I am trying to calculate a series of unadjusted odds ratios for my dataframe in R. Some of the crosstables contain zero cell sizes, and from what I've read I believe using the Haldane-Anscombe correction to add 0.5 to all cell sizes in those tables is an appropriate next step.
I can't share my actual dataset, so I created a random little sample dataset below that treats "male" as the outcome and "eye_color" and "hair_color" as the predictors, showing how I'm currently calculating the ORs below. In this sample dataset, there are 0 people who are male with green eyes.
#Creating sample dataset
male <- c(1,1,1,1,1,1,1,1,0,0,0,0,0,0,0)
eye_color <- c("blue","blue","blue","blue","brown","brown","brown","brown","blue","blue","blue","brown","brown","green","green")
hair_color <- c("brown","brown","brown","black","black","brown","brown","blonde","blonde","blonde","black","brown","brown","black","black")
df <- data.frame(male, eye_color, hair_color)
#Crosstable stratified by male
crosstable(df, c(eye_color,hair_color), by=male, percent_digits=2) -> ctable
#Calculating list of ORs
vars <- c('eye_color','hair_color')
cols <- df[vars]
ors_list <- lapply(as.list(cols), function(x) glm(male ~ x, data=df, family=binomial(link="logit")))
#Creating tibble from list of ORs
do.call(rbind, lapply(ors_list, broom::tidy, exponentiate=TRUE, conf.int=TRUE)) -> ors
The only examples of the Haldane-Anscombe correction I've been able to find show people manually adding 0.5 to all cell sizes and use different methods of calculating the ORs. Is there a way to incorporate the correction into the code I'm using? Or a way to apply the correction using different code but generating the same result?
Also, please let me know if Haldane-Anscombe is not appropriate for this case and there's another method I should be considering. Thank you!
