ABSTRACT
Logistic regression uses the same kind of modeling techniques as the author have seen for least-squares regression. Models may contain continuous variables, coded-categorical variables, and products of those variables. With the basic tools of logistic regression outlined the author can compare the same types of more complex models that the author investigated using least-squares methods. Statistical power is important for the same reasons it was in least-squares regression. With low power, there will be increased chances of making Type II errors. Logistic regression analyses solve these problems by modeling logits rather than the individual dichotomous values of the dependent variable. Finally, the authors’ conclude with a brief discussion of statistical power, noting that ordinary least squares analyses of a continuous dependent variable is generally more desirable than logistic regression analyses of a dichotomous dependent variable as the former is more statistically powerful.
