Head of the department said statistics exams must be done without software, otherwise it's cheating. What can I do as a lecturer? [closed]

Question

I am a statistics lecturer at a university. For midterms and short exams, I have students use software such as StatCrunch or Excel for computational questions, but not theoretical questions.

The head of department, from a physics background, said to me that using such software to answer computational questions is just a type of cheating to get marks. She said any statistics exams must be conducted in the traditional way. Is what she said correct?

NB: I am the instructor and required my students to use software as we are in 2021 and no one needs to compute it manually as they are finance and economics students.

Clarifications:

• The chair's concern seems to be that if students are allowed to use software, the course is too easy and the grades will be too high. This is obviously false, since there is much more to the course than just computing numbers (students have to understand the theory).
• The chair does have a history of being very controlling and abusing or firing instructors who don't submit.
• Traditionally, instructors in this course have mentioned that statistical software exists, but have not necessarily required that students use the software extensively.
• I do not believe it is just a miscommunication; she clearly said that "students using software is an attempt to cheat and get better marks" (even if I allow them to use software)

Answer 1

There are two issues here: the pragmatic and the pedagogical. On the pedagogical front I think your head of department is just wrong. Being able to do the arithmetical computations in statistics is useful only to the very limited extent that it helps with intuition. The American Statistical Association agrees

Modern data analysis involves the use of statistical software to store and analyze (potentially large) datasets. While there may be value to performing some calculations by hand, it is unrealistic to analyze data without the aid of software for all but the smallest datasets. At a minimum, students should interpret output from software. Ideally, students should be given numerous opportunities to analyze data with the best available technology (preferably, statistical software).

From a pragmatic point of view, then, what can you do? Nothing, probably, for a course in progress. In the longer term, you might try to get a department colloquium talk from a statistics education researcher -- easier now that remote talks are standard -- to get some discussion. The American Stats Association documents might help to provide some external authority, and I think the Royal Statistical Society may have something similar

Answer 2

University policies need to be followed and, in most cases, specific instructions from the head need to be given serious consideration. If you are tenured faculty then it might be reasonable to oppose the head on something like classroom/exam management, but for low level faculty it might not be wise - or even possible.

But that leaves the philosophical question and for that there can be a difference of opinion. My own would be that in the earliest courses, when data sets are mostly small and students are learning principles, manual calculations are probably valuable. Later, when doing serious things with serious data, tools might reasonably be employed. After all, those tools weren't developed for the simplest of cases.

So, doing manual calculations at the start might help the student gain insight, whereas a tool might just be a magic box. Students not only need to be able to get answers, they need to be able to make quick judgements as to the likely validity of the answers so that tools aren't misused accidentally or by design.

Of course, your opinion might be different, but you need to be able to make some reasonable argument (to the head) to justify it.

Answer 3

It sounds to me like the comments from the head of your department were made in some sort of informal chat.

The question of allowing use of computational aides beyond a pocket calculator is an significant pedagogical decision. It is therefore important that:

1. This matter be discussed seriously at least at the departmental level, if not the university level, rather than decided semi-off-hand by each instructor of each course. (And that's true even if the instructors have the autonomy to make their own decision.)
2. A clear policy be set, at whatever level the decision is taken, on this matter: What instructors will use when teaching; what students will be encouraged to use in class, in homework and in exams; and what students will be allowed to use in class, homework and exams.
3. The matter be reflected to students consistently, at the very least throughout a semesterial course.

I therefore suggest that you:

1. Check whether discussions have been held on this question (and locate the minutes/summary/recommendations).
2. Check whether official policy has been set on this matter (and whether it's binding).
3. Consult instructors of statistics in your department in previous semesters, and other departments currently, about the practice in their courses (not necessarily to copy it).
4. Assuming this matter has not been given thorough enough treatment in the past - schedule a meeting with relevant members of faculty and the head of the department, or bring it up in a department council session or another appropriate forum, to start such process.

While you're doing this, I also suggest that you get back to the head of the department, explain that you believe that this issue merits more thorough treatment - flattering them for bringing up this point, so that the main impression is not that you're brushing them off, but rather that you've taken their words to heart. Emphasize the need for a consistent policy vis-a-vis the students. Assuming the department head agrees with your general outlook, that will set the background well for you to argue that it would be inappropriate vis-a-vis the students to change course policy mid-course, and that based on the more thorough treatment of the matter, you will reconsider the policy before next semester.

Answer 4

Cheating is behavior that violates the rules on what is acceptable to use during the exam. If you didn’t make a rule that using software is forbidden, students who use such software cannot be considered to be cheating.

