When the Forward Testing Effect Confounds the Testing Effect: Implications for Theory and Individual Differences

Recall from episodic memory has been shown to enhance learning of both the retrieved information (e.g., relative to a restudy control condition; the testing effect, or TE) and subsequently studied materials (the forward testing effect, or FTE). The TE paradigm may be subject to an FTE confound when training phase testing precedes restudy, attenuating TE magnitude. Across one new cued-recall experiment and five large n cued-recall datasets from the literature, we provide evidence for that confounding effect. We further show that the predictions of the dual-memory model of TE can hold when there is and is not a confounding FTE. If there is no FTE confound, the model equation can successfully predict proportion correct in the test condition across the entire cumulative distribution. When there is an FTE confound, that prediction holds only for the lower portion of the cumulative distribution (i.e., for participants with relatively low task ability or motivation), but we show that the model can still account for proportion correct across the full distribution, provided that the prediction is based on restudy proportion correct for a matched control group that is not subject to the FTE confound. An account of the FTE confound effect is proposed, according to which a difficult test promotes more effective strategy use, particularly among higher ability participants, during subsequent restudy. Implications for testing effect theory and application are discussed. Given the greater information that the cumulative distribution analysis provides, we recommend that approach as a routine supplement to mean proportion correct.