[R] Marginal Effects of continuous variable in lrm model (Design package)

[Frank E Harrell Jr]

Minyu Chen wrote:
> Dear all:
> 
> When I am trying to get the marginal effects:
> 
> summary(result7,adv_inc_ratio=mean(m9201 
> $adv_inc_ratio),adv_price_ratio=mean(m9201$adv_price_ratio), ...(SOME  
> MORE CONTINUOUS AND DISCRETE VARIABLES BUT I AM NOT LISTING)... regW=c 
> (0,mean(m9201$regW),1), regWM=c(0,mean(m9201$regWM),1))
> 
> It gave out an error message:
> 
> Error in summary.Design(result7, adv_inc_ratio = mean(m9201 
> $adv_inc_ratio),  :
> 	ranges not defined here or with datadist for adv_inc_ratio  
> adv_price_ratio agem1 change2 yr1Libor yr1Libor1mtLag yr1Libor1yrLag  
> change21yrLag change21mtLag fwdReal4yr fwdInfl4yr
> 
> But I remember from my previous operation a few months ago (I  
> recorded the commands) that to summary the marginal effect, I don't  
> have to specify the ranges for the discrete variables. However, I use  
> the command again (with slight modification because of the newly  
> added variables) it doesn't work. I don't know what went wrong.
> 
> Thank you very much.
> 
> Thanks,
> Minyu Chen

Please read the package's help files especially about the datadist function.

Frank


-- 
Frank E Harrell Jr   Professor and Chair           School of Medicine
                      Department of Biostatistics   Vanderbilt University