TABLE 1.

Coding Scheme for Estimation of Effects

Code and Value Label	Examples of Estimation Effects
1: High estimation	Example 1 Results include the statement that the effect was nonsignificant but no quantitative data was given. Lower estimate was entered as 0 and higher estimate was entered as the largest possible r that would be nonsignificant given the degrees of freedom for the test.
	Example 2 Results reported that the effect was significant, but no quantitative data or test statistics were reported. The lower estimate was the smallest possible significant effect given the degrees of freedom for the test. The higher estimate was left missing, because there is no reasonable upper limit.
2: Moderate estimation	Example 1 Beta weights for predicting response to intervention from IQ or initial ability and from additional covariates such as phonological awareness and rapid naming are reported, but the correlations among the predictors is not included. R2 and r are computed by using correlations from large population studies as best guess for intercorrelations among predictors.
	Example 2 F statistic or change in R2 was reported, but the direction of the effect was unknown. We always coded a positive effect.
3: Some estimation	Example 1 In two studies the authors divided subjects into groups based on the amount of growth they showed in response to the intervention (Case, Speece, & Molloy, 2003; Vellutino, Scanlon, & Lyon, 2000). The authors also reported the IQ means, standard deviations, and sample sizes for each group. With this data it is possible to compute an eta2 for the linear relation between the ordered groups and the ability measure. We calculated sums of squares within, the sums of squares between, and the sums of squares due to the linear contrast only per Maxwell and Delaney (1990) and then formed the ratio of the sums of squares linear over the sums of squares total to arrive at an eta2 for the linear contrast.
4: Slight estimation	Example 1 The correlation matrix was given in the article, and we input the matrix into SAS (Arthur, Bennett, & Huffcutt, 2001) to compute the R2 change between models. To the extent that the reported correlations are not as precise as raw data, this approach might result in a small amount of misestimation.
5: No estimation	Example 1 Correlation coefficient was reported in the study.
	Example 2 Data was sent to our team by the authors of the study, and we analyzed it to obtain effect sizes.