Below is a reproducible example that illustrates the error that I've had with character vectors and summaryM. This pull request is seems to fix the problem (but I haven't tested it thoroughly).
Regards,
Matt
## load Hmisc package
library('Hmisc')
## Hmisc version on my machine:
## "Hmisc 4.0-3 2017-04-30 /home/matt/R/x86_64-pc-linux-gnu-library/3.3"
version('Hmisc')
## create list of words to sample
word_list <- replicate(100,
paste0(letters[sample.int(26,10)],collapse=''))
## create data frame
dat <- data.frame(
strata=sample(c(T,F),1000,rep=T),
words=sample(word_list,1000,rep=T))
## ensure that 'words' is character
## By default, it's converted to factor
## when the data frame is created.
dat$words <- as.character(dat$words)
## call that generates error in call to 'quantile.default'
## "Error in (1 - h) * qs[i] : non-numeric argument to binary operator"
summaryM(words ~ strata, data=dat)
Frank,
Below is a reproducible example that illustrates the error that I've had with character vectors and summaryM. This pull request is seems to fix the problem (but I haven't tested it thoroughly).
Regards, Matt