WHAT DRIVES CHILDREN’S ACHIEVEMENT?

Extensive research shows that children’s early test results are a strong predictor of a variety of outcomes later in life: the high achievers are more likely to have higher earnings and better health; and less likely to have out-of-wedlock births, be on welfare or participate in crime. But what are the key determinants of children’s ‘cognitive achievement’? New research by Professors Petra Todd and Kenneth Wolpin, published in the latest issue of the Economic Journal, explores the difficulties involved in assessing the relative importance of home, school and inherited abilities.

Children’s achievement depends on parental and school behaviours as well as on inherited abilities. For example, it is likely that a child’s score on a reading test depends on how much time parents read at home with their child as well as on the quality of a child’s teacher at school. Child development is a cumulative process, so at any age, achievement will depend on the entire history of time and resource investments made in the child at home or in school.

Researchers across many disciplines are interested in studying the determinants of cognitive achievement in the hope of gaining a better understanding of how different kinds of policy interventions can influence achievement. A large literature in economics, sociology, psychology and education seeks to establish relationships between home and school investments and test score outcomes. Unfortunately, many questions are still unresolved, because researchers often reach different conclusions even when using the same datasets.

A major challenge to estimating the determinants of achievement is that the data available to researchers are often deficient. For example, data are typically available either on the home environment or the school environment, but only rarely on both. Additionally, historical data on investments in the child at younger ages are often unavailable and direct data on inherited ability are never available. Researchers make different kinds of assumptions in an attempt to overcome these data deficiencies.

Todd and Wolpin’s research develops new approaches to analysing the relationship between test scores and home and school investments in the presence of data limitations. Their study proposes a unifying framework for thinking about the cumulative nature of the cognitive achievement process. Within this framework, they consider different kinds of data problems that researchers commonly encounter and propose ways of overcoming these problems.

For example, a common approach to dealing with the problem of missing historical data is to assume that a child’s test score depends on home and school investments in the current year and on the child’s test score from the previous year (the so-called value-added model). That is, given the test score from the previous year, it is assumed that only the current year’s investments in the child matter in determining the gain in test scores. Todd and Wolpin criticise this approach as relying on strong assumptions about the relationship between observed and unobserved determinants of achievement.

Another approach to dealing with data limitations is to include in the analysis all the characteristics of school and home that are available. For example, a study might include parental income or race as determinants of child test scores, even though these variables are poor proxies for the true parental behaviours that actually influence test scores, such as time spent teaching the child.

Todd and Wolpin discuss the dangers of adopting a kitchen sink approach to analysing the determinants of cognitive achievement and they argue for a more interpretable and parsimonious approach. More generally, they discuss the advantages and disadvantages of various approaches and offer explanations for why different studies reach different conclusions.

Todd and Wolpin also consider the value of social experiments as a way of learning about the determinants of cognitive achievement. For example, the STAR experiment in the United States randomly assigned young children to small and large classes. Researchers used the experiment to measure the effect of class size on test scores.

Todd and Wolpin show that the experiment does not uncover only the effect of changing class size. For example, parents whose children were assigned to larger classes may decide to spend more time at home studying with their children to compensate for the reduced attention provided to children in larger classes. In that case, the social experiment uncovers both the effect of children attending a smaller class as well as the effect of changes in parental behaviour towards their children in response to the change in the school environment.

ENDS

Notes for Editors: ‘On the Specification and Estimation of the Production Function for Cognitive Achievement’ by Petra Todd and Kenneth Wolpin is published in the February 2003 issue of the Economic Journal. The authors are at the University of Pennsylvania.

For Further Information: contact Petra Todd via email: petra@athena.sas.upenn.edu; or RES Media Consultant Romesh Vaitilingam on 0117-983-9770 or 07768-661095 (email: romesh@compuserve.com).