I did a retrospective study to compare the efficacy of two different treatments for the patients with infertility problems. I got negative results. How can convince the readers that this is not an underpower study? What is the point of calculate power if the study has been completed?
Hi,
Statistical power lets you gauge the smallest detectable difference within a certain power, cohort size, and the variable standard deviation used in your analysis. Hence, you could frame a retrospective power analysis in terms of not detecting any meaningful (i.e., relevant to the clinical community) difference within a certain power envelope. Still, the failure to reject the null hypothesis is not the same as concluding that your two treatments are not different.
Practically, you can use the Design of Experiment tab to look for the Sample Size and Power menu. In your case, it appears that you are comparing two means with a defined population size. By entering these parameters, JMP will display a curve of Powers by difference to detect (i.e., the number of standard deviations between the means of the two treatments).
This answer is my humble attempt at providing a practical answer. Others may have more refined explanations.
Best,
TS
Hi @fgaochori : "What is the point of calculate power if the study has been completed?". As your rhetorical question implies...the short answer is "there is little point" in doing such calculations. Yes, it can be calculated (just like you can put a golf ball down the toilet)...but should you?. No.
Here is a good article with good references. https://library.virginia.edu/data/articles/post-hoc-power-calculations-are-not-useful
One of the references here: https://www.zoology.ubc.ca/~bio501/R/readings/hoenig%20&%20heisey%202001%20am%20stat%20-%20fallacy%20of%20power%20calculations%20for%20data%20analysis.pdf
