Table 3. Intervention effects on the flanker task and academic performance.
Intervention | Control | |||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|
Outcome | n = | Baseline | Follow-up | Within-group change | n = | Baseline | Follow-up | Within-group change | Adjusted difference in changea | 95% CI | p-value | ConditionalICC |
Flanker task | ||||||||||||
Accuracy (%) | ||||||||||||
Congruent trials | 180 | 94.8 (6.1) | 96.7 (5.2) | 1.8 (5.2) | 404 | 94.9 (6.2) | 96.9 (4.3) | 2.0 (5.5) | -0.3 | -1.0–0.4 | 0.36 | NA |
incongruent trials | 180 | 80.9 (11.9) | 86.6 (10.1) | 5.7 (9.3) | 404 | 83.2 (11.9) | 87.0 (10.4) | 3.8 (10.3) | 0.8 | -0.7–2.3 | 0.31 | NA |
Interference score | 180 | 14.0 (9.5) | 10.1 (7.6) | -3.9 (7.9) | 404 | 11.8 (9.2) | 10.0 (8.6) | -1.8 (9.1) | -1.0 | -2.3–0.3 | 0.13 | NA |
Reaction time (ms) | ||||||||||||
Congruent trials | 180 | 470.2 (74.6) | 458.9 (61.7) | -11.3 (42.7) | 404 | 461.9 (64.1) | 449.8 (61.2) | -12.1 (50.2) | 2 | -5–10 | 0.57 | NA |
incongruent trials | 180 | 557.4 (92.4) | 534.4 (70.8) | -23.0 (54.9) | 404 | 541.2 (81.5) | 516.6 (70.3) | -24.7 (59.1) | 6 | -2–15 | 0.16 | NA |
Interference score | 180 | 87.2 (41.7) | 75.4 (31.1) | -11.7 (34.0) | 404 | 79.3 (35.3) | 66.7 (28.3) | -12.6 (32.5) | 5 | 0–9 | 0.03 | NA |
Academic performance | ||||||||||||
Mathematics skillsb | 191 | 19.5 (10.2) | 22.1 (10.9) | 2.6 (5.4) | 419 | 20.3 (9.9) | 22.8 (11.1) | 2.4 (4.9) | -0.2 | -1.6–1.1 | 0.73 | 0.08 (class) |
Baseline, follow-up and within-group changes are unadjusted values while between-group comparisons are adjusted. Values are mean (SD) unless stated otherwise. Mathematics performance can range from 0–50 points. RT: reaction time, CI: confidence interval, ICC: Intraclass correlation coefficient conditional on covariables. NA: not applicable. NOTE: A negative difference in change for accuracy means worse performance by intervention schools (smaller increase in accuracy from baseline to post-intervention). A positive difference in change for RT means worse performance by interventions schools (smaller decrease in response speed from baseline to post-intervention). A positive difference in change in interference scores means worse performance by intervention schools (smaller decrease in interference score).
a Estimates are the unstandardized beta-coefficients (control coded as zero, intervention coded as 1) from linear mixed models with the change in outcome adjusted for baseline value of outcome, gender and grade level. If a random effect did not improve the model fit, ordinary least squares regression was used.
b Between-group difference in change is further adjusted for class cluster as a random effect.