What was the process for editing and imputing data?

As in 1991, a key role of the processing stage is to identify missing, incomplete and inconsistent answers in the information provided by respondents. As in 1991, the entered data was subjected to a process of editing and imputation. The process is similar to that used in 1991 (see above). Inconsistent answers are subjected to an automated decision making algorithm that corrected inconsistencies based on the principle of making least change to the data. Regarding missing data, as in 1991 missing items were imputed by taking the value from a "doner" person or household in the same area. However, a notable difference from 1991 was that where more than one item on a record needed to be imputed, the same "doner" provided the value for each item (as this has been found to retain the integrity of the data better than using multiple doners for a single record).