The document discusses recoding variables in an SPSS data file to conduct statistical tests. It describes computing BMI from height and weight variables, then recoding BMI into categories of underweight, normal, and overweight. The recoding is done to test for associations between weight classification and the outcome of small for gestational age (SGA) babies. Steps shown include using the compute variable and recode functions in SPSS to generate new variables for weight classification based on BMI.