61.29% of studies showed positive effects of token reinforcement, but only 6.45% had sample sizes sufficient for generalizable large effects¶

Claim · All claims
Evidence · 1 study · quant-synthesis r2 · q2
Currency · newest study 2022 · not yet checked against its sources · not reviewed by a person

Subclaims¶

q2 i? 19 of 31 studies (61.29%) showed positive results favoring token/extrinsic interventions and 12 (38.71%) had negative effect sizes, with Yager (2008) largest (g = 18.505) but low precision. → Kelly C. Dreger 2022 q2 i? Only two studies (6.45%) had the sample sizes required to claim a large effect was generalizable. → Kelly C. Dreger 2022

Evidence¶

Kelly C. Dreger 2022¶

Kelly C. Dreger, Ed.D. and Steve Downey, Ph.D. (2022). Reinforcement Practicality for Middle School Students: A Meta-Analysis. Valdosta State University. https://eric.ed.gov/?id=EJ1355660

q2 · i? · quant-synthesis · r2

Per-study effect size comparison across the 31 aggregated studies (Table 1, forest plot). The article reports "19 out of 31 studies (61.29%)" positive and 12 negative; Yager (2008) had the largest gain (g = 18.505) with low precision, and only two studies (6.45%) had adequate sample sizes.

"There were 19 out of 31 studies (61.29%) that showed positive results in favor of the use of token and/or extrinsic interventions. There were 12 studies (38.71%) that had a negative effect size."

Discussion¶

Related Claims¶

• The overall effect size for token reinforcement interventions in middle schools is large (g ≈ 1.04) but not statistically significant — related
• Reporting and publication bias pervade the token reinforcement literature: 38.71% of studies reported large significant effects while 29.03% lacked statistical significance — related
• Five moderators of token reinforcement effect sizes: study type, outcome type, effect size direction, publication year, and variance — related
• Aggregated effect sizes differ by outcome type: behavior highest (g = 2.38), performance large (g = 0.91), motivation small and negative (g = -0.39), none statistically significant — related