Intended for healthcare professionals

Letters

Postcodes don't indicate individuals' social class

BMJ 1999; 319 doi: https://doi.org/10.1136/bmj.319.7215.1003 (Published 09 October 1999) Cite this as: BMJ 1999;319:1003
  1. Philip McLoone, research associate,
  2. Anne Ellaway, research
  1. MRC Social and Public Health Sciences Unit, University of Glasgow, Glasgow G12 8RZ

    EDITOR—In their report Danesh and colleagues said that an individual's postcode of residence could be used as a marker for his or her social class.1 This inference was based on the association between income reported in a survey and estimated household income for postcodes supplied by a commercial company. The accompanying commentary by Ben-Shlomo and Davey Smith cautioned that researchers should not attribute population characteristics to individuals—the ecological fallacy.2 These authors questioned the meaning of the association reported by Danesh et al and wondered how this correlation might change with different levels of aggregation or how it differs from associations found with indicators based on census data.

    In Glasgow we are evaluating the utility of similar income data supplied by a different commercial company.3 Using data at higher levels of aggregation than used by Danesh et al and self reported household income for 2175 people in 1579 enumeration districts,4 we report that the correlation between enumeration district income and self reported income was 0.48 (99% confidence interval 0.44 to 0.52). At postcode sector level this correlation was 0.38 (0.33 to 0.42). In comparison, the correlation between Carstairs scores (a deprivation indicator derived from census data) and self reported income at enumeration district level was −0.44 (−0.48 to −0.39) and −0.35 (−0.40 to −0.30) at postcode sector level. These associations are of similar magnitude to the correlation reported by Danesh et al. It is also worth noting that there was little difference in these correlations whether Carstairs scores or commercially supplied income data were used. On the basis of these findings, however, we do not suggest that enumeration districts or postcode sectors can be used as a proxy for an individual's social class. The figure shows the distribution of household income in two groups containing enumeration districts that are estimated to have the highest or lowest average income in Scotland. Each group was defined to contain 1% of all households. It shows a broad range of income with a substantial overlap in the two distributions. Such variation in individual income is likely to occur if postcodes were also used.

    Figure1

    Income distribution in enumeration districts grouped by high or low average income. Each group contains 1% of all households in Scotland

    The ecological fallacy apart, the correlations we have found mean that it would be unsafe to assume an individual's social class on the basis of his or her address.

    References

    1. 1.
    2. 2.
    3. 3.
    4. 4.