🚀 go-pugleaf

Deanna Wild <dkw4@NIOSHE2.EM.CDC.GOV> writes...
> Anyone familiar with a FORMAL 'test for (linear in the log odds) trend'
> in logistic regression?
> We tend to use an informal approach here - i.e.; checking to see if quadratic
> terms or the like are needed beyond the linear term.  Or, we categorize the
> continuous independent variable and check to see how consistent the change in
> the resulting indicator variables' coefficients is.
> ...

Hosmer & Lemeshow (Applied Logistic Regression, Wiley 1989, p.90)
discuss the problem of scale identification.
They consider an informal "eyball assessment" approach (categorizing
the independent variable, plotting the estimated coefficients versus
the midpoints of the groups, viewing) sufficiently accurate. They
refer to other methods as well, eg. the Box-Tidwell approach
which adds the term x.ln(x) to the model.


Vaclav Fidler