topic How to obtain the two DF in Proc Ttest e.g. F(df1, df2) = XXX in SAS Procedures

How to obtain the two DF in Proc Ttest e.g. F(df1, df2) = XXX

Mutua — Fri, 20 Nov 2015 14:02:38 GMT

Dear Community,

I recently recently performed a Ttest on my data following the Proc below:

Proc Ttest;

Class Trt;

Var PFG Male Female;

Run;

I am able to get only one F value, yet the reviewer of my manuscript would like me to report me the F-values with their two DF each. Am not sure how to go about this procedure.

I will really appreciate your help if you can guide me on the procedure to follow.

Best regards,

Peter

Re: How to obtain the two DF in Proc Ttest e.g. F(df1, df2) = XXX

Rick_SAS — Fri, 20 Nov 2015 14:34:29 GMT

Is the reviewer asking whether the variances of each group are equal to each other? The Equaliity of Variances table provides information about whether that data indicate whether the Satterthwaite (unequal variance) test might be prefereable to the pooled (equal variance) test. The table includes entries that give the "numerator DF" and the "denominator DF" for the "folded F test." The TTEST doc includes a descrition of the F test, and references:

http://support.sas.com/documentation/cdl/en/statug/68162/HTML/default/viewer.htm#statug_ttest_details14.htm

Re: How to obtain the two DF in Proc Ttest e.g. F(df1, df2) = XXX

slchen — Fri, 20 Nov 2015 14:35:18 GMT

Two DF are the degrees of freedom of the numerator (Num DF) and the degrees of freedom of the denominator (Den DF) in F test to check equality of variances of two samples.

Re: How to obtain the two DF in Proc Ttest e.g. F(df1, df2) = XXX

Mutua — Fri, 20 Nov 2015 14:57:58 GMT

Dear Rick,

Many thanks for the feedback.

The reviewer of the manuscript states that i should report all F-values with their two df each. No other information is given. My assumption is they want to check the credibility of the statisitics provided.

Being a basic user of SAS, i am not sure how to program my SAS Procedure to obtain this information.

I apprecite your feedback. Am able to get the information you stated i.e. Satterthwaite (unequal varince) and equal varince test results. Also the folded F test results. I have looked at the link you gave

Is there a standard Procedure in SAS to do this automatically in SAS?

Best regards,

Peter.

Re: How to obtain the two DF in Proc Ttest e.g. F(df1, df2) = XXX

Mutua — Fri, 20 Nov 2015 15:09:29 GMT

Dear Slchen,

Thanks too for the feedback. What then does it mean when i have similar DF for both Numerator and denominator?

Here are the details of my results

pooled DF=14

Satterthwaite (unequal) DF = 11.96

Folded F: Num DF = 7, Den DF=7, F value = 2.41.

So in this case, is my F(DF1, DF2) = F(7,7)= 2.41? If yes, what does this mean?

Best regards,

Peter.

Re: How to obtain the two DF in Proc Ttest e.g. F(df1, df2) = XXX

Rick_SAS — Fri, 20 Nov 2015 15:15:10 GMT

The Equality of Variance is printed automatically by PROC TTEST. For example, run the following statements:

proc ttest data=sashelp.class;
class sex;
var height weight;
run;

Look in the output for the Equality of Variance tables. There is one for the HEIGHT variable and another one for the WEIGHT variable.

Equality of Variances
Method	Num DF	Den DF	F Value	Pr > F
Folded F	8	9	1.03	0.9527

The first one says that DF1=8 and DF2=9. The p-value is large, which indicates that the data does not reject the null hypothesis that the variance of height is equal for men and women. DF1=8 because there are N1=9 females; DF2=9 because there are N2=10

males.

The second table is similar, except DF1=9, which means that the males are being consider as the numerator when testing the ratio of variances.

Re: How to obtain the two DF in Proc Ttest e.g. F(df1, df2) = XXX

Mutua — Fri, 20 Nov 2015 15:27:52 GMT

Many thanks for that informative email. I really appreciate.

I am getting the same kind on information when i run my analysis.

Going by your example, it means that F(DF1, DF2) = F(8, 9) =1.03 right?

What does it mean when the DF1 and DF2 are the same? and when the P-value is small enough to enable rejection of the null hypothesis?

Best regards,

Peter