1

I'm an R newbie who is struggling with some basic problems relating to factors for an analysis of likert items on a survey.

Specifically, there are two issues I'm struggling with - first, I'm not sure how to change similar factor names across columns of a data frame, and second, I'm looking for an efficient and easy way to collapse factor categories simultaneously across multiple columns of a data frame. For example, if columns 1 to 3 all have a seven-point scale of "Extremely dissatisfied", "Moderately dissatisfied", etc, etc, up to "Extremely satisfied", how do I collapse categories into "Dissatisfied" and "Satisfied", simultaneously across all three columns?

I'm also hoping to achieve this while working with a data frame instead of transforming to a matrix, or other data object. Perhaps my understanding of factors needs improving, but I'm aware that factors have an underlying number. I'm hoping to find a method to achieve the results while using the factor names, and not referring to the numbers. Or am I missing something obvious about factors here?

I have played around with recode from the car package, but I feel that there is probably an easier way. I also haven't been able to use recode across multiple columns by using an apply function. Is there a way to use the Plyr or Dplyr packages for this type of multi-column recoding?

Similarly, what's the easiest way to rename similar factors across columns simultaneously? For example, renaming "Extremely satisfied" to "Ext. sat" across columns 1 to 3.

Do I use "apply" or "lapply" together with a function?

I have searched and Googled for an answer to what seems like it should be a relatively easy and common task in R, but couldn't find anything that worked or was simple enough for me to understand.

Any suggestions would be greatly appreciated! This is also my first Stack overflow post, so let me know how I can improve my question format, wording, etc.

Thanks! Mike

Mike
  • 2,017
  • 6
  • 26
  • 53
  • Welcome to R and SO. Here is a way with `replace` http://stackoverflow.com/questions/28637880/renaming-factor-variables-if-a-condition-is-satisfied-in-separate-column and one with `revalue` here http://stackoverflow.com/questions/34221338/how-can-we-replace-elements-in-a-vector-in-r. If you add some example data you may receive more detailed help. – Pierre L Feb 25 '16 at 05:08
  • It would be a lot easier to help and offer specific recommendations if you provided a [reproducible example](http://stackoverflow.com/questions/5963269/how-to-make-a-great-r-reproducible-example). It's easier to write working code rather than just talk about code or general strategies. Stack Overflow is for specific Q&As, not general discussions. – MrFlick Feb 25 '16 at 05:32
  • Thank you for the comments and advice, they were helpful. – Mike Feb 26 '16 at 04:54

2 Answers2

0

You can assign to levels of a factor to make groups relatively easily, while retaining the factor structure. E.g.:

df <- data.frame(replicate(3, factor(1:4,levels=1:4, 
                 labels=c("extr.dis","dis","sat","extr.sat"))))
df

#        X1       X2       X3
#1 extr.dis extr.dis extr.dis
#2      dis      dis      dis
#3      sat      sat      sat
#4 extr.sat extr.sat extr.sat

df[] <- lapply(df, function(x) {
  levels(x)[levels(x) %in% c("extr.dis","dis")] <- "dissat"
  levels(x)[levels(x) %in% c("sat","extr.sat")] <- "sat"
  x
})
df

#      X1     X2     X3
#1 dissat dissat dissat
#2 dissat dissat dissat
#3    sat    sat    sat
#4    sat    sat    sat
thelatemail
  • 91,185
  • 12
  • 128
  • 188
0

We could use recode from library(car) (example data from @thelatemail's post)

library(car)
df[] <-  lapply(df, recode, 
            'c("extr.dis","dis")="dissat";"extr.sat"="sat"')


 df
 #    X1     X2     X3
 #1 dissat dissat dissat
 #2 dissat dissat dissat
 #3    sat    sat    sat
 #4    sat    sat    sat

str(df)
#'data.frame':   4 obs. of  3 variables:
# $ X1: Factor w/ 2 levels "dissat","sat": 1 1 2 2
# $ X2: Factor w/ 2 levels "dissat","sat": 1 1 2 2
# $ X3: Factor w/ 2 levels "dissat","sat": 1 1 2 2
akrun
  • 874,273
  • 37
  • 540
  • 662
  • Thanks so much, this worked and is an easy solution! – Mike Feb 26 '16 at 04:56
  • Another quick question. I have changed factor names before by using: levels(df$Q1)<-c("Newname1","Newname2",etc) where Q1 is the factor, and "Newnames" are the new factor category names. I find this quick and easy to use, but I can only use it for one column. How do I apply this across multiple columns with similar factor names? – Mike Feb 26 '16 at 04:58
  • Also, is there a way to achieve the above results (your original answer to my question with recode) by using "ifelse" or "if...else"? Thanks again! – Mike Feb 26 '16 at 05:03
  • @Mike I am not sure why you need `ifelse`. Can you elaborate a little more or update it in your post. – akrun Feb 26 '16 at 06:17