1. ## Multiple Groups Regression

Hello to all who view my thread. And thank you in advance for any assistance.

I am trying to compare changes in the associations between multiple measures with increasing age. So, I have run multiple regression on my whole sample (young, middle-aged, and old adults) and now I want to determine if the associations are different between groups.

Can anybody out there help? What test(s) should I be doing? Do I simply re-run the multiple regression for each sub-group? If so, how do I determine if the regressions are significantly different between groups?

Thanks again for any help!

2. put the different groups as categorical or indicator variables. Do you use spss?

3. ## Yes

I do use SPSS and do not see where the categorical variable would go...

4. ## What I am doing

Currently I am doing multiple regression models (stepwise) to see the associations between my dependent and independent variables. Now I want to see if the associations are statistically different among the three sub-groups (young, middle-aged, and old).

So I have multiple independent variables (6) attempting to predict the dependent variables (2). So I have multiple models.

5. the age is the dependet variable and you have six more continuous variables, right? do this: Analyze>General Linear Model. enter the dependent and the categorical variable(s) at fixed factor and the continuous variables at covariates. You can also add ineraction terms from the model button.

6. There's an SPSS tutorial online which explains how to do a problem very similar to yours:
http://www.ats.ucla.edu/stat/spss/faq/compreg2.htm
http://www.ats.ucla.edu/stat/spss/faq/compreg3.htm

7. ## Re: Multiple Groups Regression

Hi,

I need help with a multivariate linear analysis. I am terrible at statistics and could use any help I can get. I am trying to determine the difference in expression of a certain chemical in the lungs of people (continuous variable with values ranging form 0-2000), but I want that compared by age group (scored 1= young, 2= middle aged, 3= old) but to rule out a burn score (0-100%) and injury grade (scored 1-5). If anyone can please help me. I am a struggling scientist that is good at experiments, but terrible at stats. Thanks so much, I really appreciate it!

8. ## Re: Multiple Groups Regression

If I understood correctly you have one dependent and 3 independent variables.
Do you use SPSS?

9. ## Re: Multiple Groups Regression

Yes, but one of my independents has a category value (young= 1, average=2, old=3). Yes, I have SPSS and Stata.

Thanks for the help!

10. ## Re: Multiple Groups Regression

Listen what you 'll do:

Go to Analyze>General Linear Model> Univariate.
In the dialogue box which appears you put the dependent variable in the first white box.
The cateogrical variable you just mentioned goes to Fixed Factor(s) white box.
The other two continuous variables go to Covariate(s) white box.
Choose the options and another dialogue box will appear. There you click on Parameter estimates and Homogeneity tests. Then press ok and you back to the first dialogue box. Press OK and that's it.

11. ## Re: Multiple Groups Regression

Between-Subjects Factors
Age Group N
1 Age Group 1 22
2 Age Group 2 16
3 Age Group 3 17

Ok, that is what it spit out at me. Where is the p value to compare the three groups of ages? Thanks so much for the help and quick responses I really appreciate it!

12. ## Re: Multiple Groups Regression

LOL, is this what i told you to do?
You did something wrond man.
You have splitted the data based on the ages? What is TBSA, Baux and InhalationGrade?
How many independent variables do you have?

13. ## Re: Multiple Groups Regression

Sorry. I split the data earlier based on some other post. My bad. Here is what it came out with unsplit. I am just trying to find the p value between the three groups.

TBSA= Total body surface area
Inhlation grade= inhalation injury severity score
Baux = Score base don burn injury

Basically, I want to look at the age difference in expression of a chemical, but I need to account for the differences in the TBSA, Inhalation injury grade, and the Baux score. If that makes any sense to you, sorry I am a terrible explainer.

Thanks.

14. ## Re: Multiple Groups Regression

The p-values you want are in the last column, (Sig.). As you can see only inhalation grade is significant.
But, try this also.
Go in the dialougue box when you enter all the variables. There select Model.
You will see another dialogue box.
Choose one by one the variables and transfer them to the right box. Make sure the choice in the middle (between the two boxes) is on main effects.
AFTER you transfer them on the right change the middle option into interactions.
Now, pick all the pairs formed between age group and the other variables. For instance, select age group and Inhalation grade and tranfer it on the right.
Then select age group and Baux and transfer it on the right. Finally do the same for TBSA and age group. Then press OK and you return to the main dialogue Box. Press ok and send it again.
You did not send me the Levene's test (The homogeneity test you were suppose to selcet in the options tab).

15. ## Re: Multiple Groups Regression

I have no idea what any of this means, but here is what it spit out. If you could enlighten me?