Can't use factors() any more in lrm() about rms

[harrelfe]

Run y ~ Year then y ~ scored(Year) which will give you a test of association that is the same as using factor() as well as something you don't get - a test of linearity against all alternatives to linearity (factor). Hear Year is numeric interval-scaled when stored in the data frame/data table.

[harrelfe]

To open an issue without providing the code you wrote that caused the error is not a good way to report problems.

I am fairly certain that you used factor( ) or as.factor( ) inside the model formula. It was the design of rms that this never worked. It only worked by accident before. Now it is explicitly disallowed. You need to define the type of variable it is in the data frame or data table.

To go to the trouble of downgrading to older versions is quite a waste of time when the solution is extremely easy.

[MarcoPortmann]

My apologies, I assumed it should be clear that the complete code would look like that:
lrm(y~x1+factor(x2), data=mydata)
where x2 could be any type of variable. Yes, it took me quite some time to downgrade but I really assumed that this must be a bug in the latest version of either Hmisc or rms. I’ve been using this notation extensively in many of my R scripts for years without any issues and it still seems to work in other functions like lm(). If it’s not a bug then I can close this issue.
I always found it extremely handy to use this notation as it allows to quickly adapt a regression from something like
lrm(y~x1+Year, data=mydata)
to
lrm(y~x1+factor(Year), data=mydata)
where Year enters as numeric variable in the first case while the second version estimates the model with year fixed effects. May I kindly ask how I can achieve the same without using factor() in the formula and without defining the variable as a factor in the data frame?

[MarcoPortmann]

Thank you very much for answer. With scored I get even more information and I can adapt my scripts with search+replace.