Using the plasma retinol data in Section 1.3.3, and removing leverage points, ignore the former smokers and use the R function logrsm to estimate the likelihood that a participant was never a smoker given their plasma beta-carotene value.

Does this suggest that a basic logistic model is reasonable? Test the hypothesis that there is no association using the basic logistic model. The data are stored on the author’s web page in the file plasma dat.txt. Note that the first 15 lines describe the variables and need to be skipped when reading the data into R.