Note that there are a number of links available for studying binary HPV outcomes such as probit, complementary log-log link or identity links. Alternative links provide different measures of exposure–disease associations: for example, an identity link provides an estimate of the difference in prevalence (i.e., attributable risk). These other links can sometimes give different results than logit or log links. The decision regarding which link functions are correct depends on the underlying biologic relationship and usually can not be determined solely on a statistical basis. However, the logit link (standard logistic regression) is the most common link function used in HPV research.
