When to Use Fancy Words

Yes, you can say "prior"

I saw this tweet last week, and I think that it is wrong for interesting reasons that tell us something about effective communication.

I usually agree with the idea that you should simplify your writing when possible, and accept the underlying premise that a lot of writing, particularly in academia, seeks to hide behind jargon and an obfuscatory style in order to signal intelligence, make work seem more profound than it is, and prevent engagement with others.

At the same time, I do believe that there are often good reasons to say prior instead of assumption or stochastic instead of random.

Compare these two sentences.

1. I have a prior that every presidential election in the modern era will be close.

2. I have an assumption that every presidential election in the modern era will be close.

At first glance, these two sentences appear to be saying the same thing. Yet there are subtle differences that a writer communicates by choosing prior or assumption here (I agree that Bayesian prior would be much too haughty and provide no benefit over just prior). A prior is something you have good reason to believe based on previous evidence. When you bring your prior with you to analyze a question, you are evaluating new data in light of it.

In contrast, saying you have an assumption indicates that it’s something you haven’t thought that carefully about. The folk wisdom that “when you assume, you make an ass of u and me” gets at the idea that the beliefs we refer to when we use that term are often unfounded. No one would say something similar about priors, since we understand that some priors are always necessary, and the only question is whether the ones we hold are justified. The word functions as an invitation to debate.

By using prior, I am therefore telling you that I have good reasons for my belief, or at least what I think are good reasons. Moreover, I’m communicating something about myself and the way I think. Having a prior means I’ve studied statistics, or at the very least I’ve heard of Bayes Theorem. You have been given a reason to think that my analysis might be worthwhile. Your prior about me as a writer should in fact shift.

This kind of signalling might be undesirable if it’s used, as the tweet above implies, to needlessly make one’s writing less accessible without contributing any new information about the writer, his process, or the content of the work itself. Yet prior is not an obscure term. If you haven’t heard of it, you probably are not that well-read. And even if you haven’t run into the word before, the meaning is obvious enough from the context in which it is used most of the time. Prior is even shorter than assumption, so that takes away another argument against using the fancier word.

Now, if you want to say assumption so that your writing has as wide of an appeal as possible, that’s fine. In my case, I’m usually willing to risk losing lazy and unintelligent readers in order to communicate in a way that is clearer to those who are more sophisticated. One can carry out a similar analysis with stochastic versus random.

Another consideration here is that I’d like to nudge people towards learning some basic statistics, and want to create a norm encouraging them to do so. Stats is like no other area of study, as it opens up a large number of literatures and fields of research. When I was a graduate student, I was talking to a professor who had a PhD in political science but now had an appointment in genetics, and I asked how he did it. His response was that “it’s all the same stats,” which is true. During covid, there was a fight between a CNN anchor and Peter Navarro, a Trump administration adviser, over whether hydroxychloroquine worked. This is how the conversation went.

White House trade adviser Peter Navarro on Monday said he was qualified to engage and disagree with Dr. Anthony Fauci on the use of an anti-malarial drug as a coronavirus treatment – which is not yet proven as effective – saying, “I’m a social scientist.”

“Doctors disagree about things all the time. My qualifications in terms of looking at the science is that I’m a social scientist,” he told CNN’s John Berman on “New Day.” “I have a Ph.D. And I understand how to read statistical studies, whether it’s in medicine, the law, economics or whatever.”

Navarro’s remarks follow reports that he clashed with officials in the Situation Room over the weekend about the unproven efficacy of hydroxychloroquine in treating coronavirus. While the task force was discussing the latest on the anti-malaria drug, an exasperated Navarro lashed out at Fauci, the nation’s top infectious disease expert, who has urged caution around the drug, a person familiar with the meeting told CNN.

I think Navarro is a pretty goofy figure, but he’s right on his qualifications here, regardless of what one thinks about whether hydroxychloroquine works. The gold standard in medicine is the Randomized Control Trial, and when the results come in such studies are analyzed using the same tools that are common across the social sciences. Most social science is in fact bunk, but it’s useful to know enough statistics to understand why, and be able to appreciate the few cases where it goes right.

I think that the problem with the advice to always simplify is that it assumes that language is simply about clear communication, ignoring other goals and the wider social context. It is not good to take an argument that could be a blog post and make it into an article five times as long by adding clutter in the form of useless citations and the repetition of the same arguments over and over again. This is what academia incentivizes. When I was writing for journals, there were multiple occasions where I had basically said all I wanted to say in an article but knew I had to get the paper up to a certain word count for it to be published. We need less of that.

Academia appears to have evolved these norms because if papers could be as long as blog posts, how would anyone distinguish the scholars from the hobbyists? Journal papers in the social sciences are longer than blog posts on the same subjects less because academics have more to say than the fact that professors need to justify their positions as credentialed experts rather than amateurs. A lot of smart people can write decent blogs, but no one has the time to track down all the citations you need to be published in the American Political Science Review unless they’re working full-time as a professor. On the surface, more words and more citations signal “I know more about this,” but in reality they just show that one has devoted more time to it, in the hopes that observers can’t tell the difference.

Academia has made a specialty of bad kinds of signalling, but this doesn’t mean that there aren’t good kinds. When you read my articles, I want you to know that I am a smart person who has thought carefully about the issues I write about, and I have some of the basic skills and training necessary to make sense of the world. There are legitimate choices here to be made regarding tone, writing style, and word choice, and other things to consider besides broad accessibility.