15.40.* Logistic regression has infinite maximum likelihood estimates when the cases with y = 1 are separate from the cases with y = 0 in the space of explanatory variable values.When this happens, most software merely reports large estimates with huge standard errors. Check what your software does when

(a) y = 0 at x = 1, 2, 3, but y = 1 at x = 4, 5, 6.

(b) in a 2×2 table, the numbers of (0, 1) outcomes are

(5, 0) for females but (0, 5) for males. (In practice, infinite estimates occur whenever a factor has only y = 0 or only y = 1 for some category.)