PH490KR_Session17

/**********************************
THIS PROGRAM PERFORMS ACTIVITIES FOR
SESSION 17, PH490KR SPRING 2018

EMILY LYNCH
Modified: March 29, 2018
*******************************/

*Start by assigning libname to folder where dataset; 
libname ph490kr "C:\Users\emilylynch\Desktop\ph490kr";

data sess17; /*read in data into a temporary dataset that is stored in work and called sess17*/
set ph490kr.vitdsess17; /*this is the dataset being read in*/
run;
*read in dataset to a temporary dataset;
data sess17;
set ph490kr.VitDsess17;
run;

*1a. check age_cig;
proc means data=sess17;
var age_cig;
run;

proc freq data=sess17;
table age_cig*ev_smk;
run;

*fix by setting 99 to missing;
data sess17;
set sess17;/*note: this code will overwrite our temporary dataset*/
age_cigkr=age_cig;
if age_cig=99 then age_cigkr=.;
if age_cig=0 then age_cigkr=.;
run;

*check our work;
proc means data=sess17;
var age_cigkr;
run;

proc print data=sess17 (obs=50);
var age_cig age_cigkr;
run;

*1.b. check age_qt;
proc means data=sess17;
var age_qt;
run;

proc freq data=sess17;
table age_qt*cur_smk;
run;

*set values of 99 to missing;
data sess17;
set sess17;

age_qtkr=age_qt;
if age_qt=99 then age_qtkr=.;
run;

*check work;
proc means data=sess17;
var age_qtkr;
run;

proc print data=sess17 (obs=50);
var age_qt age_qtkr;
run;

*1c.check cig_day;
proc freq data=sess17;
table cig_day*cur_smk/missing;
run;

*set 99 to "0" for those who answer "no" to current smoking;
data sess17;
set sess17;

cig_daykr=cig_day;
if cig_day = 99 and (cur_smk eq 0) then cig_daykr=0;
if cig_day = 99 and ((cur_smk eq .) or (cur_smk eq 1)) then cig_daykr=.;
run;

*check work;
proc freq data=sess17;
table cig_day*cig_daykr cig_daykr*cur_smk cig_daykr/missing;
run;

proc freq data=ph490kr.vitdsmk;
table cig_day;
run;

*1d. check yrs_oc;

proc means data=sess17;
var yrs_oc;
run;
proc freq data=sess17;
table yrs_oc*ever_oc/missing;
run;

*need to set 99 to "0" for those who answer "no" to ever OC use;
/*also, many said no to ever_oc but then provided a number of years of OC use--how to handle?
1. could set to "0," essentially applying skip pattern (if they had skipped the question as
they were supposed to, they would have been set to 0)
2. set to missing, since answers are inconsistent
3. keep answers and set ever_oc to yes since they are indicating having used OCs

Any of these options are ok, as long as you are consistent, document, and have a good rationale.
Let's go with option #1*/


data sess17;
set sess17;

yrs_ockr=yrs_oc;

if yrs_oc=99 & ever_oc=0 then yrs_ockr=0;
if yrs_oc=99 & ever_oc ^=0 then yrs_ockr=.;
run;

*check our work;

proc print data=sess17;
var yrs_ockr;
where ever_oc=0;
run;

proc print data=sess17;
var yrs_ockr ever_oc;
where ever_oc ^=0;
run;

*1.e. check curr_oc;
proc freq data=sess17;
table curr_oc*ever_oc/missing;
run;

/*have 14 who answered yes to current but no to ever; 
we will set to no to be consistent with our decision of how to handle yrs_oc above
also have values of 99 that need to be set to no
and some values of 99 when ever_oc=1 or missing that need to be set to missing*/

data sess17;
set sess17;

curr_ockr=curr_oc;
if curr_oc=1 & ever_oc=0 then curr_ockr=0;
if curr_oc=99 & ever_oc=0 then curr_ockr=0;
if curr_oc=99 & ever_oc^=0 then curr_ockr=.;
run;

*check our work;
proc freq data=sess17;
table curr_ockr*ever_oc/missing;
run;

*2. label each variable and its values;
proc format library=ph490kr;
value ynf 0 ="0_no" 1= "1_yes";
value cig_dayf 1= "1_1-10" 2= "2_11-20" 3= "3_21-30" 4= "4_31-40" 5= "5_41+";
run;

options fmtsearch=(ph490kr);

data sess17;
set sess17;

format ev_smk cur_smk ever_oc curr_ockr ynf.;

label ev_smk="ev_smk: ever smoked 20+ packs cig";
label cur_smk="cur_smk:currently smoke cig";
label cig_daykr="cig_daykr:cigarettes per day";
label age_cigkr="age_cigkr:age started smoking";
label age_qtkr="age_qtkr:age quit smoking";
label ever_oc="ever_oc:ever used OCs";
label yrs_ockr="yrs_ockr:total years of OCuse";
label curr_ockr="curr_ockr:current OC use";
run;

proc contents data=sess17;
run;

*let's clean up our dataset by dropping the variables we recoded;

drop cig_day age_cig age_qt yrs_oc curr_oc;

run;

*3. save a permanent copyl;
data ph490kr.VitDsess17_working;
set sess17;
run;
proc contents data=ph490kr.VitDsess17_working;
run;