy and significantly decrease relative accuracy in evaluating self-generated examples — related
• Providing a rubric reduces bias in self-assessment, moving judgments closer to expert ratings — related
• Generating examples produces overconfident judgments of learning, especially for principles and procedures — related