topic Re: multiple regression model explanation in Statistical Procedures

multiple regression model explanation

samer_badri75 — Thu, 08 May 2025 14:49:24 GMT

Greeting All

I have a Q related to explain the multiple regression model explanation which is for example:

X= 0.4563 + 0.0134 °C + 0.43 h

if I need to explain by cofficient but the problem they are diffiernt.So, it is better to explain that using ANOVA result?

Thanks

Re: multiple regression model explanation

PaigeMiller — Thu, 08 May 2025 14:52:10 GMT

Please explain your question in more detail.

Why do you think the coefficients should be the same? Why is it a problem that they are different? Normally, coefficient on variable A will be different than the coefficient on variable B.

Re: multiple regression model explanation

samer_badri75 — Thu, 08 May 2025 14:56:23 GMT

Many thanks for your quick respond

Please note that the ranges of temperature (70-90°C) and time (12-18 h) differ numerically; therefore, the effect of a unit change in each coefficient on the outcome will vary. A direct comparison of coefficient magnitudes is misleading without considering the scale of their respective variables.

That what I get from editorial bord related to one munscript.

Re: multiple regression model explanation

PaigeMiller — Thu, 08 May 2025 14:58:52 GMT

I agree with what you wrote. So what is your question?

Re: multiple regression model explanation

samer_badri75 — Thu, 08 May 2025 15:00:53 GMT

I am asking what is the better way to explain the modle ?

Is it depend on p-valu or the cofficient

Re: multiple regression model explanation

PaigeMiller — Thu, 08 May 2025 15:03:23 GMT

You could standardize the input x-variables, to have mean 0 and variance 1, and then run the regression again, and now you can compare the coefficients directly.

Does that help? Is that what you want?

Re: multiple regression model explanation

samer_badri75 — Thu, 08 May 2025 15:21:33 GMT

Looks good but how to do that

for example

x Temp hour

0.064 75 12

. . .

0.0368 90 15

I have 27 number for each colume

Re: multiple regression model explanation

PaigeMiller — Thu, 08 May 2025 15:26:01 GMT

You seem to be using X as the response variable, while I am using x to represent the independent variables.

In SAS, use PROC STDIZE to perform standardization on the independent variables.

https://documentation.sas.com/doc/en/pgmsascdc/9.4_3.4/statug/statug_stdize_toc.htm

Then run the regresssion.

Re: multiple regression model explanation

Ksharp — Fri, 09 May 2025 01:37:30 GMT

I do not understand since temperature (70-90°C) and time (12-18 h) are totally different variables ,why do you want to comprare their coefficient ? Maybe want to find out the importance of variables in model ?

As a matter of fact, you can compare different variables estimated coefficient by STB option of PROC REG/PROC GLMSELECT/PROC GLIMMIX.

Check @Rick_SAS blogs:

https://blogs.sas.com/content/iml/2018/08/22/standardized-regression-coefficients.html

https://blogs.sas.com/content/iml/2021/05/17/standardized-coefficients-glimmix.html

https://blogs.sas.com/content/iml/2023/07/17/standardize-reg-coeff-class.html