Monday, September 21, 2015

Cheat-Proofing Exams I

Today I was as nervous as my students for our first Genetics test of the semester. Today, after two semesters of teaching DISCOVERe (tablet computer-based instruction) courses at California State University, Fresno, I gave my first all-digital exam. I learned a lot (which, as you teachers already suspect, means that a lot went wrong). And, as you might also suspect, because I'm writing about it, there are some best practices to share!

The Background
As a geneticist, providing students with authentic experiences (even in the classroom) is one of my top priorities. One of the reasons I joined DISCOVERe is to ensure that all of my students would have computers in the classroom so that we could all practice analyzing and interpreting data. The first time I taught a genetics tablet class (which was my first tablet course), I hadn't done all of the things I should have done (like providing lots of deliberate exercises using digital workflows) to ensure that everybody was comfortable with taking digital exams. So, that class used almost entirely paper-based exams. There was no talking, no notes, and no tablets allowed during the exams in the tablet class.

The Theory
My goal is to have students demonstrate to me their understanding of the material by engaging in mid-level (and up) Bloom's: interpreting data, making predictions, applying knowledge in new situations, and being creative. I also feel like giving group exams for at least two reasons: 1) they help students collaborate and teach each other, which is extremely valuable, and 2) this also helps me observe students working in group settings (which helps me write letters of recommendation for those who ask!) To facilitate group exams, and to facilitate exams with tablets, I need to develop exams that are cheat-proof - and I also have to change my attitude: what we call "cheating" in the classroom is called "collaboration" in practice. I need to incorporate that into the classroom.

The Concerns
How many colleagues have asked me how I would prevent cheating on exams if I let the students have tablet computers? Plenty. And how did I respond? I told my students on the second day of class this term: I embrace collaboration. We know that employers want to hire people who work well in groups and who have excellent communication skills. The practice of scientific research has been making deliberate moves over the last decade (and more) to foster interdisciplinary collaboration - and I definitely subscribe to Woodrow Wilson's philosophy: "I not only use all the brains that I have, but all that I can borrow." So, how can I have group exams, and incorporate tablets into exams (to access online DNA sequence databases and web-based analysis tools, for example), without facilitating cheating?

The Approach

  1. Individual Exam (80% of points, 15 minutes, no talking; open note/internet/textbook)
  2. Arranging into groups (5 minutes)
  3. Group Exam (20% of points, 25 minutes)
  4. Exit Survey (5 minutes)

Following my attitude adjustment, "It isn't cheating - it is collaboration," what do my tests look like now? We have a two-part exam (like I have seen at, in which students first complete an individual exam. They download a PDF of the exam, annotate it, and return it to me while we're all in the classroom. There is no talking, although it is open-digital-textbook and open-internet (no print materials allowed). As at UBC, I made the individual exam worth most of the test points (~80% in my case). This helps assuage fears about how group performance might impact individual performance. Additionally, the individual exams incorporate most Bloom's levels. The goal of this 15-minute portion of the exam is to distinguish students who have not prepared at all (the D and F students, say) from the other students. Students who have not studied might spend quite a bit of time looking up simple factual information and not earn many points in this relatively quick individual exam. Although the questions are fast to answer (often matching or multiple choice, for example), they do require reflection and application. Although this is an open-digital-resource exam, this phase contains no talking and also no audio. I don't want students distracting each other by playing movies, for example (say, of my lecture captures). I don't even let students bring earphones, because I can't guarantee that every student would have that resource. The individual exam should be first, so that any students who happen to arrive late don't miss out on the group exam time.

