Independent Sample T-Test
Does anyone know how to run this test for a known mean and standard deviation, but without the actual values?
For example, I am abstracting data from published clinical trials which report the mean and standard deviation, but would like to determine the other parameters not reported (e.g. variance). The data used to calculate the mean and standard deviation is not available.
Thanks!
Comments
I am not sure if I understand correctly but regarding your first question you can calculate t statistic value from mean and SD (if you also have a sample size) and use this value to obtain any Bayes Factor you are interested in (in example via ttest.tstat from BayesFactor R package or on http://pcl.missouri.edu/bf-two-sample). Regarding second part, according to my (amateur) knowledge variance is not fully encoded in mean, SD and df so it is not possible, you can only try to simulate those values (please correct me if I am wrong).
Sample size and t-statistic are sufficient, meaning that no additional information is needed and no information is lost through this summary (as far as inference is concerned). With sample size and t-statistic in hand, you can indeed conduct the Bayes factor analyses in R, and in JASP you can do this through the "Summary Stats" module.
Got it, Thanks!