Figure 4.

Correlated gain scores. Correlated gain scores between a training variable and a transfer variable can occur regardless of transfer. (A) Here, a Monte Carlo simulation (N = 1000) shows normally-distributed individual scores (M = 100, SD = 15) on a training task and a transfer task at two consecutive testing sessions (pretest in orange, posttest in blue). The only constraint on the model is the correlation between the two tasks at pretest (here, r = .92), but not at posttest. (B) In this situation, gain scores—defined as the difference between posttest and pretest scores—will be correlated (here, r = .10), regardless of transfer. This pattern can arise irrespective of training effectiveness, due to the initial correlation between training and transfer scores.