Beyond Atlanta: Fixing Corruption in High Stakes Standardized Testing

Although corruption in educational systems is viewed as a pervasive problem in developing countries, wealthy countries have had their fair share of educational corruption as well. In the United States, for example, the harsh prison sentences in the recent cheating scandal in the Atlanta school system cheating received extensive news coverage this past spring. While it’s true that what happened in Atlanta was particularly wide-spread, involving 44 separate schools (and dozens of principals and hundreds of teachers), this is hardly the first time a significant teacher or administration-driven cheating scandal has come to light. In the last few years, teachers and principals have been caught cheating all over the United States: twenty teachers in Houston were removed from the classroom for cheating on elementary school tests while in Philadelphia, there was a multi-year investigation that involved 138 educators in 27 schools. The National Center for Fair and Open Testing claims that in the last five years, there have been reports of standardized exam cheating in 37 states and the District of Columbia. They have catalogued over fifty different ways that educators helped their students cheat. And although anticorruption efforts in education frequently revolve around the exchange of money or sometimes sex for grades, the sort of cheating involved in these scandals is also a form of educational corruption. Alteration of student tests, even when the students themselves may not benefit from it, is a perversion of the system and a way for teachers and principals to put themselves in line for undeserved awards, given that many educational systems in the U.S. now operate under a high-stakes testing regime in which student performance has a significant impact on teacher and principal evaluations.

Indeed, it is clear that we are not dealing with a few corrupt “bad apples,” but rather with a widespread pattern of teacher and principal corruption. A significant contributor to this problem is the high-stakes testing system described above, which gives teachers and principals to manipulate test results for their own material benefit. Fortunately, there are a few fairly simple steps that would eliminate most of the opportunities for this sort of corruption:

  1. Find disinterested parties to administer the tests. Having teachers administer the test to their own students is a clear conflict of interest. The referee should never have such a strong interest in one side. There are many things that the test administrator can do to give students an illegal boost. The National Center for Fair and Open Testing has found incidents where teachers have posted formulas or other hints on the whiteboard, where they have told students the correct answers, where teachers have ignored students copying from each other, and where they have allowed extra time, among many other cheating tactics. Having someone without any material stake in how the students in the classroom do would eliminate the conflict of interest that the teacher faces and reduce opportunities for corruption.
  2. Use software to detect patterns of cheating. Certain types of cheating – like teachers changing their students’ answers after the test is over – leave telltale signs like a significant number of erased incorrect answers. Other types of cheating are harder to detect, but will produce patterns such as dramatic swings in particular student performance from year to year. While some districts already analyze the testing data with software, others should follow suit. This is yet another instance where big data can sniff out corruption.
  3. Secure the tests immediately after administration Since several of the scandals involved altering students’ test answers after they were turned in, if tests were scanned to a central database within an hour or two of submission it would greatly cut down on the opportunity to change them.
  4. Require make-up tests. Some schools have been caught suspending poorly performing students right before testing day in an attempt to remove them from the testing pool. This type of cheating is particularly toxic as it is punitive to individual children who do not perform as well. There are two ways to deal with this. First, suspensions could themselves be suspended on testing days. Second, there could be a make-up testing day for those who were sick or suspended on the first testing day.

Many commentators have decried the extremely stiff sentences in the Atlanta case, and others have argued that it is an inevitable byproduct on a system that puts toxic pressure on teachers to perform unattainable miracles. Whether high-stakes testing makes sense is a debate for a different forum. But as long as it is in place, it is crucial to make sure that the results are not corrupt. Corruption in education is particularly harmful: as the discussions we’ve had on this blog have emphasized, education is crucial in training students for a lifelong belief that corruption is wrong. When their teachers cheat on their behalf, students learn that cheating is the way to get what you want. As the chair of the Atlanta School Board said when the cheating scandal came to light, the corrupt cheating is “absolutely devastating… It’s our children. You just don’t cheat children.”

1 thought on “Beyond Atlanta: Fixing Corruption in High Stakes Standardized Testing

  1. Very nice post, and all your suggestions seem sensible.

    I don’t have much to add, but I did want to point out a strong thematic similarity between at least one aspect of your post and Richard Bistrong’s guest commentary from a little while back about how international salespeople’s incentive-laden compensation packages contribute to corruption (

    In both that context and the educational testing context you discuss, we of course can and should use monitoring and control systems, as well as the threat of ex post sanctions, to deter, detect, and punish wrongdoing. But in both contexts, we’ve set up a system where certain agents (salespeople, teachers) are under extraordinary pressure to produce significant (arguably unrealistic) results in the short term, without adequate recognition that the ultimate goals (learning improvements in disadvantaged communities, sales growth in emerging markets) take a long time to achieve, are affected by numerous factors beyond the agents’ immediate control, etc. At the same time, in both contexts the principals (governments, senior managers) have reasonable concerns about “shirking,” and an understandable desire to both measure performance quantitatively and give the agents strong incentives to work hard (and to weed out poor performers).

    I suspect that a lot of corruption challenges present a similar structural problem.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.