I realize that we’re approaching the end of the semester and I, like all of you, am very tired. Indeed, I feel like a dehydrated man crawling painfully through a desert lusting after the cool, refreshing waters of yon oasis. Granted, yon oasis is, in my case, a mirage since the end of the semester is always punctuated by a hurricane of grading, but I digress. My point is that I understand and, indeed, sympathize with your exhaustion. All that having been said, there is a small matter that we really need to discuss- a problem that I have noticed several times over the course of the semester. A problem that I have actually explicitly addressed with the class as a whole and with several of the more egregious offenders in person. That problem is simply this:
The words that you use to describe the reasoning process? Yeah- they do not mean what you think that they do.
Let’s start with the word “assume”. To assume does not mean “To make an a@~ out of you and me” as many hackneyed grammar teachers might quip, but rather is to take something for granted, as if it were true, based upon presupposition without the preponderance of facts. In other words, an assumption can be thought of as something we accept as given, even though we can’t prove it, so that we can move our logical reasoning forward. Assumptions are a necessary part of logical reasoning- for example, I assume that I am communicating with someone just now rather than simply rambling in my own demented fever dream*- but nevertheless, we must always keep in mind their status as unproven beliefs.
Next, let’s consider the word “conclude”. A conclusion is a logical consequence of some thing or things that we know to be true. So, for example, if the statement “All frogs are green” were true,** then the statement “Kermit the Frog is green” would have to also be true. In the context of a social science or, indeed, any science, conclusions are generally based on evidence- so, in other words, rather than deriving from formal logical statements, our conclusions reflect the quality and nature of the empirical evidence we have produced. The critical point here is that, unlike an assumption, a conclusion is something we take to be true based upon explicit evidence. Thus, in this case a conclusion might be something like, “Given that we found using a large, nationally-representative data set that female incomes tend to be lower than male even while controlling for a variety of confounding factors, we conclude that gender plays a role in determining compensation.” Note than a conclusion can be erroneous while, yet, remaining a conclusion.
Finally, let’s consider the term “inference”. To “infer” something is to make a guess based on the balance of evidence, but regarding a subject about which you do not have direct evidence. So, in other words, an inference is sort of mid-way between an assumption and a conclusion- it’s supported by some evidence, and therefore isn’t an assumption, but the evidence doesn’t support it directly, and thus it is not a conclusion. An example of an inference is something like, “One of my students was doing consistently poorly and then, before the last test, came to tell me that he needed to take it late because his grandmother had died. Given my knowledge of the Dead Grandmother/Exam Syndrome, I infer that this student is most likely lying to me.” I don’t know that the student is lying to me- my evidence doesn’t speak directly to that point- but I have evidence that leads me to infer that he most likely is.
I bring all this up because you are driving me to madness by insisting upon writing things like, “Based on her evidence Bethany Bryson assumed that people avoid liking the musical styles of lower-status individuals.” And you see that’s just not right- if it’s based on evidence, it’s not an assumption. It might be a conclusion, or it might be an inference- depending on the nature of the evidence and the claim- but it isn’t an assumption. This may seem like a trivial matter, but these terms denote very different parts of the reasoning process and very different levels of certaintly. And, frankly, it distresses me more than a little that college students- even advanced college students- cannot discern the difference between a claim that is taken to be true, despite a lack of evidence, in order to facilitate reasoning and a claim that has strong evidentiary support. I realize that a variety of media outlets seem almost to be conspiring to stamp out good, rigorous thinking,*** but that is no excuse. And frankly, every time I read a paper where someone says “assume” when they mean “conclude,” or “infer” where they mean “assume,” my train of thought spectacularly derails, killing masses of helpless neurons.
Please, please stop.
All the best,
Drek the Uninteresting
* Given the quality of my writing, the fever dream option may be far more parsimonious.
** In point of fact, all frogs are NOT green, but give me a break here.
*** I'll decline to specify whom I am referring to here. Fill in your own preferred villain at your leisure.
As a side note: No students were directly quoted in this post.
7 thoughts on “you don’t even want to think about what they do to contractions”
“…And, frankly, it distresses me more than a little that college students- even advanced college students- cannot discern the difference between a claim that is taken to be true, despite a lack of evidence, in order to facilitate reasoning and a claim that has strong evidentiary support”
You seem to be inferring (or less likely, assuming ;-) that the students don’t realize there is a difference between assumption, conclusion, and inference. An alternative could be that they realize there is a difference between these things, and just do not know the correct term to use in referring to each. I personally think that is often the case. Sloppiness is also a problem, however.
Did that Dead-Grandmother guy really collect data for “over twenty years,” then publish it in the Connecticut Review? I hate to spread it around if it’s just a good joke, but I’d love to spread it around if it’s true.
KMD: You may be right that they don’t know the correct term. Actually, you probably are correct in the sense that they’re aware that each of these things exist. On the other hand, if they’re unable to talk about them in a coherent way, one is forced to wonder how they interpret material that they’re exposed to that involves assumptions, inferences and/or conclusions.
Philip: No idea on the data, you’d have to ask him. I believe it did see publication, however, though that does not preclude it from being a joke.
It seems to me unlikely that a student would inform their professor about every death in the family unless each death was directly relevant to that student’s performance in a component of the class which could strongly affect his or her grade. Therefore, the data collected on dead grandmothers is likely to be fallible given that a student will presumably only inform the professor of the death of his or her grandmother if it happens to coincide with an upcoming exam. This data selects on the dependent variable.
This is what I also assumed (but did not conclude) when I taught … that grandmas succumbed at a pretty constant rate throughout the semester, but that a student would only inform me of her passing if they were going to miss an exam that affected their grade materially. Otherwise, what business is it of mine?
I also assumed that the grade grubbers were not going to let something like mourning come between them and an opportunity to pile on the points – hence, hence the constant FDR for the A students regardless of exam timing.
Given that the author of the Dead Grandmother article/grant proposal still has his job, one can infer (HT to Drek) that he didn’t poll his students about deaths in their families.
I seem to recall seeing the “grandmother” piece but don’t have time to look for it now. I believe the methodology involved calculating the expected number of grandparent deaths within a four month period (or a three-day period?) in a group of size N and then comparing that value to the number of reported grandparent deaths among students in a class and the clustering around exam times. So the hypothesized non-report bias for people who don’t need accommodation would have made the result of this project (if I am remembering it correctly) a conservative estimate.
Of course, grandparents do die, and some of them die just before exams or when papers are due, even the grandparents of A students.