I am using two sample sizes for the independent variables: transformational leadership and intrinsic motivation. One dependent variable turnover intention and a moderating variable organizational commitment. The moderating variable is used for the independent variables.
When determining the sample sizes using G*Power I am a bit confused as to which tests to use, I do know it is the priori, two tail, power .90, medium or large effect size, and alpha 0.01.
Any insights, advice, or pointing me in the right direction to find the information would be greatly appreciated. Thanking in advance.
I don't think that you will get any useful advice here other than, possibly, that you should go to a local (bio-)statistician.
You did not formulate any statistical hypothesis you plan to test. You did not even specify by what kind of random variable you model your DV ("turnover intention" does not seem to be something that can be modelled in a sensible way by a normal-distributed random variable). You did not say anything about the characteristics of the IVs and the moderator and their assumed relationships with the DV. It is extremely unlikely, to my experience, that you will be able to give
these details here so that a statistician can understand your problem and effectively help you. Therefore I resommend to talk face-to-face to a statistician so that seh can ask, you can answer and that both of you can approach a meaningful statistical model and define a meaningful statistical hypothesis to test, for which the required sample size can be determined -- what might require a simulation approach not provided by G*Power, so that you would need further assistance of help in running and interpreting such simulations.
Or you forgent about the sample-size calculation and just take the largest possible sample that is feasable within your time and money budget. As sampling does not seem to harm any creature, taking a very long time or being very expensive, you would not risk much in just aquiring a sample as-large-as possible and see if the data are sufficiently conclusive. However, this still requires you to set up a proper statistical model and to map your research question to a testable statistical hypothesis within this model; something for what you will likely need the help of a statistician, still.
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Sampling Design
- Sample Size