topic Re: Power Analysis for Slope of a Regression Line in Discussions

Power Analysis for Slope of a Regression Line

Skrombee — Tue, 25 Feb 2020 21:31:35 GMT

How do I do a power analysis for the slope of a regression line in JMP? I'm trying to figure out how many samples I need to collect to look at the relationship between two quantitative variables.

Re: Power Analysis for Slope of a Regression Line

Mark_Bailey — Tue, 25 Feb 2020 22:23:53 GMT

I suggest that you start with DOE > Custom Design. Define your response. Define your continuous factor. Define your model terms. Try some number of runs, make the design, and then open Design Evaluation > Power Analysis. You have two choices here. You can specify effect size and random variation on a relative basis. Leave RMSE at 1 and enter coefficients as multiples divided by 2. So if you expected an effect to be 3 standard deviations, enter 1.5 for the anticipated coefficient. Or you can evaluate it on an absolute bases. Enter the actual RMSE (response standard deviation) and the coefficient. This value will be one half of the total change expected in the response over the factor range.

Re: Power Analysis for Slope of a Regression Line

Skrombee — Wed, 26 Feb 2020 16:38:42 GMT

Thank you! Is there an example somewhere I can follow?

Re: Power Analysis for Slope of a Regression Line

Mark_Bailey — Wed, 26 Feb 2020 17:46:48 GMT

See the same book and search for power analysis. It is fully explained and it is demonstrated with examples that you can follow.

Re: Power Analysis for Slope of a Regression Line

lala — Wed, 04 Mar 2020 13:26:07 GMT

DOE 、NICE!