The analyses presented in exemplar 1 are based on the FRS household file for the financial year 2002/03. The original data were obtained from the ESRC data archive and a small file was produced from this with just a few variables.

During this process steps were taken to ensure that no individual could be identified from the data. In particular the following steps were taken:-

• The identifiers for the primary sampling units were changed so that they no longer convey any geographic information
• The case identifiers (SERNUM in the file) have been changed.
• Random noise was added to the weights and to the income data

The programs below are provided to allow you to see how the data were extracted from the large survey files and made ready for analysis by different packages.

It is not expected that you would want to use these, but they are provided to show how the data relate to that on the archive.

The following steps were used :-

• - The Stata data file from the archive was exported to SAS and a SAS program was used to select variables and make the data anonymous : ex1_prep.sas   It produces a SAS data set for analysis (ex1.sas7bdat) and an export file  (ex1.xpt)  to read into other systems.
• - The code to read the file ex1.xpt into R is in the file  ex1_prep.R  produces ex1.Rdata.
• - To read ex1.xpt into Stata is  exp1_prep.do was used and produced ex1.dta
• - To read ex1.sas7bdat into SPSS the code ex1_prep.SPS was used and produces ex1.sav.

The data sets ( in bold )  mentioned here are restricted to the variables needed for these analyses. They can all be accessed from the main page of exemplar 1.