Re: [R] 'mgcv' package, problem with predicting binomial (logit) data

ut why the NA's caused so much trouble but was eventually driven to exclude them in the subset step by mismatched lenght errors from plot(), Thanks, Jef -----Original Message----- From: David Winsemius [mailto:dwinsem...@comcast.net] Sent: maandag 30 augustus 2010 17:22 To: Jef Vlegels Sub

[R] 'mgcv' package, problem with predicting binomial (logit) data

Dear R-help list, I’m using the mgcv package to plot predictions based on the gam function. I predict the chance of being a (frequent) participant at theater plays vs. not being a participant by age. Because my outcome variable is dichotomous, I use the binomial family with logit link function.