Statistical procedure

[mounikag]

Hi Team,

I have data as below 

subject  dose1  dose2  dose3  dose4

101       0.32      0.93   0.54      0.61

102       0.23      0.99   0.54      0.71

103       0.32      0.23   0.84      0.71  

104       0.33      0.93   0.64      0.61

105       0.22      0.73   0.44      0.66

I need to calculate GMR(geometric mean ratio) acrooss doses like dose1-dose2, dose2-dose3 and dose3- dose4. I wanted to find the geometric
mean across each variable for the whole subjects. so i want the output has below

Doses               GMR value(geometric mean ratio)

Dose1

Dose2            xx

Dose3             xx

Dose4             xx

so what procedure i need to use to calculate GMR across doses . is it possible to calculate GMR in proc TTest procedure if yes, wat is the option available to use in syntax?

[Reeza · Accepted solution]

Can you please illustrate the full calculation? I'm familiar with geometric mean but that doesn't look like what you're looking for here so want to be absolutely sure.

Making assumptions: 

data diff;
set have;

dose2_dose1 = dose2-dose1;
dose3_dose2 = dose3-dose2;
dose4_dose3 = dose4-dose3;
run;

proc ttest data=diff dist=lognormal;
var dose2_dose1 dose3_dose2 dose4_dose3;
ods select confLimits = WANT;
run;

proc print data=want;run;

https://blogs.sas.com/content/iml/2019/10/02/geometric-mean-deviation-cv-sas.html

---

@mounikag wrote:

Hi Team,

I have data as below 

subject  dose1  dose2  dose3  dose4

101       0.32      0.93   0.54      0.61

102       0.23      0.99   0.54      0.71

103       0.32      0.23   0.84      0.71  

104       0.33      0.93   0.64      0.61

105       0.22      0.73   0.44      0.66

I need to calculate GMR(geometric mean ratio) acrooss doses like dose1-dose2, dose2-dose3 and dose3- dose4. I wanted to find the geometric
mean across each variable for the whole subjects. so i want the output has below

Doses               GMR value(geometric mean ratio)

Dose1

Dose2            xx

Dose3             xx

Dose4             xx

so what procedure i need to use to calculate GMR across doses . is it possible to calculate GMR in proc TTest procedure if yes, wat is the option available to use in syntax?

---

[mounikag]

Hi reeza ,

thanks for your reply i want to calculate GMR across doses(dose1-dose2, dose2-dose3, dose3-dose45) on whole population. what stsitsical procedure i can use . when I seached online it showing me below syntax:

PROC SURVEYMEANS data=my_data geomean;
var dose1 dose2;
run;

is there any other procedure to get values?