I'm still not sure whether I need microdata or area data!

How do I decide?

Think though your study. Is your focus and unit of analysis the area (at whatever spatial scale) or the individual (with "place" as a secondary concern) or some aspect of process/determination?
- Since the standard tables were mainly released for 100% of the population, start by seeing whether the area data fulfil your needs. Check the variable cross-tabulations you require are available, if so, use the area data
- If you need to carry out an analysis for geographical areas at a scale below that of local authority district, then you need area data. (You may, however, be sacrificing socio-demographic detail)
- If you need cross-tabulations of variables that are not available in the standard area
- If you want to model process or determinants, then you almost certainly need microdata

The following example might help (see Dale et al., 2000: 35):
- Unemployment shows considerable variation from one area to the next, with very high pockets of unemployment concentrated in specific areas. If the requirement is to identify the areas of highest unemployment in order to target resources then area data are essential. If your analysis is along these lines, then choose area data.
- Alternatively, the requirement may be to investigate the determinants of unemployment. Unemployment varies with age and sex, there tend to be differences in ethnic group demographic profiles and education level and family responsibilities are both important predictors of unemployment. Thus, at least five variables may be required but the Census Area Statistics (CAS) do not provide this detail in a single table. Moreover, using multivariate models with area-level data risks the ecological fallacy, especially if drawn from different tables. If your analysis is similar to this in approach, then choose microdata