Using residents and experts to evaluate the validity of areal wombling for detecting social boundaries: A small-scale feasibility study.

Several studies have explored the relationship between socially constructed neighbourhood boundaries (henceforth social boundaries) and ethnic tensions. To measure these relationships, studies have used area-level demographic data to predict the location of social boundaries and their characteristics. The most common approach uses areal wombling to locate neighbouring areas with large differences in residential characteristics. Areas with large differences (or higher boundary values) are used as a proxy for well-defined social boundaries. However, to date, the results of these predictions have never been empirically validated. This article presents results from a simple discrete choice experiment designed to test whether the areal wombling approach to boundary detection produces social boundaries that are recognisable to local residents and experts as such. We conducted a small feasibility trial with residents and experts in Rotherham, England. Our results shows that participants were more likely to recognise boundaries with higher boundary values as local community borders. We end with a discussion on the scalability of the design and suggest future improvements.

Estimating the effects of crime maps on house prices using an (Un)natural experiment: A study protocol.

Street-level crime maps are publicly available online in England and Wales. However, there was initial resistance to the publication of such fine-grained crime statistics, which can lower house prices and increase insurance premiums in high crime neighbourhoods. Identifying the causal effect of public crime statistics is difficult since crime statistics generally mirror actual crime. To address this question empirically, we would ideally experiment and introduce a source of random variation in the crime statistics. For instance, we could randomly increase or decrease the number of offences displayed in crime statistics and measure their effects on local house prices. For obvious reasons, we cannot pursue this research design. However, street-level crime maps contain intentional errors, which are the product of a geomasking algorithm designed to mask the location of crimes and protect the identity of victims. This project leverages features associated with the geomasking algorithm to estimate the effect of public crime statistics on house prices.

Social work contact in a UK cohort study: Under-reporting, predictors of contact and the emotional and behavioural problems of children.

BACKGROUND: There is little evidence on the degree of underreporting of social work contact in survey studies. There are also few studies about the emotional and behavioural problems of children in families who have contact with social workers, despite the adversities they face. OBJECTIVE: We examine underreporting of social work contact; the predictors of social work contact with families; and links between social work contact and emotional and behavioural outcomes for children. PARTICIPANTS AND SETTING: The Avon Longitudinal Study of Parents and Children follows the health and development of 14,062 children born in 1991-2 to women living in and around the City of Bristol. The study includes self-reported information about mothers' contact with social workers and children's emotional and behavioural outcomes using the Strengths and Difficulties Questionnaire (SDQ). METHODS: Self-reporting of social work contact in ALSPAC is compared with administrative data from the child protection register to examine rates of underreporting. We use a fixed-effects model to account for measurement error in estimating the association between social work contact and potential predictors, namely gender of the child and maternal factors: marital status, trouble with law, hospitalisation, cannabis and alcohol use, employment, financial difficulty and experience of cruelty from a partner. SDQ scores are explored using linear regression with lagged indicators of social work contact. RESULTS: The probability of a false negative for self-report of social work contact ranged from 22% to 34% across three survey waves. Mothers who married within the last 12 months were less likely to receive social work contact (OR: 0.13, CI95%: 0.01-1.34). Those who had either been hospitalised (OR: 1.52, CI95%: 1.01 - 2.28) or increased their rate of alcohol consumption (OR: 2.14, CI95%: 0.91 - 5.07) within the last 12 months were more likely to receive such contact. Overall children whose mothers report social work contact were much more likely to have emotional and behavioural problems within the first seven years of their lives (p < 0.01). CONCLUSION: There is potentially a high degree of underreporting of social work contact in social surveys and cohort studies. Researchers should adopt methods to account for this issue in the future. The risk of emotional and behavioural problems is greater among children whose mothers have had contact with social workers compared to other children with seemingly similar adversities.