After the individual exam, students form their own groups of three or four. I don't choose the groups; I hope that students (if they don't already have study groups) will establish long-lasting relationships in the class by realizing that there will be additional group exams throughout the semester. The group exam is 25 minutes long. Here I differ from the UBC example. I don't give the same exam to the groups. Instead, I write a short exam that is worth ~20% of the total exam points and that builds on the individual exam. For example, I might ask the group to describe why they chose the answer(s) they did on the individual exam, or to perform group brainstorming (one of my favorite questions from today was: come up with three hypotheses to explain why this unexpected experimental result was observed). These higher-level Bloom's questions, making up a small portion of the points, are designed to help me distinguish the A, B, and C students. I adhere to the UBC example by asking each group to submit only one exam per group. So, the first thing each group does, after introductions, is to select one member to be the "scribe," who will write all of the group member names on the exam, annotate it, and submit the PDF to me.

Active learning at its best: during a group exam

After the group exam, I distribute an exit survey in which students rank each others' contributions and state whether they agree with the group consensus answers. This is where those who dissent can explain why. All of this serves a simple goal: to help me collect evidence of understanding the critical concepts.

With my unorthodox scheme of assigning letter grades (0-20% of points = F, 20-40% = D, 40-60% = C, 60-80% = B, 80-100% = A), this exam structure lets the individual student earn up to a B individually. Any additional points on the group section can push any student into the A grade range. Thus, one letter grade of points is reserved for students who are creative and work well in groups and know the material so well that they can apply it in novel circumstances (such as making predictions and formulating hypotheses)

The Execution
Here's where today was a mixed bag. On one hand, I (as always) had a backup plan. It turns out, today was the day when I had to implement the backup plan. And the only reason that we (the class and I) were successful today was because I had specifically had my students practice digital workflows on relatively trivial exercises, every day, in class.

My approach was that I had the individual and group exams prepared as PDFs. They were to be distributed, one copy per student, via Google Classroom. Before class, I had uploaded and saved the PDFs as attachments to draft Assignments. The plan was to press the "Assign" button one minute (or so) before I wanted every student to have access to each exam. I designed the exit survey in Socrative. I also, just in case, posted both PDFs to Blackboard; they were to become visible to students at designated times during the exam.

It turned out that, for whatever reason, although the students generally had no internet connectivity issues (every student in attendance did, actually, successfully submit their exam to me by the end of class), none of my devices (laptop, two iPads) were able to connect to wireless networks on campus.

So, prior to the exam, I was not able to launch the Google Classroom assignments. Even with my thunderbolt-to-ethernet adapter and the ethernet cord provided in the classroom, I still wasn't able to access the web (still working on why!) Apparently, Google Classroom assignments can't be launched by the Google Classroom apps (my iPhone wasn't able to launch the exams either!) So, I was at least able to tell all of the students to download the exam PDFs from Blackboard, which worked smoothly. The main drawback was that the submission of exams was to me, directly, by e-mail attachment of annotated PDFs. That means that I spent about 1.5 hours after the exam downloading attachments and renaming files in a consistent format with student names. With Google Classroom, all of this would have happened on the fly during assignment submission.

Alas, the network let me down today. But, being prepared won the day! Now I need to go grade some exams.

The Bottom Line

  • As always, have a backup plan.
  • As always, be progressive and be willing to try new (to you) things that others have proven can succeed!
  • As always, use practices in the classroom that will help students develop skills and proficiencies that are both universal and also relevant to your discipline!

Until joining DISCOVERe, I never thought I would see the day (today) when I would be able to ask every student in the classroom, during an exam, to analyze a 280-nucleotide DNA sequence by searching online databases to tell me which species and gene that sequence likely comes from! Today, I was transformed - and my classes were transformed forever. And, that DNA sequence came from the tra-2 (transformer 2) gene of Caenorhabditis elegans.

1 comment:

  1. This is a very interesting perspective. After sitting through our college's graduation today, several of us went out and spent a good part of the afternoon discussing how rampant cheating is in our classrooms. We teach history, and it's amazing how innovative the students can be in the ways they find to cheat. Your approach is certainly interesting, and I've forwarded this article to them. Personally, I'm not sure if it's enabling them, or "abling" them to function better in the real world, but it's certainly worth considering! Thank you so much for publishing this!


Have an insightful comment, best practice, or concern to share? Please do!