enormous SE

[mcallen]

Any ideas why this nest survival model has such a large SE on the slope parameter (see below)? The I.V. was a dummy variable (1=mowed, 0=unmowed). Number of nests were: 4 mowed, 16 unmowed. Thinking it may be related to the fact that no mowed nests failed (i.e., DSR = 1).

Mike

grsp - patuxent

LOGIT Link Function Parameters of {NESTMOW}
95% Confidence Interval
Parameter Beta Standard Error Lower Upper
------------------------- -------------- -------------- -------------- --------------
1: 2.9475134 0.4190881 2.1261 3.7689
2: 17.928609 4044.0206 -7908.3520 7944.2092

[gwhite]

Mike:
You identified the problem. Without any unsuccessful nests for the mowed group, the estimator of the mowing effect is infinity. You get the same problem in logistic regression when you have all failures for values of x less than some threshold, and all successes above this threshold.
Gary

[mcallen]

Thanks!