Induction and latent periods are distinguishable concepts referring respectively to the period between causal action and disease initiation, and the period between disease initiation and detection. A disease cannot be characterized as having a long or short induction period, except in relation to a specific etiologic component. Inappropriate assumptions, explicit or implicit, about the length of the combined induction and latent period (the "empirical induction period") in an analytic study result in nondifferential misclassification and bias toward the null. Repeated analyses, varying the assumptions about the length of the empirical induction period, can be used to minimize such misclassification, thereby providing estimates for an undiluted measure of effect and the mode of the empirical induction period.
Kenneth J. Rothman (Sat,) studied this question.