Delayed JOL bias: multiple text readers are underconfident, rereaders overconfident at one day, and single readers overconfident at one week¶

Claim · All claims
Evidence · 1 study (3 entries) · causal r3 · q3 · i3 large
Currency · newest study 2025 · not yet checked against its sources · not reviewed by a person

Subclaims¶

q3 i3 A significant main effect of reading approach on the bias index (F(2, 180) = 14.54, p < 0.001, ηp2 = 0.14) and a significant interaction with time of test (F(2,180) = 4.44, p = 0.013, ηp2 = 0.05). → Peter Seban 2025 q3 i? On the one-day delayed test, MTR learners were significantly underconfident while RR learners were overconfident (p < 0.01); the group difference was non-significant at one week (p = 0.13). → Peter Seban 2025 (2) q3 i? On the one-week delayed test, the single-reading group was significantly overconfident compared with the MTR group (p < 0.001). → Peter Seban 2025 (3)

Evidence¶

Peter Seban 2025¶

Peter Seban, Kamila Urban, Radovan Sikl. (2025). Comparing the effectiveness of multiple text reading and rereading on knowledge retention and metacognitive accuracy. Instructional Science, 53, 337–364. https://doi.org/10.1007/s11251-024-09686-4

q3 · i3 · causal · r3

Two-way ANOVA on the computed bias index (delayed JOL minus actual score, divided by maximum points) from the Results section. The article prints the approach main effect "F(2, 180) = 14.54, p < 0.001, ηp2 = 0.14" — ηp2 = 0.14 — plus a non-significant time effect (ηp2 = 0.00) and a significant interaction (ηp2 = 0.05).

"This revealed a significant main effect of the reading approach (F(2, 180) = 14.54, p < 0.001, ηp2 = 0.14), a non-significant main effect of time of the test (F(1, 180) = 0.77, p = 0.38, ηp2 = 0.00), and a significant effect of interaction (F(2,180) = 4.44, p = 0.013, ηp2 = 0.05)."

Peter Seban 2025 (2)¶

Peter Seban, Kamila Urban, Radovan Sikl. (2025). Comparing the effectiveness of multiple text reading and rereading on knowledge retention and metacognitive accuracy. Instructional Science, 53, 337–364. https://doi.org/10.1007/s11251-024-09686-4

q3 · i? · causal · r3

Tukey HSD post-hoc comparison of delayed-JOL bias on the one-day delayed test: MTR learners were "underconfident about the knowledge they acquired" while RR learners were overconfident, with a significant difference (p < 0.01). At one week the trend was non-significant (p = 0.13) as rereaders became more accurate.

"learners in the multiple text reading group (MTR) were underconfident about the knowledge they acquired while those in the rereading group (RR) were overconfident, and the difference was significant (p < 0.01, 95% C.I. = [0.04, 0.38])."

Peter Seban 2025 (3)¶

Peter Seban, Kamila Urban, Radovan Sikl. (2025). Comparing the effectiveness of multiple text reading and rereading on knowledge retention and metacognitive accuracy. Instructional Science, 53, 337–364. https://doi.org/10.1007/s11251-024-09686-4

q3 · i? · causal · r3

Post-hoc contrast of bias on the one-week delayed test: the single-reading group "did not expect they would have forgotten as much as they did and were overconfident in their estimates" relative to the MTR group, significant at p < 0.001 with the printed 95% CI.

"We have also found a significant difference in the bias index on the one-week (WD) delayed test between the multiple text reading group (MTR) and the single reading group (1R): learners in the 1R group did not expect they would have forgotten as much as they did and were overconfident in their estimates (p < 0.001, 95% C.I. = [0.13, 0.47])."

Discussion¶

Related Claims¶

• On a one-week delayed test, multiple text reading and rereading both outperform single reading, but on a one-day delayed test the differences are non-significant — related
• Rereading raises judgements of learning after each reading while multiple text reading raises them only after the second reading (significant interaction, ηp2 = 0.07) — related
• Reading approach and test delay each significantly affect knowledge-test performance, with no significant interaction (ηp2 = 0.02) — related
• Multiple text reading shows more stable retention than rereading: non-significantly higher one-week scores and a smaller decline from one day to one week — related
• Delaying judgments of learning after study markedly reduces, but does not eliminate, the foresight bias — related
• Anchoring-and-adjustment explains the underconfidence-with-practice effect, with JOLs anchored between 30 and 50% correct recall — related
• JOL magnitudes decline across successive lists in both JOL groups, reflecting learners' belief that recall decreases with continued new learning — related
• Students' JoL ratings became more consistent over time and evolved from under-rating to over-rating to accurate judgement — related
• Judgments of learning add no new self-assessment information beyond performance estimates for single tasks — related
• Metacognitive monitoring bias and precision modulate when students stop studying relative to their goals — related
• Non-interactive teaching produces stronger under-confidence (metacognitive bias) than restudy immediately after learning — related