How to verify authenticity of submitted assignments

Question

When teaching 250 students in the same subject, knowing each student can be quite difficult. In smaller classes, it may be possible to learn enough through classroom interaction that when assessing the assignments it may be easy to tell if it is their work or not. For example, if a student can never form a coherent argument when asked in class, but that same student submits extremely high quality work, the teacher might want to dig a little deeper to see if the student is just better at writing or if they are hiring a ghost writer to produce their reports for them. However, in large classes, there is too little interaction with each student to form a strong opinion of everyone. Those who can 'hide' in class are the least likely to be 'caught.'

I see one option as incorporating an oral exam into the overall mark and using that mark as a basis from which to judge the student's future work. It's not perfect but it's something. One problem is that with 250 students, oral exams end up consuming so much class time that little time remains for lectures.

Running through something like TurnInIt will only catch if they are taking from an existing work but it will clearly not catch ghost writing.

What are the most effective ways of verifying authenticity of student assignments?

Answer 1

I'm going to give you an answer you may not want to hear: it isn't worth it for a class that big.

The pedagogue in me doesn't like my own answer, but as they say, you have to choose your battles, and I believe there are more important issues when teaching a class that large than to worry about catching the homework cheaters. Mitigation can include reducing the impact that homework has on the overall grade (but this leads to other issues, such as students wondering why homework is necessary if it isn't worth much in the final grade). Another (probably less successful) method is to instill the fear of [insert deity of choice] into your students when it comes to cheating. If you do catch a cheater, make an example of that person that the other students won't forget -- I threaten to fail any cheater instantly if I can prove they cheated in any way, and I say it as matter-of-factly as I can on the first day of class: "Look, if I catch you cheating, I'll fail you, end of story."

If you have a balanced set of assignments that includes in-class exams, then poor performance on exams should produce final grades that indicate that the student hasn't learned the material.

If you can work out the oral exam, that might help balance out any poor knowledge, but beware that some students are absolutely terrified of oral exams, and they might know the knowledge as well as anyone else on paper but not be able to perform out loud.

I suggest not losing sleep over a few cheaters in a class that big, and to ensure that you're giving good assignments that are challenging and provide a good means for learning the material.

Answer 2

I agree with Chris Gregg's answer and want to add some additional thoughts and strategies to it.

Make cheating irrelevant

1) Homework becomes a collaborative venture.

Students can work individually or in groups on all out-of-class assignments. Every student needs to turn in their own copy of the assignment. Every student has to list whom they worked with on the assignment. Homework is a tool for me to give the students to help the students learn the material. Why should I care if they do it in groups or individually? If you encourage students to go to each other for help, then only the most diehard cheaters (whom you cannot stop) will keep going to the faceless masses on the internet. All involved learn more through the interaction. Most "real-world" work is done collaboratively anyway. When I give these kinds of assignments I work very hard to come up with questions and prompts that are not internet searchable - because they must use specific resources, the questions are opinion-based, or the questions focus on specific narrow local events, etc.

• Which of the two figures on the accompanying handout best explains the observation that carbon-carbon single bonds in strain cyclic molecules are weaker than those in unstrained cyclic molecules? Provide a brief rationale.
• Use the data in Table 8.1 on page 115 to determine which of the the following is growing faster: China's population, China's per capita GDP, or China's per capita energy consumption. Do a similar analysis for the US 100 years ago (use the data on the handout). Comment on any similarities.
• What are the three key points of the speech given last week by the Dean of the College? With the information I've given you about the budget of the College, which of the Dean's action priorities for this year are financially feasible? Which action priority are you most interested in seeing completed?

The third one on my list would be very challenging for someone to get outside the institution help on.

If the out-of-class projects are significant, then you can encourage the students to keep each other honest. On big collaborative projects, I have the students rate (not grade) each other on the following questions. Trends appear when someone has not pulled his/her own weight:

• This person contributions to the project were (less than, the same as, greater than) mine.
• This person deserves a grade on the project that is (less than, the same as, greater than) mine.
• This person's greatest contribution to the project was _______.
• The hardest thing about working with this person was _______.

2) As Chris Gregg said - make exams worth much more than homework.

For example:

• Homework 20%
• Midterm Exam 1 20%
• Midterm Exam 2 20%
• Midterm Exam 3 20%
• Final Exam 20%

In this scenario, exams count for 80% of the grade (and so cheating on homework will not have a huge impact), but homework still counts for something. If you phrase your grading descriptions as "homework counts as much as an exam", then you get students to take it seriously. One instructor I know even goes the extra step to rename his homework assignments as "Take-home Exam Part 1, 2, 3, etc." (and then treat them like the collaborative homework I described above. If a student has been cheating on the assignments, then he/she will not do well on the tests, and course grades will follow.

Both of these strategies focus on the purpose of the homework (to guide student learning and studying) which is different from the the purpose of tests (to assess student knowledge, skills, etc.). If homework is a guiding assignment and exams are an assessment assignment, then worry about cheating on exams (easier to spot and control) and ignore cheating on homework. Some students will not be guided the way you want them to be. Just as you cannot make an individual student learn, you also cannot make an individual student learn in the fashion you choose.

Answer 3

I've never seen it done before, but in theory, it could work...

The day you collect the assignment in class, hand out a surprise quiz that is identical to the assignment the students completed at home (or maybe just a couple questions from the assignment to make it faster/easier to compare). Have the students complete the quiz during the class period.

If a student completed the assignment legitimately, they will likely have very similar answers on the quiz. If the student did not complete the original assignment legitimately, they will likely struggle with the quiz and their answers will not even remotely match the answers they "provided" on the assignment.

The key is to look for quizzes whose answers are significantly worse than those provided on the assignment.

Two problems with this method

1. It is kind of insulting to students that didn't cheat. FIX: Don't tell the students until afterwards, but don't count the quiz for normal credit. Instead, just give students who didn't cheat a few extra credit points.
2. It does nothing to catch cheaters that don't show up to class. FIX: Make a big deal out of requiring students to hand the assignment in in-person on the day of the quiz to make sure whoever hands in an assignment takes the quiz.

I admit, this is a pretty wacky idea, but it seems feasible, although a bit elaborate.