This is the Framingham data. 

W_2 = average of the two SBP's at exam 2
W_1 = average of the two SBP's at exam 3

Z   = age, smoking status, cholesterol at exam 3

Y   = CHD status






VARIABLE NAMES and DEFINITIONS:    

COL 01 -- OBS = observation number (1--1615)
COL 02 -- AGE = age at exam 2
COL 03 -- SBP21 = first systolic blood pressure at exam 2
COL 04 -- SBP22 = second systolic blood pressure at exam 2
COL 05 -- SBP31 = first systolic blood pressure at exam 3
COL 06 -- SBP32 = second systolic blood pressure at exam 3
COL 07 -- SMOKE = present smoking at exam 1
COL 08 -- CHOLEST2 = serum cholesterol at exam 2
COL 09 -- CHOLEST3 = serum cholesterol at exam 3
COL 10 -- FIRSTCHD = indicator of first evidence of CHD occurring at exam 3
                     through 6, i.e., within an eight-year follow-up period to 
                     exam 2 


GENERAL INFORMATION:

1. The data are for MALES only.
2. The data contain complete records only.
3. I have used a very similar data set in thr SIMEX paper with John Cook. The 
   difference being that we did not use CHOLEST3 in the subsetting process and 
   therefore ontained a slightly larger data set (1660).
4. I pulled the data off the tape that I obtained from the NHLBI and thus we 
   must print a disclaimer whenever the data are used in a published document.  
5. The variates in the data set, with the exception of CHOLEST3, were suggested 
   initially by John Cook as being potentially ``interesting'' for examples. In
   our SIMEX paper we did not use CHOLEST? in order to avoid conflict with 
   any results that Merck people derive (Merck sells a cholesterol lowering
   drug). Since I obtained an independent copy of the tape we are under no 
   legal restriction regarding the use of CHOLEST? variates. However, as a 
   matter of professional courtesy, I would like to give John the option of 
   seeing in advance anything that we intend to publish that uses the CHOLEST? 
   data.