Marii Paskov, @MariiPaskov UK Data Service Data Impact Fellow and Research Officer at the University of Oxford, shares her journey about how, when it comes to social mobility research we need to be inventive in using both survey and administrative data.
Why is intergenerational social mobility so difficult to measure?
Intergenerational social mobility indicates to what extent an individual’s social and economic position is influenced by that of their parents. The extent to which life chances depend on family circumstances is a fundamental question of social science. A highly mobile society is thought to be providing more equality of opportunity because an individual’s life chances are less strongly conditioned by characteristics over which they have no control over (nobody can choose which family they are born into).
Despite widespread interest in social mobility, robust empirical evidence on social mobility rates remains limited. This is largely because good data for analysing social mobility is hard to find and the data requirements are high – we need information about the socio-economic position of both parents and their adult children. Typically, this information is attained from national survey datasets (for example, the European Social Survey). In surveys, respondents are asked about their own socio-economic position (e.g., income, social class, occupation, education) and the socio-economic position of their parents when they were growing up. One of the problems with this approach is the fact that individual’s ability to recollect and reliably report on their parents’ socio-economic characteristics are limited. This is why measurement of mobility in terms of social class or occupational mobility has been more common. With such items the problem of recollection is less severe – people are likely to know the type of work their parents did or what education they had. This was the approach we took in a recent paper where we analysed social class mobility across 30 European countries, using cross-sectional data from the European Social Survey.
Next to intergenerational class mobility, there is widespread interest in intergenerational income mobility. Income mobility, however, provides an even larger challenge in terms of data. It is highly unlikely that survey-respondents would be able to adequately remember their parent’s earning or family income. Hence, most surveys only include information on parental occupation or education and not on parental income. This has led some researchers to impute incomes or earnings of parents based on the information about parental education and occupational. Such imputations are highly problematic and unlikely to be true reflections of parental income or earnings.
Are there alternatives to survey data?
Using survey data for monitoring intergenerational social mobility presents various challenges. These problems are typical issues relating to survey data, including a relatively small sample size and the question whether the sample is representative of the population. More recently, academics are increasingly trying to utilise alternative data sources, including administrative data. Administrative data are generally attained from administrative systems, which are typically collected by government agencies for registration, transaction and record keeping or other purposes. With adequate information, such datasets provide the opportunity to construct large panel datasets linking parents and children. One recent example of this is the work of Franz Buscha (University of Westminster) and Patrick Sturgis (University of Southampton) who turn to census data. They analysed data from the Office for National Statistics Longitudinal Study (ONS LS), which allowed them to link records of individuals and their parents from the five decennial censuses between 1971 and 2011. Using census data represents a step further in social mobility research by providing much bigger samples. There is some potential for comparative social mobility research using harmonised international census data.
While using the census data Buscha and Strugis are still limited to observing class and status mobility, administrative data can also provide an opportunity to observe income mobility. Income mobility is studied in the Scandinavian countries where administrative data is easily accessible. Other countries need to be more creative as such data sources are not readily available.Raj Chetty and colleagues have done impressive work using administrative data for studying social mobility in the US.
My time as a Visiting Postdoctoral Fellow at the Stanford Center of Poverty and Inequality (CPI) in Spring 2017 was highly illuminating in learning about administrative data sources for social science research and policy evaluation. The CPI is actively involved in projects to improve the data infrastructure for monitoring poverty, inequality, and mobility in the US, and they do this by exploiting administrative and other forms of “big data” more actively than previously; an inspiration for exploring similar opportunities in the UK and in other countries. However, we need to be aware that there are some problems to using administrative data. The fact that different data sources can give different answers to the same question is illustrated recently by Stephen Jenkins who investigates whether income inequality in the UK has increased using different data sources. When it comes to social mobility research, we need to be inventive in using both survey and administrative data.