5

I would like to access some elements of an Anova summary in R. I've been trying things like in this question Access or parse elements in summary() in R.

When I convert the summary to a string it shows something like this:

str(summ)
List of 1
 $ :Classes 'anova' and 'data.frame':   2 obs. of  5 variables:
  ..$ Df     : num [1:2] 3 60
  ..$ Sum Sq : num [1:2] 0.457 2.647
  ..$ Mean Sq: num [1:2] 0.1523 0.0441
  ..$ F value: num [1:2] 3.45 NA
  ..$ Pr(>F) : num [1:2] 0.022 NA
 - attr(*, "class")= chr [1:2] "summary.aov" "listof"

How can I access the F value? I've been trying things like summ[c('F value')] and I still can't get it to work.

Any help would be greatly appreciated!

Community
  • 1
  • 1
user3801533
  • 321
  • 1
  • 6
  • 15

2 Answers2

6

You have the anova object inside a list (first line of str output is List of 1). So you need to get the "F value" of this single element, like:

 summm[[1]][["F value"]]
tfc
  • 596
  • 3
  • 7
3

As an addition to the answer above I'd recommend to start using the broom package when you want to access/use various elements of a model object.

First, by using the str command you don't convert the summary into a string, but you just see the structure of your summary, which is a list. So, str means "structure".

The broom package enables you to save the info of your model object as a data frame, which is easier to manipulate. Check my simple example:

library(broom)

fit <- aov(mpg ~ vs, data = mtcars)

# check the summary of the ANOVA (not possible to access info/elements)
fit2 = summary(fit)

fit2

#             Df Sum Sq Mean Sq F value   Pr(>F)    
# vs           1  496.5   496.5   23.66 3.42e-05 ***
# Residuals   30  629.5    21.0 


# create a data frame of the ANOVA
fit3 = tidy(fit)

fit3

#        term df    sumsq    meansq statistic      p.value
# 1        vs  1 496.5279 496.52790  23.66224 3.415937e-05
# 2 Residuals 30 629.5193  20.98398        NA           NA


# get F value (or any other values)
fit3$statistic[1]

#[1] 23.66224

I think for the specific example you provided you don't really need to use the broom method, but if it happens to deal with more complicated model objects it will be really useful to try it.

AntoniosK
  • 15,991
  • 2
  • 19
  • 32