Diagnostic analyses show the pooled effect sizes for achievement, attitude and retention are robust and not attributable to publication bias¶

Claim · All claims
Evidence · 1 study · quant-synthesis r3 · q3
Currency · newest study 2016 · not yet checked against its sources · not reviewed by a person

Subclaims¶

q3 i? Publication-status moderator analysis, Egger's regression intercept test and Rosenthal's Fail-safe N all indicated no publication bias. → Kanadlı 2016

Evidence¶

Kanadlı 2016¶

Kanadlı, S. (2016). A meta-analysis on the effect of instructional designs based on the learning styles models on academic achievement, attitude and retention. Educational Sciences: Theory & Practice, 16, 2057–2086. https://doi.org/10.12738/estp.2016.6.0084

q3 · i? · quant-synthesis · r3

Diagnostic analysis of the 30 included studies. Publication-status moderator analysis was non-significant for achievement (Q = 0.016, p = .901), attitude (Q = 1.585, p = .208) and retention (Q = 0.000, p = .992); Egger's intercepts (−0.236, p = .84; 2.440, p = .41; 7.756, p = .09) did not significantly deviate from zero, and Fail-safe N values were large.

"As a result of Rosenthal's Fail-safe N, it was computed that 3512, 409 and 416 studies with the mean effect of zero were needed to nullify the common effect size of the achievement, attitude and retention, respectively."

Discussion¶

Related Claims¶

• The meta-analytic result is robust to publication bias, with a fail-safe number of 33,896.6 null studies needed to nullify the observed effect — related
• Learning-styles-tailored instruction shows a large pooled effect on students' attitudes toward courses (d = 1.113) — related
• Learning-styles-tailored instruction shows a large pooled effect on retention, with inconsistent printed values (d = 1.763 in Table 7; d = 1.290 in the abstract and discussion) — related
• Publication-bias checks indicate the pooled effect is unlikely to result from publication bias — related
• Publication bias is unlikely to explain the consistency of SRL intervention effect sizes — related
• Publication-bias analyses indicate a low, negligible risk of publication bias in the synthesized effect sizes — related
• The pooled self-efficacy–partnership correlation appears robust to the file-drawer problem — related
• Publication bias tests give mixed evidence: Egger's test indicates bias, but fail-safe N calculations lead the authors to conclude no publication bias — related