Can someone look into whether SAS option 'dist=lognormal' gives the same results, as simply taking the natural log transform the data, calculating the t-test p-value & the mean/ CIs on the log scale and then exponentiating them back in order to get the t-test for lognormal data. (add that SD can't be back transformed).. If these give same results as SAS then update Comparison to say although option not available in R, can do it using log / anti-log. Could also look into the EnvStats Package to see if their function is possible & reliable alternative
Please whoever does the update, check paired t-test & two-sample t-test as if they reference EnvStats & lognormal, the same update would be needed there.
[ ] check how to analysis data on the log scale
[ ] update one sample t-test
[ ] update paired t-test
[ ] update two sample t-test (if needed)
[ ] update R pages
[ ] update SAS pages
[ ] Update comparison (it can be done easily using log / anti-log & gets same results.. .(If that is true as proven above!)
https://github.com/PSIAIMS/CAMIS/blob/5e19d3d0ac003e46dca043f203b3c844bb8ecca4/R/ttest_1Sample.qmd#L73
Can someone look into whether SAS option 'dist=lognormal' gives the same results, as simply taking the natural log transform the data, calculating the t-test p-value & the mean/ CIs on the log scale and then exponentiating them back in order to get the t-test for lognormal data. (add that SD can't be back transformed).. If these give same results as SAS then update Comparison to say although option not available in R, can do it using log / anti-log. Could also look into the EnvStats Package to see if their function is possible & reliable alternative
Please whoever does the update, check paired t-test & two-sample t-test as if they reference EnvStats & lognormal, the same update would be needed there.