topic Re: PROC ANOVA - SNK in Statistical Procedures

PROC ANOVA - SNK

dht115 — Sun, 22 Nov 2020 19:24:50 GMT

Need help for setting...

I am not getting 'SNK Grouping' information by using the following code:

DATA chol_treat;
length treat $ 7;
do treat = 'A', 'B', 'Placebo';
do i = 1 to 10;
input result @;
output;
end;
end;
drop i;
datalines;
220 190 180 185 210 170 178 200 177 189 160 168 178 200 172 155 159 167 185 199 240 220 246 244 198 238 277 255 190 188
RUN;

PROC ANOVA data = chol_treat;
title 'Analysis of Cholesterol Meication data';
class treat;
model result = treat;
means treat / SNK;
RUN;

Re: PROC ANOVA - SNK

PaigeMiller — Sun, 22 Nov 2020 19:56:06 GMT

I run your exact code and I see the SNK output. Can you show us screen captures of your output by including the screen capture in your reply by clicking on the camera icon here? Do not provide screen captures as attachments.

Re: PROC ANOVA - SNK

PGStats — Sun, 22 Nov 2020 20:40:02 GMT

Add option linestable to the means statement.

Re: PROC ANOVA - SNK

SteveDenham — Mon, 23 Nov 2020 14:20:20 GMT

@PGStats and @PaigeMiller have given you everything you need to get SNK output. Now I am going to ask a question -why SNK, when it is known to be one of the poorest methods for controlling type I error amongst all the available multicomparison methods. I recommend familiarizing yourself with the work of Peter Westfall and in particular the book Multiple Comparisons and Multiple Tests Using SAS. This link from the GraphPad documentation is short and explains the reasoning well:

https://www.graphpad.com/support/faq/why-we-recommend-you-do-not-use-the-newman-keuls-multiple-comparison-test/

SteveDenham