Public debate around issues of privacy and security, surveillance and oppression are on the rise as we become aware of the algorithms and big data that determine our lives. Fear of a future Orwellian control society dictated by algorithms and massive databases managed by shadowy public-private conglomerates is not a sci-fi specter but an emerging reality. A glimpse of this future is already available in the public education system, crystallized in the practice of standardized testing.
Standardized testing is a classic socio-technical system: an imperfect technology that has been subject to an ongoing process of attempted improvements and redesigns, complicated by its entwinements with both public and private organizations and its prominent status in the daily lives of millions of Americans. Standardized testing is a slow-moving, non-shiny, and still only semi-digital technology, but nonetheless it provides a powerful metaphor for understanding our likely futures in a big-data driven society.
Unsurprisingly, the practice of standardized testing has its origins in the military. During World War I, the American Psychological Association assisted in the creation of a test for military personnel selection . The Army Alpha and Beta test— administered to nearly 2 million recruits by 1918— propelled the development of the standardized multiple choice test which could be efficiently taken and graded in massive numbers.
As with many technological innovations developed during times of war, standardized testing quickly became popular in civilian life. Throughout the 1920s this form of testing was adopted in the workplace, high schools, and college admissions. Just as the army utilized these tests to help sort recruits into various levels of service, in an educational setting they helped schools to sort students, either by placing them into “tracks” or ability groups or to determine admissions into college .
Since the 1920s, standardized tests have become an essential part of the American educational system. In addition to their usefulness as a sorting tool, standardized tests have been the lever powering multiple massive reform movements within the American educational system. Haney notes that these two dominant uses of standardized testing- reform and sorting – seem to correspond to different philosophical values about the role of education:
“When it comes to public schooling at the elementary level (and perhaps to a slightly lesser extent at the secondary level), there seems to be a broad commitment to educating all students to their full educational potential. But at the postsecondary level, the sorting and selection function of schooling comes to the forefront; indeed it is reflected in the hierarchy apparent among American institutions of higher education.”
Standardized tests have proven fairly straight-forward as regards sorting and selection activities, but their role in actually helping to educate students has been much less clear. While sorting tools like the SAT and ACT have remained largely unchanged, elementary and secondary schools have seen the tests used for reform and accountability reinvented again and again as we struggle to determine the best way to “educate all students to their full educational potential.”
These uses of data– the fairly simplistic sorting and categorization of individuals into existing societal structures and the much more complex uses of data to improve our institutions and our society — is a division that applies across the emerging realm of big data.
One oft-cited instance of data-driven sorting practices are the restrictions on insurance placed on individuals deemed in one way or another to be “high-risk”- based on traditional demographics, warehouses of activity tracker data, and or even genome sequencing. As with college admissions processes, there are good reasons for this kind of sorting in the insurance world, but it is certainly a slippery slope that tends to unjustly deny opportunities and even basic rights to individuals who have little or no control over the data points that determine their lives. The subtle encroachment of highly rational and data-driven forms of discrimination is likely to be a hallmark of the coming decades.
The other primary use of big data, for assessment and reform, is not as explicitly discriminatory but contains its own share of danger. The ability to shine a spotlight on previously shadowy phenomena, to increase accountability around social phenomena big and small, is one of the most touted features of big data. But we can see in the case of standardized testing and many others that data-driven attempts to improve an institution or behaviors often have their own fair share of blind spots and inequities. This can be seen in how quickly these practices lead to a misguided form of micromanagement. Recently, there has been a shift towards “high stakes” testing ties that test scores directly to an elaborate system of reward and punishment. High-stakes testing has radically changed the nature of accountability in education and given the federal government a much stronger form of “informational power” over previously distributed and independent state educational systems. Unfortunately, this apparently rigorous data-driven accountability system seems to do a better job of terrorizing our educators than increasing the quality of our education system.
This shift towards high-stakes testing is remarkably similar to the kind of data-based accountability that allows Uber to manage its drivers remotely while escaping the legal responsibilities of actually being an employer. While increased accountability and transparency is an important goal for both public and private industries, we cannot base these systems entirely on simplistic data points. On-the-ground systems of accountability based on the much richer data set of lived experience, while extremely messy and difficult to manage at a high level, are also the systems most capable of understanding and addressing for the complexities of human life.
Standardized testing represents a deeply-entrenched form of widespread data collection within the United States. Looking closely, we can see that this data is often used for deeply discriminatory “sorting” practices. While we cannot say that all sorting practices are de facto unjust, it is clear that they often unintentionally reify and re-enact structural injustices within our society.
There is already a robust dialogue in place about the dangers of data-based sorting practices. However, the use of data to assess and improve our institutions, while well-intentioned, results in misguided and excessively top-down structures of control and micromanagement. Within education, which has ironically applied these reforms partly in an attempt to reduce structural inequalities, falls into a misguided system of basing rewards and punishments for schools and teachers on very narrow data points that do not take into account the context and underlying problems, complicating but not addressing extremely complex socioeconomic and systemic cultural issues.
As the data points of our lives are increasingly collected, analyzed and monitored by public and private institutions, we are likely to see these same problems of unjust sorting and discrimination practices and misguided narrow-vision reform repeat themselves again and again. And while the US education system is a relatively decentralized and slow-moving behemoth, rest assured that private companies (themselves driven by the data points they present to investors and board members on a daily basis) will be much quicker to implement these systems.
 Haney, W. (1981). Validity, vaudeville, and values: A short history of social concerns over standardized testing. American Psychologist, 36(10), 1021.
 Haney, W. (1984). Testing reasoning and reasoning about testing. Review of Educational Research, 597-654
 Anagnostopoulos, D. (2013). The infrastructure of accountability: Data use and the transformation of American education. Cambridge, Mass.: Harvard Education Press.