Using the Parental HIV data calculate an overall Brief Symptom Inventory (BSI) score for each adolescent (see the codebook for details). Log-transform the BSI score. Obtain a normal probability plot for the log-transformed variables. Does the log-transformed variable seem to be normally distributed? As you might notice, the numbers of adolescents with a missing value on the overall BSI score and the log-transformed BSI score are different. Why is this the case?

Could this influence our conclusions regarding the normality of the transformed variable? How could this be avoided?