Supplementary (Online) Tables In this file, we report results in which we used an ordinary least squares path analytic framework to evaluate the moderator effects of pretest working memory, language comprehension, and reasoning on the effect between each intervention condition versus the control group. We conducted three analyses on the comparison accuracy score. In each analysis, one of the three possible moderators was tested as the moderator, while controlling for the other two cognitive/linguistic variables. We ran three analogous models for the word-problem outcome, while also controlling for pretest word-problem skill and the interaction between pretest word-problem score and condition. Sample-based standard scores across the two conditions were employed for all measures. In Table S1, we report results of the analyses for the contrast between EXP versus CON, and below the table, we identify significant moderator effects and classify them as compensatory or prerequisite ability moderators. In Table S2, we do the same for the contrast between WP versus CON. Table S1 Moderation Analyses on Magnitude Comparisons and Word Problems for the Explanations (EXP) versus Control (CON) Conditions Outcome/Moderator Magnitude Comparisons Word Problems Working Memory t (p) Language Comprehension Reasoning Working Memory Language Comprehension Reasoning t (p) t (p) t (p) t (p) t (p) Constant -0.02 (.984) 0.07 (.943) -0.02 (.983) 0.16 (.875) 0.74 (.460) 0.15 (.884) Moderator 0.52 (.603) 1.25 (.213) 2.55 (.011) 1.75 (.082) 1.73 (.087) 0.36 (.717) Condition 8.49 (<.001) 8.47 (<.001) 8.84 (<.001) -0.14 (.885) 0.38 (.703) -.27 (.789) Moderator x Condition -1.06 (.291) -0.81 (.420) -3.75(<.001) -2.54 (.012) 1.44 (.154) -2.80 (.006) Pretest WPs NA NA NA 4.00 (<.001) 4.28 (<.001) 3.92 (<.001) Pre-WPs x Condition NA NA NA -0.40 (.687) -1.07 (.287) -0.27 (.785) Working Memory NA 0.50 (.620) 0.54 (.593) NA 2.20 (.030) 1.92 (.056) Language Comp 1.28 (.202) NA 1.97 (.051) 1.17 (.243) NA 1.89 (.061) Reasoning 2.75 (.007) 2.84 (.005) NA 0.61 (.543) 0.43 (.667) NA Effects Control Variables On Magnitude Comparisons, EXP received relevant intervention via the core multicomponent fraction program and the EXP component; CON only received the schools’ fraction program. Reasoning was a significant a compensatory moderator of the effect of EXP over CON. Reasoning was correlated with outcomes in CON but not in EXP. On Word Problems, both conditions received word-problem instruction through the schools’ fraction program; however, the EXP condition received intervention on fraction understand and procedures (and the focus of the word problems was fraction). Although there was no significant effect between conditions on word problems, reasoning and working memory were significant a compensatory moderators of EXP versus CON. Reasoning and working memory were correlated with learning in CON but not in EXP. Table S2 Moderation Analyses on Magnitude Comparisons and Word Problems for the Word Problem (WP) versus Control (CON) Conditions Outcome/Moderator Magnitude Comparisons Word Problems Working Memory t (p) Language Comprehension Reasoning Working Memory Language Comprehension Reasoning t (p) t (p) t (p) t (p) t (p) Constant -0.01 (.993) 0.12 (.907) -0.14 (.885) 0.62 (.539) 0.83 (.407) 0.71 (.482) Moderator 1.63 (.106) 0.93 (.354) 3.67 (<.001) 2.95 (.004) 1.14 (.257) 2.95 (.004) Condition 5.59 (<.001) 5.62 (<.001) 5.73 (<.001) 6.13 (<.001) 6.30 (<.001) 6.18 (<.001) Moderator x Condition 0.07 (.947) -0.81 (.418) -2.53 (.012) -0.50 (.615) 1.06 (.291) 0.31 (.756) Pretest WPs NA NA NA 3.04 (.003) 3.24 (.002) 3.11 (.002) Pre-WPs x Condition NA NA NA -0.64 (.523) -0.98 (.331) -0.74 (.458) Working Memory NA 1.61 (.110) 1.72 (.088) NA 2.96 (.004) 2.91 (.004) Language Comp 1.07 (.286) NA 1.20 (.231) 0.94 (.351) NA 0.96 (.338) Reasoning 3.26 (.001) 3.33 (.001) NA 3.02 (.003) 2.97 (.004) NA Effects Control Variables On Magnitude Comparisons, WP received relevant intervention relevant explanation via the core multicomponent fraction program; CON only received the schools’ fraction program. Reasoning was a significant a compensatory moderator of the effect of EXP over CON. Reasoning was correlated with outcomes in CON but not in WP. On Word Problems, WP received relevant intervention via the WP intervention component; CON only received the schools’ fraction program. We identified no moderators of the effect between WP and CON.