topic DOE - Compare Slopes in Discussions

DOE - Compare Slopes

LadderMonster — Sat, 12 Jul 2025 03:36:08 GMT

I have data from a 3x3 DOE but testing 2 standard concentrations. I want to compare the slopes between each condition but am not sure how to do this. I was fiddling with Fit model but am unsure how to add the slope for the response comparison.

Using JMP 18.1.10.

Re: DOE - Compare Slopes

statman — Sat, 12 Jul 2025 15:07:21 GMT

First, welcome to the community. There is not very much information provided, so it is difficult to provide specific advice. It is helpful to attach your data table. Let me see if I understand...You have an experiment that has 3 factors at 3 levels, so 27 treatments, correct? What does "but testing 2 standards concentrations" mean? Are you replicating the 27 treatments for 2 different concentrations? Or are you repeating each treatment with 2 concentrations?

What are the response variables? Since you are indicating there is some sort of slope, I imagine you are taking measurements during the treatment (perhaps a measure of degradation or change in viscosity). You are therefore interested in the rate of change. This you could calculate from the data and use as a response variable (along with mean and standard deviation). If the data approximates a linear relationship, the calculation of slope is easy and you could fit a line to the data and JMP can give you the slope. If there is significant departure from linear, slopes are more difficult to calculate.

Re: DOE - Compare Slopes

LadderMonster — Mon, 14 Jul 2025 16:37:50 GMT

Hi Statman,

Apologies, it's my first post. The data is now attached.

I have an experiment that has 3 components Component A, B, and C. There were two dilutions tested (%Dilution). The responses are Y1 and Y2. I want to compare the effect of the different components on the slopes of the two responses between the two %Dilutions (15 and 60).

I wanted to use the Fit model but am unsure how to easily compare the slopes.

Thanks in advance!

Re: DOE - Compare Slopes

statman — Mon, 14 Jul 2025 17:20:36 GMT

Not sure what kind of experiment you ran? Component A is certainly biased to level 1. What model were you using to design this experiment? B&C make a 3^2, but A is not orthogonal? How were the 10 data points for each Y gotten for each treatment? Are those repeats? Are they in time series?

There are a number of outliers also detected. Perhaps you need to investigate those and then decide what to do with those data points. Once you have considered that, you will use the 10 data points per treatment to calculate a slope (rise/run) for each Y. Analyze the slopes as the Y in fit model.

I'd start by looking at the data. Also make sure the variation ids of practical significance before doing any quantitative analysis. I added some simple scripts to your data table (Multivariate, and graph builder by factor)

Re: DOE - Compare Slopes

Mark_Zwald — Mon, 14 Jul 2025 17:24:10 GMT

If you want to check if the slope of Y vs X1 is statistically different from Y vs X2, then look at the p-value of the X1*X2 regression term. If it's low, the slopes are statistically different.