Controlling, Confounding, and Construct Clarity: Responding to Criticisms of 'Revisiting the Marshmallow Test' by Doebel, Michaelson, and Munakata (2020) and Falk, Kosse, and Pinger (2020)

Not applicable - reply article

Not applicable.

The reply that clarifies what is and is not in dispute, which is what makes it worth recording.

Concedes partial replication - correlations with adolescent academic achievement were smaller than Shoda et al. reported but still positive and significant, indicating at least partial replication of the achievement correlations. Maintains that in contrast to the original, correlations with BEHAVIOURAL outcomes were null. Defends the controls as the whole point of the exercise, namely distinguishing a predictive correlation from a malleable causal lever, rejects the multicollinearity objection, and notes that Falk et al.'s confidence-interval argument cuts both ways since the original estimates rest on 35 children.

The reply is explicitly about whether controlling heritable early ability is over-adjustment. This archive sides with controlling it, while recording that the question is contested.

Not applicable.