Based on the samples, I decided to research if there’s significant difference among different ethnics on drug use. Here’s the code:
ods html close;
ods html;
libname worklib 'e:\Course\';
proc import
datafile="E:\SPM\Coursera_datasets\nesarc_pds.csv"
out=worklib.nesarc_pds;
delimiter=',';
run;
data worklib.working;
set worklib.nesarc_pds;
where ETHRACE2A=1 or ETHRACE2A=2;
if DGSTATUS=2 then DGSTATUS=1;
run;
proc sort; by idnum;
proc freq; tables DGSTATUS*ETHRACE2A / chisq;
run;
data worklib.working;
set worklib.nesarc_pds;
where ETHRACE2A=1 or ETHRACE2A=3;
if DGSTATUS=2 then DGSTATUS=1;
run;
proc sort; by idnum;
proc freq; tables DGSTATUS*ETHRACE2A / chisq;
run;
data worklib.working;
set worklib.nesarc_pds;
where ETHRACE2A=1 or ETHRACE2A=4;
if DGSTATUS=2 then DGSTATUS=1;
run;
proc sort; by idnum;
proc freq; tables DGSTATUS*ETHRACE2A / chisq;
run;
data worklib.working;
set worklib.nesarc_pds;
where ETHRACE2A=1 or ETHRACE2A=5;
if DGSTATUS=2 then DGSTATUS=1;
run;
proc sort; by idnum;
proc freq; tables DGSTATUS*ETHRACE2A / chisq;
run;
data worklib.working;
set worklib.nesarc_pds;
where ETHRACE2A=2 or ETHRACE2A=3;
if DGSTATUS=2 then DGSTATUS=1;
run;
proc sort; by idnum;
proc freq; tables DGSTATUS*ETHRACE2A / chisq;
run;
data worklib.working;
set worklib.nesarc_pds;
where ETHRACE2A=2 or ETHRACE2A=4;
if DGSTATUS=2 then DGSTATUS=1;
run;
proc sort; by idnum;
proc freq; tables DGSTATUS*ETHRACE2A / chisq;
run;
data worklib.working;
set worklib.nesarc_pds;
where ETHRACE2A=2 or ETHRACE2A=5;
if DGSTATUS=2 then DGSTATUS=1;
run;
proc sort; by idnum;
proc freq; tables DGSTATUS*ETHRACE2A / chisq;
run;
data worklib.working;
set worklib.nesarc_pds;
where ETHRACE2A=3 or ETHRACE2A=4;
if DGSTATUS=2 then DGSTATUS=1;
run;
proc sort; by idnum;
proc freq; tables DGSTATUS*ETHRACE2A / chisq;
run;
data worklib.working;
set worklib.nesarc_pds;
where ETHRACE2A=3 or ETHRACE2A=5;
if DGSTATUS=2 then DGSTATUS=1;
run;
proc sort; by idnum;
proc freq; tables DGSTATUS*ETHRACE2A / chisq;
run;
data worklib.working;
set worklib.nesarc_pds;
where ETHRACE2A=4 or ETHRACE2A=5;
if DGSTATUS=2 then DGSTATUS=1;
run;
proc sort; by idnum;
proc freq; tables DGSTATUS*ETHRACE2A / chisq;
run;
And the output is here
And the letter table:
| Group | ing | ETHRACE2A |
|---|---|---|
| A | 1 | |
| B | 2 | |
| C | 3 | |
| D | 4 | |
| B | D | 5 |
According to the result, the drug use between Asian/Native Hawaiian/Pacific Islander, Not Hispanic or Latino and Hispanic or Latino is not significantly different, but among other every two ethnics, it’s different. Be sure that statistically significant is only a part - we have to associate the result with real life too.