topic Re: Comparison of slopes of lines in Discussions

Comparison of slopes of lines

tox — Thu, 16 Apr 2020 18:19:12 GMT

Good afternoon,

I have a dataset that examines the concentration of a compound over time as a function of one of four different treatments. I want to compare the slopes (i.e. the inactivation rate) between the different treatments to see if there are significant differences. From other post in the discussion, I believe I have been able to enter the data correctly and determine that there are significant differences in the slopes, or inactivation rates, by examining the interaction term (i.e. treatment x time) in the effect tests output. Both treatment and time are also significant by themselves. Where I am unsure is how to apply the multiple comparisons tests to examine the pairwise comparisons to determine which treatments differ from each other. Any guidance would be greatly appreciated.

Thanks in advance,

Paul

Re: Comparison of slopes of lines

Mark_Bailey — Fri, 17 Apr 2020 10:22:46 GMT

If you chose the Effect Leverage emphasis in the Fit Model launch dialog, then click the red triangle next to the leverage plot for the treatment effect and select one of the multiple comparison procedures, such as Turkey HSD.

If you chose another emphasis, then open the Effect Details outline, and use the red triangle menu from there.

Re: Comparison of slopes of lines

tox — Fri, 17 Apr 2020 21:00:42 GMT

Thanks for the reply - I was able to find it. One additional question - I only want to know whether the slopes of the lines are different, and am not interested in the intercepts. Are the comparisons provided by the Tukey HSD, for example, telling me that the slopes are different , or just that the lines (slopes and/or intercepts) are different? Thanks in advance.

Re: Comparison of slopes of lines

Mark_Bailey — Sat, 18 Apr 2020 14:08:21 GMT

I see. I read your original post about using multiple comparisons and thought of the main categorical effect, which is only about the intercept.

See this discussion for an explanation of how to test slopes.

Re: Comparison of slopes of lines

HankStamper — Wed, 19 Aug 2020 15:34:27 GMT

With 4 treatments and multiple time points, you might get a whole bunch of invalid comparisons if you use Tukey's i.e. comparing treatments between different time points. To overcome this, you can use the contrast post hoc and choose the comparisons you want to make.