Constructing a composite index to measure loneliness amongst older populations

Loneliness is a social phenomenon that is gaining increased recognition in the UK. Studies have found it to be as damaging to an individual’s health as obesity and smoking. The British Government, and several charities, including The Campaign to End Loneliness, Age UK and the British Red Cross have highlighted a lack of tools for appropriately measuring and analysing loneliness in older populations. Thus, authorities do not know who is affected by loneliness most severely, or where they reside. Presently, measures of loneliness depend on surveys asking individuals whether they are lonely. These measures have many conceptual and practical drawbacks, importantly, small sample sizes mean they are not directly applicable at the neighbourhood level. A new composite index that measures loneliness in older populations at the area level is presented here. Its construction takes two primary methodological steps clearly described in this thesis. Firstly, it analyses the primary characteristics that are associated with loneliness in old age through national survey data, the wider literature, and consultation with stakeholders. Secondly, relevant indicators are selected, normalised, weighted and aggregated into a reproducible composite index that identifies neighbourhoods at highest risk of loneliness in older populations. The key characteristics associated with loneliness in older populations are widowhood, living alone, poor health and low income. Four other peripheral characteristics are also identified, these include national language ability, rates of hate crime in the local area, smoking status and provision of informal care. The index reveals that the spatial distribution of loneliness in older populations in England is clustered in coastal areas and former industrial Northern cities. The neighbourhood at highest risk of loneliness in England is in Christchurch, Dorset. Loneliness amongst older populations is found to be more common in rural areas and is not found to have a strong relationship with multiple deprivation.