Chapter 32 - Analyzing male fertility data

Summary

This chapter reviews the basics of probability, statistics, and computer modeling, with special attention to problems relating to male infertility. A probability distribution describes how the probabilities of all possible outcomes of an event are divided or distributed. Mathematically, the utility of a test can be described by various parameters, the most commonly used being sensitivity, specificity, positive predictive value, and negative predictive value. The problem confronting the andrologist is the multifactorial nature of male infertility. One reason for the popularity of logistic regression is that inferential statistical tests exist to determine if the weights are statistically significant. This allows researchers to report the predictive variables, and ignore the others in the predictive model. In discriminant function analysis, the observed data are used to create a Gaussian probability distribution centered at the multidimensional mean of each group and with the appropriate standard deviations.