It sounds like your department head and you have a disagreement on whether there should be such a rule, that’s related to a conceptual disagreement over what skills students should be tested on. Whether she can dictate to you whether you should have such a rule depends on the policies of your institution. In my institution and all others I’m familiar with in western countries, the decision on what aids students are allowed to use during an exam lies with the instructor as part of the instructor’s academic freedom. A department could make recommendations to instructors about what sorts of policies are a good idea to implement, but individual instructors could decide to do things differently if they thought that was pesagogically appropriate.

Finally, even if your department head doesn’t have the official power to force your hand on this issue, there are situations in which it might be a good idea to do what she suggests anyway.

Answer 5

I think it is high time to sit down and discuss the intentions and options.

What the head sees is a test, like any other before and later, and students are allowed to use somewhat blackboxed tool that reduce efforrt needed significantly.

What you see is a new technology that allows all to focus on the core problem and leave the number crunching for a machine.

It is very same as times where forst calculators became available...

If you left the tests and evaluations same as without the aid, the head is right; the resulting marks will be increased without increased level of student performance.

On the other hand, if your tests and marks are adjusted to this aid, then you are "right" and should be able to convince the head that there is no tech-created bias.

I can remember one of my toughest exams. We were allowed to use any aid we could bring in the room, except for laptops and phones. Quite rich help, aint'it? The questions were formulated the way nobody without knowledge would be able to find the answers in the papers.

And my most favourite teacher announced that any student can go and use Wolfram Alpha for their calculus exam, and distant students were allowed to use any help. It was a trap; without the knowledge, Wolfram's answers were useless and "smart cheaters" all left soon after because the knowledge they didn't learn led them to fail in all other subjects...

Back to your problem. Try to convince her the number crunching is not the critical skill for the course and that the bigger picture is. Meaning, as the students have tools for simplify their work you have made it harder elsewhere.

But, by the way, sometimes the old number crunching is a good exercise and forces people to go down and look "under the hood".

Answer 6

If in the end you need to go with a non-computer-based exam, here's a handful of suggestions which hopefully get at the same pedagogical points:

1. Set up multiple choice questions where the students have to determine which distribution or function is most appropriate.
2. Make hypothetical spreadsheet data with an error to find and correct. For example, one column labeled as (x-mu), the next labeled as (x-mu)^2, but with some cells clearly not squared. Or a sum that doesn't contain all of the terms (see e.g. Reinhart-Rogoff).
3. More theoretically, write subtly incorrect formulas and have them correct the errors. For example, neglecting a square root.
4. Looking at a data set and estimating the standard deviation/variance or mean or linear coefficient.
5. Solve the results yourself (using computer software), and have students interpret that data.

All of these are meant to highlight points that a finance/economics student would come across in their own use of statistics.

Answer 7

I have upvoted the most popular answers as they cover all the ground very well. As a human being who spent decades as a leader in academia, I can read between the lines. This chair is not up to the job and therefore being overly controlling, unreasonable, and outright horrible at times. You must keep your nose down until this person is no longer chair. If that is more than you can bear for the years to come, look for a transfer or a position elsewhere.

Answer 8

Here are my two cents on avoiding the problem rather than trying to actually address it, which could make matters worst - as others have noted above.

Perhaps one way to work under these guidelines is short answer questions on using the software and testing whether the task is possible or not, without (excel) add-ons. Worst case there is the option of pseudocode on paper.

For example, asking whether linear regression is possible in R, and what should be the input and output be, in terms of the dimensions and data structure.

Answer 9

I'm old enough to remember taking many quantitative exams where only pencils carried in were permitted and sometimes a slide rule. Typically, grading made allowances for these limitations (e.g., lower precision required) and were usually thoughtfully (fiendishly it often felt) designed to probe conceptual understanding. Being a little dyslexic, I often foundered on "dumb math errors." Many extra point problems of that era could probably be solved with a few keystrokes on a cheap present-day calculator or a smart watch. So, if that's one's frame of reference, using such a device would have been ipso facto cheating.

I'd guess the "cheating" part of the dictum is at least partly motivated by a desire to avoid evaluating superficial skills in the use of portable computing devices and surreptitious cribbing via the internet, and instead evaluate conceptual fluency in essential subject matter. As @Dave notes, questions can be formulated that are independent of aids, which could support exams given mit or mit out.

As others have noted, ex post facto changes to rules (by any party) are certainly questionable.

Answer 10

You could just disallow software, but take off no points for computational errors.