turn on suggestions

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

Showing results for

Find a Community

- Home
- /
- Analytics
- /
- Stat Procs
- /
- Covariate

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
- Highlight
- Email to a Friend
- Report Inappropriate Content

03-31-2015 09:43 AM

How to select the covariate for ANOCOVA? Is that OK if it is continuous and linearly related to the dependent varaible?

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Highlight
- Email to a Friend
- Report Inappropriate Content

04-01-2015 09:51 AM

Any helpful answers?

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Highlight
- Email to a Friend
- Report Inappropriate Content

04-01-2015 01:47 PM

This may not be helpful, but in any model building exercise, you select the model terms to be appropriate for your situation.

So, you ask a very general question, the best I can do is give you a very general answer.

Can you make your question more specific?

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Highlight
- Email to a Friend
- Report Inappropriate Content

04-06-2015 09:39 AM

I wish to know the guidelines before we select covariate for ANCOVA? Why it should be continous?

I couldn't find any documents for covariate. It would be helpful if you could find any document to understand covariate.

Thanks.

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Highlight
- Email to a Friend
- Report Inappropriate Content

04-06-2015 10:16 AM

There is no reason to require a variable in the model to be continuous. You can put any variable in the model that you think belongs, whether it is continuous or categorical.

This is simply a naming convention, if you have a model with both continuous predictor variables and categorical predictor variables, this model is called ANCOVA. If the model has only categorical predictor variables, it is called ANOVA.

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Highlight
- Email to a Friend
- Report Inappropriate Content

04-06-2015 10:27 AM

I'm experiencing ANCOVA at the moment. So I wish to know the guidelines before we select covariate which is a control varaible for ANCOVA.

Thanks for any help you offer me.

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Highlight
- Email to a Friend
- Report Inappropriate Content

04-06-2015 10:31 AM

As I said, ANCOVA is just a naming convention. If the covariate (or categorical variable) belongs in the model, based upon whatever model-building situation you are in, based upon whatever criteria are appropriate for you, then you put it in the model. If it doesn't belong in the model for whatever reason, then you don't put it in.

Normally, the only reason to put a term into a predictive model, is that you think (you have reason to believe) that this term will be predictive of the response variable(s).

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Highlight
- Email to a Friend
- Report Inappropriate Content

04-06-2015 10:48 AM

I agree with you.Thanks for continued response.

However, there is also a situation where I have multiple continuous explanatory variables which I think it is predictive of the response variable(s). But the results shows the other way.

So I need to work on a trial and error basis where my assumption and model results proves it is significant, That's the reason I questioned to understand the covariate before I put into the model. I was looking for a material to understand the covariate as well.

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Highlight
- Email to a Friend
- Report Inappropriate Content

04-06-2015 11:00 AM

However, there is also a situation where I have multiple continuous explanatory variables which I think it is predictive of the response variable(s). But the results shows the other way.

This happens to everyone. It is the nature of statistical modelling with real data where you don't have a complete theoretical understanding of what is being modelled, not every term in the model is statistically significant.

So I need to work on a trial and error basis where my assumption and model results proves it is significant, That's the reason I questioned to understand the covariate before I put into the model. I was looking for a material to understand the covariate as well.

I guess you have said this in a number of ways, and I have answered it in a number of ways, and it still isn't clear to me what you are asking, nor is it clear why my answers don't work for you. Perhaps someone else can answer.