Re: test for trend
- From: "Ray Koopman" <koopman@xxxxxx>
- Date: 1 Feb 2006 15:07:38 -0800
raoulreulen@xxxxxxxxxxx wrote:
> Hi,
>
> How do I perform a test for linear trend of a variable that is coded as
> a dummy variable in a logistic regression analysis? Is it as simple as
> dropping the dummy variable, and using an ordinal variable instead?
> Will the p-value of the ordinal variable give me the test for trend?
>
> an example:
>
> let's say I have the variable age which is categorized into 4 groups. I
> use the lowest category as the reference catogory.
>
> My model looks like this:
>
> log(odds) = constant + age1 + age2 +age3 (age is a dummy in which age0
> is used a reference category)
>
>
>
> Would the following model give me the p-value for the test for linear
> trend in the variable age?
>
> log(odds)=constant +age (age is coded as 1,2,3,4)
>
> thanks
>
> Raoul
Use echelon coding: grp age1 age2 age3
0 0 0 0
1 1 0 0
2 1 1 0
3 1 1 1
If the coefficients of age1, age2, and age3 are all positive and
significant then you can conclude that log(odds) increases with age.
.
- References:
- test for trend
- From: raoulreulen
- test for trend
- Prev by Date: test for trend
- Next by Date: Re: test for trend
- Previous by thread: test for trend
- Next by thread: Re: test for trend
- Index(es):
Relevant Pages
|
Loading