assosiations and model fit

#1
Hi I am working on the assigment for biostatistics where I need to investigate the association between intake of and serum concentration of beta carotene.
Also the if the association disturbed by smoking? I am not sure not sure which analysis to run , is this a simple linear regression , also if i should look for the interaction of smoking in the association of intake with serum concentration of beta carotene how do I do it correctly. Is is then multiple regression? Furthermore, if I am interested in if intake of vitamin supplements mediates the association of intake on serum concentration of beta carotene. Please can anyone tell me what to be the rigt thing to do if the task is to conduct and present an appropriate final analysis where all model assumptions are fulfilled, including relevant confounders considering your interaction and mediation analysis using the model with best model fit. I really need help with this , beacuse I am not sure how to approach this evn though I read the material. Hope someone can help?
The following variables are given to this:
  • hahanr (id)
  • sex (1= male, 2=female) categ
  • age (age in years) nr
  • kg (weight in kg) nr
  • kl (height in cm) nr
  • bvbeta – serum concentration of beta carotene (µg/dl) nr dependent variable
  • Command= xi:regress bhdlc i.bmikat age
  • bvbetacat – quartiles serum concentration of beta carotene nr (cutpoint: 43.8 µg/dl; 28.2 µg/dl; and 18.2 µg/dl) nr
  • nsvbeta - nutrition intake of beta carotene (µg per day measured in a seven day diary) nr
  • nsvbetacat: quartils of nutrition intake of beta carotene (cutpoints; 2,67 µg/day; 1,55 µg/day and 0,932 µg/day) nr
  • bchol - overall blood cholesterol (mg/dl) nr
  • btg - overall blood triglyceride (mg/dl) nr
  • suppl - intake of vitamin supplements (0= no; 1=yes) categ
  • rauch - smoking habits (1= current smoker; 2= former smoker; 3= never smoker) categ