A Simple Example of a Comparison Involving Quantal Data

Data in which the response variable has only two possible forms, success or failure, defective or non-defective, etc., arise in many fields. The present note is concerned with a particular comparison problem occurring with this sort of data. The problem could be formulated fairly generally, but the discussion here will be very largely in terms of the following specific example. Table 1 gives data of Gordon & Foss (1966) on the effect of rocking on the crying of very young babies. On each of 18 days, the babies not crying at a certain instant in a hospital nursery served as subjects. One baby selected at random was rocked for a set period, the remainder serving as controls. The numbers not crying at the end of a specified period are