turn on suggestions

Auto-suggest helps you quickly narrow down your search results by suggesting possible matches as you type.

Showing results for

- JMP User Community
- :
- Discussions
- :
- how to interpret a regression with a dummy variabl...

Topic Options

- Subscribe to RSS Feed
- Mark Topic as New
- Mark Topic as Read
- Float this Topic for Current User
- Bookmark
- Subscribe
- Printer Friendly Page

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Get Direct Link
- Email to a Friend
- Report Inappropriate Content

Sep 19, 2016 6:58 AM
(744 views)

Hi Peeps,

I'm a student from Germany and writing my bachelor thesis and I really hope that you can help me with the interpretation of my regression analysis.

I need to check the influence of a independent variable on three dependant variables in two different categories. That's why I added a dummy variable which is coded with 1's and 0's (1 for the first category and 0 for the second) as an independent variable.

Unfortunately I cannot really handle the output. I think that something went wrong.. or maybe it didn't? Maybe you can help me and say if I did the regression model right and how I can interpret the output. I'll put a screenshot beneath. Unfortunately it's in german, but actually the tools is quit the same as in english.. if you have question, just ask and I will try to answer as well as I can.

Thaaaaank yooou verry much!!!

7 REPLIES

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Get Direct Link
- Email to a Friend
- Report Inappropriate Content

Sep 19, 2016 10:46 AM
(640 views)

Vanessa,

If your dummy variable is truly a separation of categories then you try using it as a "By" variable as opposed to an independent variable. You will get two separate fits. If you believe it should be an independent variable then make sure it is coded as Nominal and then you will see the effect in your fit model.

HTH

Bill

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Get Direct Link
- Email to a Friend
- Report Inappropriate Content

Sep 20, 2016 1:59 AM
(640 views)

Hi Bill,

thank you very much for your response! I already tried the function "By" and I will use it if I don't find a good way to use it as a independant variable. As you said I coded it as nominal and it really looked better and I had a different graph for each group.. so, I think that's the solution.

But: I read here https://community.jmp.com/docs/DOC-7537 that you have to make sure, that the variable is continous for a regression and not nominal.. that's why I'm not sure whats right or wrong..

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Get Direct Link
- Email to a Friend
- Report Inappropriate Content

Sep 20, 2016 5:19 AM
(640 views)

Hey vanessa_bama,

changing the modeling type to nominal is correct. yet, for a more direct comparison between two categories you should use the ordinal modeling type.

Ii would also try to add the interaction between the independent variables to the model. this will allow you to test the hypothesis that each category has a different slope.

use the "By" option only if you strictly do not want to test for the differences between groups in their parameter estimates or you want to see the differences in model fit.

best,

ron

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Get Direct Link
- Email to a Friend
- Report Inappropriate Content

Sep 20, 2016 6:00 AM
(640 views)

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Get Direct Link
- Email to a Friend
- Report Inappropriate Content

Sep 20, 2016 6:45 AM
(640 views)

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Get Direct Link
- Email to a Friend
- Report Inappropriate Content

Sep 20, 2016 6:56 AM
(640 views)

Thaaaanks a lot ron_horne

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Get Direct Link
- Email to a Friend
- Report Inappropriate Content

Sep 20, 2016 8:50 AM
(640 views)

Perhaps regression will not yield well with your data set. Based on your output report, it does not look like you have a good model yet. Have you tried neural networks and/or partitioning? You might get better predictive models this way. Discriminant Analysis might also be useful !

. I hope this helps!

Steve