Contents
- Index
Adjust Number of Parameters
Often, the numerical procedure used in Program MARK is unable to correctly determine the number of parameters (K) in the model. The problem happens most often with the logit link when parameters are at the boundary, e.g., a survival or capture probability is estimated as 1. However, other examples of the problem occur with the closed captures models when the population size, N, is estimated close to the number of animals encountered. This problem causes the AIC to be computed incorrectly:
AIC = -2log Likelihood + 2K
and thus also causes problems with the QAIC (quasi-AIC)
QAIC = -2log Likelihood/c-hat + 2K
and the QAICc:
QAICc = -2log Likelihood/c-hat + 2K + 2K(K + 1)/(n-ess - K - 1)
where n-ess is the effective sample size.
When you recognize that the number of parameters returned by the numerical estimation routine is incorrect, you can change the value with this dialog box. Enter the correct number of parameters, and the QAICc will be corrected appropriately.
For random effects models obtained with the Variance Components analysis, the number of parameters may be a non-integer value.