Virtually Autumn: Cheat-resistant exams

Now that Assessment week is half-over, it must finally be time to address how to prevent cheating on homework, exercises, quizzes, and exams for online instruction (and even for in-person instruction).

In my mind, the solution is straightforward. Like many other cases, this solution is not free, and I firmly believe that the answer depends entirely on teacher philosophy, which each of us is free to change. So, I invite you to consider whether now is a good time to make some purposeful change in how you view assessing student work and how to address cheating.

Why resist cheating?

Many teachers view cheating as a problem and as something to avoid and to prevent. Well, I don't, and here are a couple of reasons why. First, I think it is important to consciously admit that this war is not winnable. Cheating is an arms race, and teachers are outnumbered and outmotivated. We can spend inordinate amounts of time trying to imagine (and defend against) every possible way a student might circumvent exam protections, but we won't think of them all.

Ways to address cheating

Remove incentives to cheat

Realizing this, we might just throw up our hands and throw in the towel. Or, we can also examine cheating from the student perspective by asking "What motivates cheating?" If there are actions we can take to reduce the motivation, then that would help our cause. When earning points leads directly to a letter grade, that provides incentive for a student to do anything they can to earn as many points as they need to earn they grade they desire. So, what if there was more than one way to earn points? It is not like there is a finite number of points a teacher can dole out, so why build an instructional economy that acts like points are limiting? Others make this point better than I will, so I'll summarize: providing students opportunities to revise work to improve it (and then to earn more points they might initially have "missed") is a great approach to reduce the incentive to cheat.

Also, we have the option to remove points entirely from the assessment equation. Moving to standards-based grading is a great approach. For example, consider a class where students are provided with learning objectives (e.g. compose a waltz) with rubrics that describe the parameters (e.g. the time signature, the types of instruments that should be included, the length of the piece). Instead of providing a continuum of possible numbers of points to earn for each parameter, the students have to achieve all three parameters to have met that standard. There can be multiple standards for a class, and the instructor can decide whether all standards (thresholds), or perhaps only a specific number of them, must be passed to pass the class.

Encourage cheating

If you really want to see your students' heads spin, then you'll consider designing assessments with cheating in mind - by which I mean: encourage cheating collaboration as part of the assessment! Of course, we call collaboration "cheating" in the classroom, but in "the real world," collaboration skills are sought by many employers. Maybe we should be helping students develop and hone those skills, and maybe it would even be useful for us to be assessing those skills themselves? Specifically, I'm thinking about group exams, where teams work together. Yes, it can be more difficult to assess the individual work of team members, and there are ways to do this, like peer evaluations.

Likewise, I now run all of my assessments as open-note/book/internet. I do ask students not to collaborate with each other on certain items, and I also ask students to be sure to cite their sources if they use a resource to assist with the completion of the assessment. In other words, I work to destigmatize using all of the resources at one's disposal to address a challenge. All that is required to effectively implement cheat-resistant exams (and other assessments) is to design those assessments to be more difficult!

Make cheating not worthwhile

Are you worried that students taking your online exam can Google the answer to a question? Well, then don't ask that question. Or, if you do, make it worth so few points that the student will spend more time than it is worth looking up the answer. Instead, design more high-level Bloom's taxonomy type questions (like prompts to "compare," "defend," "design," "create," etc.) I've previously written an extensive post on this idea that you can read here. To summarize the key ideas of this approach:

• make sure assessments throughout the course combine the full spectrum of question styles from easy (like multiple-choice) through difficult (like essay questions)
• based on your letter grading scale (e.g. >90% is an A), assign points to questions so that those top 10% of points students need to earn to get an A are difficult to answer - questions you don't think students who, in your mind, are B students, should be able to answer

Design assessments to require unique student responses

The reason this approach works to produce cheat-resistant assessments is that the high-level Bloom's taxonomy questions require unique student output. Even if students worked together to outline an essay, for example, no two students would turn in the exact same wording.

Yes, these question types take longer to grade…and if you're that concerned about student cheating, then you should be willing to put in the extra effort, right? There's no such thing as a free meal. Consider that cheating and time grading are opposite gradients of the same spectrum. Do you want an exam that grades itself? That is a highly cheatable exam. Do you want a cheat-proof exam? You'll spend a lot of time grading student work. The great thing about this system is that each teacher can decide where on the spectrum they want to be. To be practical:

1. start with the idea of how much time you're willing to invest in grading student work, and
2. then design the exam to meet that need, and
3. then be comfortable in knowing that you've designed an exam that is as cheat-resistant as you think it should be

I've been using this approach in many of my college classes for multiple semesters, and I'm happy to let you know that grade distributions didn't change much when I started giving open-internet exams that contained more difficult questions. In other words, for those of you that think more students will get higher grades in open-internet exams, that's definitely not the case. Likewise, the more difficult questions didn't cause the grade distribution to drop. Students quickly realized that I wasn't putting emphasis on rote memorization (T/F, multiple-choice, fill in the blank questions that are easily cheatable), and that instead they should focus on studying those higher-level concepts.

I readily admit that these approaches will not all work in every discipline, especially when memorization and recall is actually important to measure. In particular, I feel for my colleagues who teach foreign language, because it is incredibly difficult to test student vocabulary knowledge and their ability to read, write and translate a foreign language when that student has access to a web browser.

However, for many, using some of the above ideas will help make assessments more cheat-resistant, and you can decide whether to use positive incentives (e.g. encourage collaboration, allow revisions) or negative ones (e.g. more point weight on difficult questions, fewer Google-able questions), or a combination, to achieve that goal.