0
$\begingroup$

I have the following code:

data nozzle;
input nozzle vel shape @@;
datalines;
 1 14.37 0.80 1 16.59 0.81 1 20.43 0.75 1 23.46 0.77 1 28.74 0.78
2 11.73 0.85 2 14.37 0.85 2 16.59 0.92 2 20.43 0.86 2 23.46 0.81 2 28.74 0.83
3 11.73 0.93 3 14.37 0.92  3 20.43 0.89 3 23.46 0.89 3 28.74 0.83
4 11.73 1.14 4 14.37 0.97 4 16.59 0.98 4 20.43 0.88 4 23.46 0.86 4 28.74 0.83
5 11.73 0.97  5 16.59 0.78 5 20.43 0.76 5 23.46 0.76 
;
run;

proc mixed NOINFO method=TYPE1;
class nozzle vel;
model shape=nozzle vel / outp=ps residual;
lsmeans nozzle / alpha=0.05 adjust=tukey cl;
run;
ods graphics off;

proc sgscatter data=ps;
   plot StudentResid*(nozzle vel pred);
run;
ods graphics pff;
ods pdf close;

Run the proc mixed statement to find the mean comparisons.

I have the following question: Now look again at the treatments you identified as not significantly different from the treatment with the largest sample mean. Based on the adjusted p-values, evaluate the strength of evidence regarding the differences in treatment means. Make a conclusion about what treatment(s) you would prefer, using a statistical argument.

I am unsure how to answer this question because I found that according to the adjusted p-values, all the treatments are not different from treatment 4 (the largest sample mean). How do I evaluate the strength of evidence?

Any imput is helpful and thank you

PS: if there is an easier way to share my data from SAS please let me know as this is one of few posts I've made on CrossValidated. Also let me know if there is a way to share the output from SAS. I was not sure if there was.

$\endgroup$
  • $\begingroup$ Looks like a homework problem. Should tag it accordingly. $\endgroup$ – Russ Lenth Mar 2 '18 at 4:04
  • $\begingroup$ At the outset, are you interested only in comparisons to treatment 4? It doesn't sound like it, so you ought to consider procedures that compare all pairs of treatments. $\endgroup$ – whuber Mar 2 '18 at 15:14
  • $\begingroup$ I am sorry; I didn’t post the first part which asked to find the treatments that are not significantly different from the treatment with the largest sample mean (which happens to be 4), so yes, i am only interested in comparing to 4 $\endgroup$ – the boy 88 Mar 2 '18 at 15:27

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.