Here, you can download the files needed to replicate the results obtained with the software BUGS in the paper. First you need to download the software BUGS and documentation from the url: http://www.mrc-bsu.cam.ac.uk/bugs/. The software is distributed for free.

Description of the files (to download use Save Target as on these links):

• Envi_bugs: The file contains the commands and the data for fitting a one-factor model to six ordinal variables.

• Mobi1f_bugs: The file contains the commands and the data for fitting a one-factor model to eight binary variables.

• Mobi2f_bugs: The file contains the commands and the data for fitting a two-factor model to eight binary variables.

• Mobimis_bugs: The file contains the commands and the data for fitting a two-factor model to eight binary variables with missing values.

If one wants to fit latent variable models to binary, ordinal, nominal and metric variables using the full maximum likelihood approach with the E-M algorithm, s/he could use the software GENLAT. The software and documentation can be downloaded for free from the url: http://multilevel.ioe.ac.uk/team/aimdss.html (to be found on the link: go to software for chapters 7 to 9).

• Environ.inp: The file contains the input file for fitting a one-factor model to ordinal variables.

• Mobility.inp: the file contains the input file for fitting a one-factor model to binary variables. (See software documentation GenlatReadMe.txt on how to modify the input file to allow for the two-factor model to be fitted). For speeding up the program we have restricted in the version of GENLAT provided on the web the sample size to be no more than 3000 observations. Therefore, if one wants to use the program should take a random sub-sample of up to 3000 observations from the mobility data set.

NOTE: GENLAT in its present form cannot fit the model with missing values. The program will be amended to allow for that model to be fitted in due course.