JoL–mastery accuracy trajectories over the five weeks rose in the elicit condition, declined in the none condition, and were mixed in the tell condition¶

Claim · All claims
Evidence · 1 study (3 entries) · causal r1–r2 · q2 · i2–i3
Currency · newest study 2026 · checked against its sources 2026-10-20 (automated, 3 of 3 entries) · not reviewed by a person

Subclaims¶

q2 i2 In the elicit condition, JoL–mastery correlation increased consistently across three time periods, reaching r = 0.482 (p < 0.001) in the third. → Graf 2026 q2 i? In the none condition, the JoL–mastery correlation declined over time, from r = 0.428 (p = 0.026) to r = 0.110 (p = 0.537). → Graf 2026 (2) q2 i3 In the tell condition, correlations varied greatly, turning significantly positive only in the third period (r = 0.571, p = 0.003). → Graf 2026 (3)

Evidence¶

Graf 2026¶

Graf, L., Bassner, P., Anzinger, M., Dietrich, F., Krusche, S., & Poquet, O. (2026). Interactive learning dashboards: rethinking learning visualisations as engagement tools. Education and Information Technologies. https://doi.org/10.1007/s10639-026-14082-1

q2 · i2 · causal · r2

Figure 6 analysis of JoL–mastery correlations over three even time bins: elicit rose from r = 0.332 (n = 32) through r = 0.378 (n = 19) to "the correlation strengthened further (r = 0.482, p < 0.001)" (n = 53).

"We observe a positive change over time of JoL accuracy under the elicit condition. In the elicit condition, the correlations between JoL and sys-tem's assessment show a consistent increase across the three time periods."

Graf 2026 (2)¶

Graf, L., Bassner, P., Anzinger, M., Dietrich, F., Krusche, S., & Poquet, O. (2026). Interactive learning dashboards: rethinking learning visualisations as engagement tools. Education and Information Technologies. https://doi.org/10.1007/s10639-026-14082-1

q2 · i? · causal · r1

Figure 6 analysis: the none condition fell from r = 0.428 (p = 0.026, n = 27) to r = 0.176 (p = 0.221, n = 50) and further to r = 0.110 (p = 0.537, n = 34).

"The none condition presented a decline in correlation over time suggesting poor support for the accuracy of the student's judgement of learning."

Graf 2026 (3)¶

Graf, L., Bassner, P., Anzinger, M., Dietrich, F., Krusche, S., & Poquet, O. (2026). Interactive learning dashboards: rethinking learning visualisations as engagement tools. Education and Information Technologies. https://doi.org/10.1007/s10639-026-14082-1

q2 · i3 · causal · r2

Figure 6 analysis: tell-condition correlations were negative in periods one and two (r = −0.419, p = 0.084; r = −0.076, p = 0.654) before "the correlation turned significantly positive (r = 0.571, p = 0.003)" in period three.

"However, in the third time period ( n = 25), the cor- relation turned significantly positive (r = 0.571, p = 0.003)."

Discussion¶

Related Claims¶

• Students using the eliciting agent showed JoL ratings correlated with system learning metrics, strongest for mastery — a broader claim this one bears on
• Tell and no-agent conditions showed no significant JoL–confidence correlations, with mixed results for progress and mastery — related
• Students' JoL ratings became more consistent over time and evolved from under-rating to over-rating to accurate judgement — related
• Most students submitted JoLs out of curiosity about the system rating, often without attending to the LA graphs, and reported behaviour change more from JoL than from the agent — related
• In a real programming course, students with an eliciting LLM agent on a learning dashboard wrote more reflection messages than students with a telling agent — related