skip to primary navigation skip to content

Housing, mobility and the measurement of child health from the 1911 Irish census

Housing, mobility and the measurement of child health from the 1911 Irish census

Selected research areas and preliminary conclusions

Retrospective residential persistence

Image as described adjacent Image as described adjacent

Because we are interested in how long people recorded in the 1911 census had living at those addresses, our measures of persistence look backwards from 1911 rather than forwards from 1901.

Mobility was very high, with not many more than 50 percent of households in 1911 occupying the same addresses as 18 months previously, and less that 20 per cent having been resident at the same address ten years before.

Image as described adjacent

Disaggregation by a number of factors (and confirmed with a multiple logistic regression on the influences on the chance of being resident in the same place) indicates that life-cycle factors were the paramount influence on residential mobility. Married people and male headed households were more mobile than the single and female headed, but this is shown to be a product of youth and fertility.

Infant and child mortality

Reported numbers of children ever born and children surviving (by married women) have been transformed into estimates of infant mortality for the 15 or so years prior to 1911 using the demographic techniques of indirect estimation. Although the three ward-based clusters of streets (protestant Shankill, catholic Falls and middle class Windsor) showed little difference in mobility, there was distinctly lower mortality among the children of the middle class district.

Image as described adjacent

The figure below shows infant mortality estimates for three mobility groups and demonstrates that there is a clear gradient in infant mortality associated with mobility: the more mobile the family, the higher the risk of child death. Life-cycle effects are automatically controlled for here because the points are calculated on a marital status basis, and these results also hold for the working classes. The next step is to see if those who move do so between similar or different environments and types of housing.

Image as described adjacent

Housing trajectories & mortality

In order to judge whether it is reasonable to use the information on housing in 1911 as a proxies for housing when children were at risk of death we traced a sample of couples married 11-14 years in 1911 back to find where they were living towards the beginning of their marriage, in 1901. This involved some automatic linking within the dataset, but some additional searching of the Irish census on We then compared the housing of people at both points. Our research shows that residential mobility was very high in Belfast in the decade leading up to 1911. Only about 6 per cent of couples married for 11-14 years in 1911 had been living in the same accommodation since 1901. It is only for this small percentage of couples in this marital duration that it might be legitimate to assume that the housing characteristics in 1911 also reflect those when their children were at risk of death. A sizeable minority of other couples, however, appear to have lived in very similar housing despite moving frequently: although they have the same number of rooms at their disposal, the housing may differ in other ways not measureable using the census. A small proportion of couples moved into smaller houses, but the majority moved into larger ones, suggesting that life-course progression during the early years of marriage when children were young and at risk of death means that it should not be assumed that cross-sectional information gathered at the time of a census or survey can be assumed to represent conditions a decade or even a year or two earlier. The chances that they will be able to represent specific risk factors such as type of toilet facility or ventilation are slim. It is probable however that housing indicators can still reflect wealth or social-class and that they will show a correlation with mortality through this route.

We calculated estimates of early age mortality from the 1911 census using Brass techniques of indirect estimation and compared the influences housing related factors such as the number of rooms in the house as measured in 1901 and 1911 (controlling for other socio-economic and demographic factors) and concluded that mortality was more sensitive to housing circumstances as measured in 1901 than 1911. Housing in 1911, although connected to wealth and opportunity, provides little additional insight into the factors linked to infant death, but housing in 1901 appears to have an independent effect on mortality, with those living (while their children were at risk of death) in small houses with no back access having had significantly higher mortality.