Sample size calculation for multiple paired comparisons

UcheOkoro — Fri, 15 Jan 2021 19:26:58 GMT

Hello,

Please, I need help calculating sample sizes for 3 paired comparisons. I am looking at 3 measurements of blood glucose within the same individual to see if there is difference. I am also concerned about p-value correction such as Bonferroni and holm I do know how to apply it to the codes if required.

The following is my code.

Thank you

ods pdf file="C:\temp\Merged44.pdf" style = minimal;
 ods graphics on;
 title " Sample size by power";
proc power ;
 pairedmeans test=diff 
  meandiff =4.5
  std = 12.5
  corr = .20 0.25 0.30
  npairs = .
  alpha =.05 
  power = 0.8 0.85 0.9;
plot
min=0.8
yopts=(ref=280 crossref=yes)
vary(color by stddev, linestyle by meandiff, symbol by corr);
ods graphics off;
 quit;
 ods pdf close;

Re: Sample size calculation for multiple paired comparisons

Norman21 — Mon, 18 Jan 2021 09:26:44 GMT

If I understand your question, you simply apply the Bonferroni correction to the alpha; in you case, the adjusted alpha would be 0.05/3 = 0.017

More here, including discussion of adjustment for correlation between predictors: https://stats.idre.ucla.edu/sas/dae/multiple-regression-power-analysis/

topic Sample size calculation for multiple paired comparisons in Statistical Procedures

Sample size calculation for multiple paired comparisons

Re: Sample size calculation for multiple paired comparisons