Intra-urban variations in male unemployment are hypothesised to depend on residential variations in the personal characteristics of adult males, (age, marital status, family size, skill level, housing tenure), and local characteristics, (manufacturing employment, the number of married working women, and unemployment in previous years). A causal modelling approach is adopted, in which the nature and direction of the causal relationships between the highly correlated personal and local characteristics are carefully specified a priori. The hypothesised linkages between the standardised variables are tested in a systematic manner, using step-wise regression, yielding beta-coefficients. Post-code Sector data, for 29 Scottish cities and towns, from the Small Areas Statistics of the 1971 and 1981 Censuses of Population, were analysed as a whole, and in seven subsets: 4 cities, Strathclyde excluding Glasgow, new towns, and the rest of Scotland. The typical pattern suggests that unemployment is related to lack of skill, number of dependents, and manufacturing employment, but not to youthfulness, council housing tenure, nor to previous unemployment. However, the disaggregated analyses show considerable variation, confirming that relationships are highly specific to location. Thus policy recommendations should be tailored to the local situation.
