topic Re: Is there a way to get exact p-values from Tukey? in Statistical Procedures

Is there a way to get exact p-values from Tukey?

abarnard — Fri, 21 Aug 2015 14:23:52 GMT

Hello,

I am a relatively novice SAS user (currently using SAS 9.4, remote) and have what I believe is a very basic question. I have a data set (n=80) of patients with one of three cancers (prostate, lung, breast) and am comparing other serum measurements against their form of cancer. For example: Is the mean serum, albumin level different between patients with lung cancer and breast cancer, etc. For this I am utilizing a one-way ANOVA with post hoc Tukey test to control for type one error. Everything runs smoothly and I get the results I want but I would like to report the exact p-values between the multiple comparisons rather than just knowing they are significant (a<0.05) via asterisks. Is there a way to do this? Adding a lines statement before tukey does not give the p-value either.

Sample code (after data):

proc anova data=TotalINTSAS;

class cancer;

model albumin=cancer;

means cancer/tukey;

run;

Any help is appreciated! Thanks.

-Alle

Re: Is there a way to get exact p-values from Tukey?

SteveDenham — Fri, 21 Aug 2015 14:55:00 GMT

Shift over to PROC GLM so that you can use the LSMEANS statement (probably a good idea anyway as I would guess that your data are unbalanced per cancer type).

proc glm data=TotalINTSAS;

class cancer;

model albumin=cancer;

lsmeans cancer/stderr pdiff=all adjust=tukey;

run;

Steve Denham

Re: Is there a way to get exact p-values from Tukey?

abarnard — Fri, 21 Aug 2015 15:30:26 GMT

Steve-

I appreciate your quick and helpful reply! I ran the code you posted and it works quite well. My data is almost balanced (prostate=25, breast=30, lung=25) but in the future I am sure it will not be so pretty thus this is a great trick to keep in my pocket! Thanks again.

-Alle