How are missing values coded in WERS 2004?
The standard approach to the coding of missing values in WERS 2004 is to use the same three codes in each data file: -9 (Not answered / Refused); -8 (Dont know); and -1 (Not applicable, i.e. Not asked). A small number of variables have additional missing values to identify particular circumstances: an example is the use of -6 on single-coded questions in the SEQ that were inadvertently multi-coded by respondents (e.g. B2). In the SPSS data files, these codes are designated as user missing values and so are automatically excluded from any statistical procedures, such as the production of means. However, in the Stata data files, the codes are not designated as missing values and they will therefore be treated as valid values by any statistical procedures. The syntax command cited below can be used to designate the three standard codes (-9, -8 and -1) as missing values in the Stata data files. Such codes will then be excluded by statistical procedures, such as tabulate- or summarize-. How
