Generalized likelihood ratio test for the shoulder condition in line transect sampling

We propose a generalized likelihood ratio procedure for testing the shoulder condition of a detection function in line transect sampling. The proposed procedure is compared to the other existing methods for testing the shoulder condition. It is demonstrated that the new method has a much higher power than the other methods when the true detection does not possess a shoulder, while it has similar performance to the other methods when the true detection satisfies the shoulder condition. We also apply the new method to some well known data sets in line transect sampling.