So Stuyvesant High turns out to have a cheating problem–or perhaps all selective high schools do? If the high schools do, I’m sure the colleges and universities that receive their graduates do as well. And so in turn do the workplaces that hire the graduates fed to them by this system. Not a startling conclusion if you’ve read Christopher Hayes’ Twilight of the Elites. The NYT article suggests that skilled, systematic cheaters often rationalize their behavior by arguing either that everyone does it (which Hayes would argue is a structural inevitability in social hierarchies that justify stratification via meritocratic distinction) or that cheating is the only way to temporarily distinguish oneself amid uniform excellence, and when you’re done with the test, the class, the moment, you will have earned your place in a college or a job and can prove your genuine merit. As Hayes notes, that moment never comes, the cheater is never at rest, at home, able to show their true quality independent of silly tests and bullshit obstacles. The whole of life becomes a bullshit obstacle, and the search for the edge, the advantage, the trick becomes perpetual. Which doesn’t just hollow out the person, it contributes to the entire socioeconomic system dropping into an ever-accelerating pursuit of short-term gain at the cost of long-term sustainability.
The first problem with narrowly setting out to foil cheaters is that if students or employees no longer believe that tests measure anything important, simple anti-cheating techniques become another petty annoyance–particularly if they think that the testers or bosses are using tests as a crude rationing device or screening mechanism, a way to avoid grappling with difficult or nuanced evaluations. Simple tricks are equally simply defeated, and each one of them just increases the sense that testing is a sadistic and cynical exercise.
Most selective institutions, whether K-12 or higher education, promise highly individualized instruction that adapts to the learning styles, aspirations and personal distinctiveness of every pupil. But the time required to make those promises real when it comes to assessment of students is often badly short-changed in preference to covering as much content as possible in order to enable students to move on to the next subject, the next part of a sequence, the next big lump of content to force-feed the students. What makes many systems of cheating effective is the combination of highly standardized content plus some form of standardized test or assessment. The easiest ways to impede cheating turn on the delivery of distinctive, personalized instruction, which has its own pedagogical justification quite aside from making it harder to cheat. Note, of course, that this strategy is very difficult to adopt in an environment where legislators keep upping the ante on the use of dumb standardized testing for evaluating teachers: the consequence is less and less effective teaching combined with a massive increase in both individual and collective cheating.
Consider the following ideas instead:
1) Don’t use standardized textbooks or teach to the “common denominator” of knowledge about a particular subject or discipline. Build a class intended to teach standard knowledge around a distinctive case, situation, or application of that knowledge, and change the situation or case each time the class is taught. One of my Swarthmore colleagues has taught a fantastic course of this kind in statistics that has students learning basic statistics through delivering statistical studies to local non-profit or community groups based on what those groups would like to know. My colleagues in Biology do a lot of customization on their introductory sequence. This is not just an approach for higher education: it could absolutely be adopted widely in K-12 education. The catch is that this takes constant work for teachers, it takes teachers who are confident in their own understanding of the subject matter and can adapt it to changing circumstances and cases, and it absolutely cannot happen in an environment where highly standardized testing is frequently imposed from above. If your class isn’t like any other class on the subject (even when it’s covering some shared or common knowledge), and your essay questions and problem sets aren’t like any others, and you don’t reuse them, it’s going to be pretty hard to cheat beyond the most local scale (a single class in a single semester), and maybe not even then.
2) Assess individual students on a continuous, spontaneous basis, and weight grades heavily on their demonstrated ability to recall, apply and repurpose what they are learning. Every teacher knows that a formal, scheduled test is really a fairly poor way to understand what someone really knows, and how much they are able to use that knowledge. If you wanted to know whether someone was a good driver, for example, what would you rather have? Five weeks worth of surreptitious webcam recordings of them behind the wheel or a multiple-choice test asking them about the laws and formal rules governing driving? How many times have you seen a person who can pass any formal test and yet can’t use anything they supposedly know? Or inversely, a person who doesn’t do that well on formal tests but who can make very effective use of what the test is trying to measure? I see both fairly often. Tests, multiple-choice or otherwise, aren’t how we use what we know in any other real-life context. We use standard tests because ongoing, constant assessment seems too labor-intensive, and because we believe that you can’t use knowledge until you’ve consumed a sufficiently large baseline amount of information and acquired a sufficiently large baseline of skill. The former might be true in large, poorly financed educational institutions, but it shouldn’t be true at wealthy, selective schools. The latter I think is demonstrably not true: you can put into practice what you’re learning from the very first moment that you are learning it, and be assessed continuously based on how well you apply what you know and how much you improve in your application over time.
3) If these two approaches seem impractical, even a standardized-testing approach can work better with a modest amount of individuation. Let’s say you’re teaching high school biology in a standardized curriculum and you have a final exam that’s a mix of multiple choice and problem sets of some kind. Have all the teachers in that system work together to generate a test bank of 2,000 questions. Change it each year–toss out 200 or 300 of those questions and add 200 or 300 replacements. Have every single test for every single student be randomly generated from the test bank. Every student gets a test with their name on it at the top that is not the same as any other student’s test. Match each student’s test to each individually generated answer key. That surely kills off some common cheating techniques (looking at another student’s test, photographing a test and sharing it with others, memorizing the same identification question, etc.) If you object by saying, “Well, what if the students get a hold of all 2,000 questions in the test bank and memorize or prepare for all of them?”, I say, “That would be mission accomplished, then: what’s the difference between a student who has memorized 2,000 questions worth of content you’re going to test them for and a student who just happens to know all that content well enough to answer any questions from that 2,000 you might randomly choose to ask of them?” You can’t exactly scribble the answers to 2,000 questions on your body, and if you don’t know which of thirty or fifty or one hundred variant versions of the test you’re going to get, then the only way you can prepare is to actually learn the content.