Periodic Reporting for period 1 - CHILDCARE STRATEGIES (Complex Contagion of Childcare Strategies amongst Low-Income Parents)
Reporting period: 2022-07-01 to 2024-12-31
Access to high-quality childcare significantly improves children's long-term outcomes. However, access to and use of childcare services are unevenly distributed, leading to growing inequalities in later life. Families with low incomes are 50% less likely to use formal childcare than high-income families, and when they do, it is often for older children and less frequently. Research has shown that cost, location, and quality influence childcare usage, but policies designed to address these barriers haven't effectively reduced the gap. This project takes a fresh approach, combining insights from sociology, economics, demographics, and social policy within a framework known as "complex contagions." The goal is to understand how formal childcare spreads—or doesn't—among low-income families and to identify the barriers that prevent its wider adoption. In a "complex contagion," people only adopt a new behavior, like using formal childcare, when they are repeatedly exposed to it from multiple, diverse people in their social networks. Low-income households tend to have personal networks with long, weak ties and closely-knit groups, making it harder for new childcare strategies to take hold. This project explores how childcare strategies within work, family, and neighborhood networks influence the choices of low-income families. The research uses unique, linked administrative and survey data to examine how childcare strategies spread across these networks. By analyzing this network data with advanced techniques like multichannel sequence analysis, the project aims to uncover how networks shape childcare decisions over time. Ultimately, this study will not only change how we understand childcare strategies but also offer new insights into broader social policies and how behaviors spread within communities.
This project introduces a groundbreaking method for social research: graph sampling, used within a social survey. This technique marks a major advance in how social surveys are conducted and is part of a collaboration with Statistics Netherlands. Traditional surveys rely on random sampling, but this project uses population-scale networks—maps of large-scale social connections—as a sampling frame. This approach is new and could transform social survey design. After consulting with stakeholders like Statistics Netherlands, the European Commission, and Erasmus University’s Ethics Committee, the research team refined the original plan, replacing a random-walk approach with a community identification process. This adjustment made the process easier to implement while ensuring it met ethical and legal requirements. The team ran simulations to test how the community identification process would work, especially in light of non-response rates and the distribution of the target population. These simulations helped fine-tune the sampling method, making it more efficient and practical. The results were shared in a working paper presented at a Network Survey Sampling workshop in September 2024. This paper was also submitted for publication in Survey Research Methods. By the end of 2024, the final survey design will be confirmed, and the survey will be conducted in mid-2025 with data available for analysis in Autumn 2025. Efforts to share this new methodology extend beyond academic circles and the approach is set to be presented in a diverse range of international conferences. Additionally, Statistics Netherlands is eager to explore how they can apply this innovative graph sampling technique in their own research. The project has the potential to reshape the field of social science research by pushing the limits of what is possible with large-scale network data. Its success could influence how social surveys are conducted, offering a new standard that can be adopted by other research institutions and national statistics agencies.
There are several elements of the project which can be considered as going beyond the state of the art. The most significant is the network sampling. This is something that is, to the best of our knowledge, genuinely unprecedented and if successful would have far reaching and broad consequences. In this project we will utilise network sampling to examine whether the childcare strategies that are observable in an individual's close ‘community’ influence their own childcare strategies. At the data collection stage, this requires us to collect data which provides a sufficient density of observations in a local community in order for us to make inferences. In subsequent analysis, various identification strategies using this highly novel data will also be applied. If this is successful, the potential applications of such an approach are vast and would have extensive ramifications for the analysis of societal dynamics. It would mean that social researchers could design social surveys and relax a fundamental assumption that all observations are independent and study social dynamics at scale on a very broad range of topics. For example, voting, housing markets, demographic behaviour, attitudes, values, could all be viewed from a social dynamics perspective for the first time outside of an online context. In other, more incremental ways, the project goes beyond the state of the art. This is particularly true for the Multi-channel sequence analysis which is uniquely applied to high-resolution administrative data for the first time. Similarly, the reconceptualization of family network layers and the consequences for kin-matrix analysis also go beyond the state of the art and will have broad applications beyond the scope of this project.