Consistency and confoundings: Reply to Slamecka.

Two issues are discussed. First, the vertical comparison method favored by Slamecka (1985) is subject to scaling problems that yield inconsistent conclusions, both within and across experiments. The existence of such inconsistencies is illustrated with data drawn from the literature, and the reason for such inconsistencies is illustrated with a hypothetical example. The second issue involves confoundings. Slamecka (1985) notes that the horizontal comparison method requires that degree of learning be confounded with list age. Similarly, the vertical comparison method requires that degree of learning be confounded with the performance level over which forgetting is assessed. The confounding noted by Slamecka is identified with a potential failure of Loftus's (1985) assumption that equal levels of performance imply identical states of the cognitive system. In situations where this assumption does fail, forgetting of high-learning and low-learning lists must occur for qualitatively different reasons, and any comparison of forgetting curves is therefore inappropriate unless made within the context of a strong theory.