I submitted my first paper to a peer-reviewed journal and got lots of feedback, questions and criticism back.
One point was related to the number of digits I report for estimated values (e.g. for Cronbach's alpha values or for regression coefficients).
The reviewer asks me whether it is justified that I use e.g. four digits for my Cronbach's alpha estimates.
Actually, I just took the output from Stata and thought that if Stata reports these estimates, I can report them as well.
I guess the question implies that is not realistic (and not necessary) to report four digits.
How do I know how many digits are okay?
Thanks a lot!
Best Answer
Different journals may have different style guides about this. Many follow the APA (American Psychological Association) guide. The APA doesn't have a specific rule about Cronbach's alpha, but they recommend reporting correlations to two digits (e.g. 0.45). That seems adequate for Cronbach's alpha as well. R's pscyh package reports alpha to two digits.