CODEWITHSUNDEEP
rglmlapply
nofunsally
nofunsally

Reputation: 2091

Extracting p-value from lapply list of glm fits

I am using lapply to perform several glm regressions on one dependent variable by one independent variable at a time. Right now I am specifically interested in the Pr(>|z|) of each independent variable. However, I am unsure on how to report just Pr(>|z|) using the list from lapply.

If I was just running one model at a time: coef(summary(fit))[,"Pr(>|z|)"] or summary(fit)$coefficients[,4] Would work (as described here), but trying something similar with lapply does not seem to work. Can I get just the p-values using lapply and glm with an accessor method or from directly calling from the models?

#mtcars dataset
vars <- names(mtcars)[2:8]
fits <- lapply(vars, function(x) {glm(substitute(mpg ~ i, list(i = as.name(x))), family=binomial, data = mtcars)})
lapply(fits,summary) # this works
lapply(fits, coefficients) # this works
#lapply(fits, summary(fits)$coefficients[,4])# this for example does not work

Upvotes: 5

Views: 7656

Answers (2)

mnel
mnel

Reputation: 115392

When you run lapply(fits, summary) it creates a list of summary.glm objects each of which is printed using print.summary.glm

If you save this

 summaries <- lapply(fits, summary)

You can then go through and extract the coefficient matrix

 coefmat <- lapply(summaries, '[[', 'coefficients')

and then the 4th column

 lapply(coefmat, '[', , 4)

Upvotes: 6

David Robinson
David Robinson

Reputation: 78610

You want to do:

lapply(fits, function(f) summary(f)$coefficients[,4])

However, if each item is just a p-value, you would probably rather have a vector than a list, so you could use sapply instead of lapply:

sapply(fits, function(f) summary(f)$coefficients[,4])

Upvotes: 9

Related Questions