R 'memisc' package: why has "as.data.frame()" changed 0/1 values of data.set to 1/2 in data.frame?

[MarcoK]

Hi all, I'm trying to prepare an SPSS .sav data file with survey data for performing analyses in R.
Now I have an issue that some variables with binary values 0/1 (signifying no/yes) have been transformed unexpectedly.

I have used the memisc package to import the data as a data.set object.

Dset.core <- spss.system.file(file="C://..../data_coded.sav",
                            varlab.file=NULL,
                            codes.file=NULL,
                            missval.file=NULL,
                            count.cases=TRUE,
                            to.lower=FALSE      ## set to FALSE since we want to keep the upper case VarNames
)

This worked all fine, from what I saw from str() and codebook() outputs. One example of a 0/1 variable $AMEVYES (labels are 0=no, 1=yes) is shown here:

str(Dset.core)
Data set with 1999 obs. of 106 variables:
(...)
$ AMEVYES : Nmnl. item w/ 2 labels for 0,1 num 0 0 0 0 0 0 0 0 0 1 ...

I now want to convert the special data.set object created by memisc into a data frame with:

Dset2Df.core <- as.data.frame(Dset.core)

As intended, the nominal 0/1 variable was changed into a factor variable with corresponding levels. But for some strange reason, this procedure also changed the values of the variables, from 0/1 to 1/2, like in this example output:

str(Dset2Df.core)
'data.frame': 1999 obs. of 106 variables:
(...)
$ AMEVYES : Factor w/ 2 levels "No","Yes": 1 1 1 1 1 1 1 1 1 2 ...

Why did this happen, and most importantly, how can I stop this from happening?
Many thanks for a hint!

PS: I'm rather new to R and new to this forum, so please excuse if I missed any best practices when formulating my question.

[andresrcs]

Doy you have to necessarily use nemisc for any reason? If not, you can try haven package to read .sav files directly as tibbles (tibbles are a modern take on data frames), and if you want, you can use as.data.frame without any problem

[jdlong]

@MarcoK

As intended, the nominal 0/1 variable was changed into a factor variable with corresponding levels. But for some strange reason, this procedure also changed the values of the variables, from 0/1 to 1/2 , like in this example output:

I think there's a gap in your understanding of factors.

Factors map your character values to a index. By default the index starts at 1 and goes up to n where n is the number of unique values in your field.

It's hard to know without having some dummy data, but I don't think there's any conversion from 0/1 happening. I think you are seeing R map the values "Yes" and "No" to a factor with levels "Yes" and "No" with underlying index of 1, 2.

Intro to factors:
http://monashbioinformaticsplatform.github.io/2015-09-28-rbioinformatics-intro-r/01-supp-factors.html

[MarcoK]

Thanks to all! I actually had assessed both package descriptions of 'haven' and 'memisc'. Then I selected the latter, because I was impressed with their functionalities around defining/manipulating various label types of survey data sets, possibility to generate a codebook() etc. - like what I know from traditional social science statistics software.
But you are right, if I'm not able to solve this problem above, I will likely switch to 'haven' to complete my work. Also because it's comfortable to use as a built-in RStudio import manager for .sav files.