AI-generated note
Methods — full extraction notes
Source: Cognitive load and delayed recall — Methods · J. Educ. Psych. (2024) · 116(3) · pp. 412–414
Research question
Methods- Primary RQ: does spaced retrieval practice reduce extraneous cognitive load and improve delayed recall vs. rereading?
- Secondary RQ: does the effect interact with topic complexity?
- Hypothesis 1: spaced retrieval > rereading on delayed recall (one-tailed).
- Hypothesis 2: spaced retrieval < rereading on self-reported extraneous load (NASA-TLX).
Participants
Methods- Sample size: N = 128 undergraduates enrolled in introductory biology.
- Demographics: ages 18-22 (M = 19.4), 62% female, 38% male, recruited at single R1 university.
- Recruitment: course credit incentive; opt-in via study portal; informed consent obtained.
- Exclusions: 4 participants excluded for failing attention check (final N = 124 analyzed).
- Power: a priori power analysis (G*Power) indicated N = 110 for 0.80 power at α = .05, η² = .10.
Experimental design
Methods- Type: within-subjects, fully counter-balanced (each participant did both conditions).
- Conditions: spaced retrieval practice vs. rereading control.
- Pre-registration: design, hypotheses, and analysis plan filed on OSF (osf.io/x42q8) before data collection.
- Random assignment: Latin-square randomization of condition order, balanced across topic complexity.
- Why within-subjects: controls for individual differences in baseline ability; required smaller N.
Materials
Methods- Study passages: 2 biology chapters (cell signaling, ecology), each ~1,200 words, matched on Flesch readability (60-65).
- Retrieval prompts: 5 short-answer questions per passage; appeared 4 times spaced across 2 sessions.
- Distractor task: Sudoku puzzle inserted between study and retrieval, prevented immediate rehearsal.
- NASA-TLX: 6-item cognitive load self-report; administered immediately after each study condition.
Procedure
Methods- Session 1 (Day 0): all participants studied both passages — half via retrieval, half via rereading. Counterbalanced.
- Delay (Day 0-7): no contact with study material; participants returned exactly 7 days later.
- Session 2 (Day 7): free-recall test for both passages; NASA-TLX retrospective rating.
- Duration: Session 1 = 45 min, Session 2 = 20 min; total participant time ~65 min.
Outcome measures
Methods- Delayed recall (primary): free-recall test administered one week after the final study session.
- Scoring: items recalled from a 50-item idea-unit coding scheme; expressed as proportion (0-1).
- Cognitive load (secondary): NASA-TLX self-report immediately after each study session.
- Time-on-task: logged automatically by the testing platform; controlled for in analysis.
Scoring & reliability
Methods- Two independent raters: graduate students blind to condition; double-coded all 256 protocols.
- Inter-rater reliability: κ = .91 — strong agreement (Landis & Koch 1977: 0.81-1.0 = 'almost perfect').
- Coding scheme: 50 idea units pre-specified by content experts; binary present/absent per unit.
- Disagreements: 11 of 256 protocols required adjudication by third coder.
Statistical analysis plan
Methods- Primary test: 2 × 2 repeated-measures ANOVA: study condition × topic complexity.
- α level: .05, two-tailed (despite directional H1, for transparency).
- Effect size: partial η² reported alongside F-test; planned a priori for η² ≥ .06 (small).
- Post-hoc plan: Bonferroni-corrected pairwise comparisons if main effect is significant.
- Sensitivity analysis: models re-run with time-on-task as covariate to rule out time-spent confound.
