topic Re: Coding Question in SAS Procedures

Coding Question

Wendy13 — Wed, 08 Apr 2015 03:07:46 GMT

I am struggling with coding with following questions:

I need to find difference between male and female from dataset and find significance between male and female.

First, should I subset or split dataset?

There are variables for each male and female so I think I need to create a new variable with add number of variables.

What is a code to get significance between male and female?

Please help me,

Re: Coding Question

RW9 — Wed, 08 Apr 2015 13:13:34 GMT

Hi,

You shouldn't need to split your dataset, SAS does by group processing so just use gender as a by or class variable. E.g.

proc sort data=sashelp.classfit out=tmp;

by sex;

run;

proc means data=tmp noprint;

var age;

by sex;

output out=myres n=n mean=mean stddev=stddev min=min max=max;

run;

This will give you means results by sex from the sashelp table.

As for significance, not sure, your best off posting that question in Statistical Procedures sub forum.

Re: Coding Question

ChrisHemedinger — Wed, 08 Apr 2015 13:20:05 GMT

is correct -- no need to split the data. However, one improvement on this answer: you don't need to use BY groups or SORT the data first with PROC MEANS. You can use the CLASS statement instead:

proc means data=sashelp.classfit noprint; 
     var age;
     class sex;
     output out=myres n=n mean=mean stddev=stddev min=min max=max; 
run;

Result:

The first record is the stats for ALL observations, and the next two records are for F and M, respectively.

Chris

Re: Coding Question

art297 — Wed, 08 Apr 2015 14:16:35 GMT

If you only need to analyze one variable, you could get everything you want by running proc ttest. e.g.:

proc ttest data=sashelp.class;

class sex;

var weight;

run;

Re: Coding Question

PGStats — Wed, 08 Apr 2015 15:19:04 GMT

... IF (big if) your data has a normal distribution. Otherwise, or instead, since there is not is not much power lost, go for non-parametric analyses :

proc npar1way data=sashelp.class Wilcoxon;

class sex;

var weight;

exact Wilcoxon;

run;

Re: Coding Question

Ksharp — Thu, 09 Apr 2015 14:02:09 GMT

IF you have more than two levels, Don't forget to use variance analysis .

proc glm

proc anova

Xia