Usually in SP surveys a number of choice obsrvations is taken from each individual to reduce the cost of data collection. Thsi raises the so-called "repeated measurements" problem: observations taken from the same individual are not independent, which means that simple analysis method may be biased. In this article we describe the application of hte Jackknife and the Bootstrap techniques to attack this problem in the case of a simple logit model, and compare their relative performance. The Jackknife and Bootstrap are used to investigate bias due to serial correlation, the influence of hte size of hte Jackknife sample, the influence of the error distribution for each respondent individually and across respondents, and the variance estiamte of the model coefficients obtaiend in the course of macimum-likelihood estiamtion.
展开▼
