Hello,

I am a wildlife researcher who helps conduct alligator surveys in Florida. I gathered data on alligator behavior to determine how often and when they are underwater or above water (a binary response) and correlated that to a variety of environmental data (time of day, air and water temperature, solar radiation, wind, season, etc.) by running a regression. I developed an equation where one could plug in all environmental variables and predict the probability that an alligator is at the surface and therefore available to be counted during a survey. Analysis was done in SAS, proc genmod.

However, all I end up with is a probability, one number. First, I need help calculating confidence intervals for my final equation. I am also thinking of ways to verify the accuracy of my equation. I am thinking of re-analyzing my data while leaving a randomly picked subset of data out. I would then go back and test my equation against this subset to predict its accuracy. Only by developing an estimate of accuracy, as well as calculating confidence intervals, will my equation be of real use in the field. Is there anybody who knows how to calculate CI's for logistic equations? How do I go about testing my equation against a subset of data to predict accuracy? Any help would be greatly appreciated!