The authors show how data from the 2% Sample of Anonymised Records (SA
R) can be combined with data from the Small Area Statistics (SAS) data
base to investigate the causes of the ecological fallacy in an Enumera
tion District (ED) level analysis. A range of census variables are exa
mined in three 'SAR districts' (local authority districts with populat
ions of 120 000 or more, or combinations of contiguous districts with
smaller populations) in England. Results of comparable analyses from t
he 1986 Australian census are also given. The ecological fallacy arise
s when results from an analysis based on area-level aggregate statisti
cs are incorrectly assumed to apply at the individual level. In genera
l the results are different because individuals in the same area tend
to have similar characteristics: a phenomenon known as within-area hom
ogeneity. A statistical model is presented which allows for within-are
a homogeneity. This model may be used to explain the effects of aggreg
ation on variances, covariances, and correlations. A methodology is in
troduced which allows aggregate-level statistics to be adjusted by usi
ng individual-level information on those variables that explain much o
f the within-area homogeneity. This methodology appears to be effectiv
e in adjusting census data analyses, and the results suggest that the
SAR is a valuable source of adjustment information for aggregate data
analyses from census and other sources.