Data Class — Part I

Data in Education: The Oil of the 21st Century

Learn More’s Data Class is a series of monthly articles, devoted to the bridging of data and education. Getting things started is Nadir Zanini, as he argues the power of bringing together these two fields.

“Without data you’re just a person with an opinion”. So said William E. Deming (1900–1993), a pioneer of using data to make improvements in work and daily life.

Data constitutes the objective and robust base of evidence upon which transparent and credible decisions can be made. This is why businesses, governments and other organisations are investing so much time and money into leveraging data in market intelligence, healthcare, manufacturing, defence and security.

Nowhere is data more important than in education.

Education can be thought of as an efficient way of acquiring knowledge and skills. In modern democracies, education is the key to providing equality of opportunity. If all students were offered the same quality of learning — irrespective of their cultural and socio-economic background — education would enable better job prospects for all, thus promoting social mobility and enhancing social inclusion. With this goal partly in mind, governments are increasing investment in education: across OECD countries, more than 10% of total public expenditure is on education. This figure is growing. This type of budget decision is drastic, and cannot be taken without the support of data-driven evidence. England’s school accountability system and the What Works Clearinghouse initiative are just two examples of how data can be used to improve school systems and ultimately help to direct public expenditure.

Data is especially relevant in this field because education can be particularly subjective. Everyone has an opinion on how learning should best occur. But these views are biased by our own experiences: the area we lived in, the courses we took, the support and expectations of family and friends, our capabilities and aspirations. Data can help clear things up.

We live in a time when high-quality, objective information on education is available, and in a larger quantity than ever before. Education providers, testing agencies and governments manage a vast amount of data on learners, their educational choices and their outcomes. In addition to this administrative data, we can also look to national and international sample-based surveys (for example OECD’s PISA and IAEA’s TIMSS) and to the in-depth evidence produced by experimental research on small groups of individuals. If this is not enough, the Big Data wave has hit education too. Recent technological advances have not only changed the 21st century classroom and the way learning happens, but also broadened the variety of data available, increased its volume and the velocity with which this data becomes accessible. Real-time written communication, web searches and geographical locations are among these new sources of data.

Data in education can be thought of as oil. With oil, comes power. Likewise, monitoring the educational experience of millions of learners generates the power to improve educational outcomes. Trialling a new ICT tool on a small number of schools allows us to understand if it is effective before adopting it nationwide. Data enables us to target interventions, such as specific teaching practices or financial support, to specific groups of learners. There is power in learning what kind of education our economy and society need. There is power in making education accessible and fair to everyone. There is power in providing the best possible learning experience to everyone.

But power requires responsibility. While the use of data in education can help us to improve the world we live in, its misuse — intentional or unintentional — can damage it. One of the misconceptions is that in order to be informative, data must be summarised into headline figures. Such figures can be wrong or misleading. A captivating account of this is IFS’s Director Paul Johnson’s ‘confessions of a number-cruncher’. In education, when coupled with our subjective perceptions, reductionism translates into even greater risk of misinterpreting the data. As a statistician working in government on educational research and policy advice, my job is to minimise this risk.

In the same way oil is not of use in its raw form, neither is data. In order to exploit the power of data, we need knowledge and skills to analyse and interpret it in as clear and objective a way as possible. Along with data strategies to store, process and share data, statistics, machine learning and artificial intelligence are among the tools with which we mine the data and extract useful information. In education especially, skilled experts are very sorely needed to refine data and produce knowledge from both large-scale studies databases as well as small-scale, in-depth researches.

This change won’t happen if we don’t demand it. All of us must create the expectation for objective and robust evidence-based data in education.

Nadir Zanini
Research Fellow at Ofqual, the regulator of qualifications, exams and assessments in England. The views expressed in this article are those of the author.