This is the data package for Lugtig (submitted) The relative size of measurement error and attrition error in a panel survey. Comparing them with new Multi-Trait Multi-Method model.
questions? p.lugtig@uu.nl
last update: 03-11-2017


In order to replicate the results in the paper, the following steps are necessary:

1. Software requirements
Need SPSS 24.0, R 3.3.0 and Mplus 8.0 to do the analyses.
2. data requirements:
The study uses two publicly available datafiles:
a. au09a1.op_en.sav . This is the original datset as available on www.lissdata.nl and cited in the paper. The codebook is included in the package (codebook MTMM LISS.pdf).
b. participation_panelmember_fieldworkperiod.sav. This is a file that is composed from the first 48 full months of data of the LISS panel. It contains information on whether a respondents participated in any given wave of the study. The data can be created using the publicly available datafiles at www.lissdata.nl, but given that this needs more than 100 separate datafiles, I opted to provide this file here. Another reason for doing this is that the LISS panel sometimes retroactively deletes respondents from the panel to comply with privacy and data protection laws. By providing the cumulative response file, I avoid such problems.

3. steps in the analyses
a. First run the spss syntax file "syntax common metric".  This does some data management (merging the two datafiles for example) and creates an R-ready file
b. run the mplus input file "meref basic trust institutions.inp". This generates the descriptive statistics of table 1.
c. run the R syntax "syntax R-Mplus MTMM model.R. Within this syntax a.txt template is used for building 35 different Mplus.inp files that are used to run Mplus models automatically. This file is called "template MTMMpolitical trust meref.txt".
The syntax generates the tables and figures used in the paper automatically in a folder called "paper"

4. results not reported in the paper in detail (but mentioned in discussion) that study the effects on two other MTMM-experiments are shown in a separate R-syntax (extraa analyses in folder "additional analyses"). Run this file to replicate the models for leftright placement MTMM and social trust. Note again that these results are not discussed. Results do differ from the political trust scales (including some conversion problems in social trust), 
but I found the differences in results not to be meaningful. 
5. For completeness, the complete workspace (output) of R is given in ".Rdata"
	

