I often hear academics say things like: “It is not necessarily the case that the evidence would support that.”
Is this sentence meaningless or just trivial? I don’t know, but I am still surprised by how many otherwise reasonable people hide behind such verbiage. Other common examples of the defensive use of modal diction:
1. “It could be impossible.”
2. “It’s certainly possible.”
3. “It mustn’t be inherently so.”
4. “It must indeed be admitted both that it would require powerful arguments to counter the strong presumption in favor of such a conclusion and that these arguments have not yet been adequately stated.” (Hayek)
What’s the alternative? Plain, simple, clear, clean, bog-standard, unadorned probabilities.
Instead of “It could be impossible,” say “The probability is 3%” or “The probability is 17%” or whatever. Instead of saying, “It’s certainly possible,” say “The probability is greater than zero,” or “The probability is roughly 10%.” Instead of “It mustn’t be inherently so,” say “The probability is only 86%.” Do specific numbers seem overly precise? Then switch to “highly unlikely,” “even odds,” “almost certain,” and such.
Above all, instead of talking like Hayek, talk like Orwell.
You could reply, “We’re in a Prisoners’ Dilemma. Academia crushes anyone who doesn’t quality their assertions with a pile of modal terms.” But frankly, this seems paranoid to me. Outside of Grievance Studies, academics will crush you for unwarranted certainty. Otherwise, though, you can qualify your assertions like an eagle – or like a weasel. So why be a weasel?
READER COMMENTS
nobody.really
Jan 16 2020 at 10:19am
Richard Feynman, The Character of Physical Law, Cornell University Messenger Lectures (1964)
Doug Gabbard
Jan 16 2020 at 12:44pm
When researchers are communicating with a general audience, I frequently hear sentences that begin, “There is no evidence that . . .” This ambiguous clause should be excised from public discourse. Instead, the researcher should choose one of the following:
“This question has not been extensively studied, so we don’t really know whether your position is correct.”
OR
“This question has been studied extensively, and the evidence indicates that your position is incorrect.”
john hare
Jan 16 2020 at 4:51pm
Isn’t this just jargon as applied to economics? I’ve noticed people in several fields that try to use specialized jargon to keep outsiders out. An amusing game some of us used to play was finding complicated and obscure ways of saying things. “Up the proverbial tributary without feasible means of locomotion” is the one I remember from several decades ago.
RPLong
Jan 17 2020 at 8:51am
One problem is that, if you are a frequentist, it’s not possible to generate these kinds of numerical likelihoods without extensive statistical analysis. Meanwhile, if you are a Bayesian, then the number is absolutely as vague and weaselly as the non-numerical language: Just because I say the likelihood of something is 5% doesn’t mean it actually is; and even if the event does eventually occur, that doesn’t mean that the likelihood was greater than 5% all along — it might still have been 5%.
So dressing up intuition with numerical likelihoods doesn’t actually solve the problem here.
Michael Stack
Jan 29 2020 at 12:45pm
I don’t think the issue here is dressing up intuition with numbers – if it were, I’d agree with you. I think the issue Bryan is combating is around clarity of language.
I remember reading Hayek and absolutely hating the way he wrote. I’m glad I’m not the only one who felt that way.
Phil H
Jan 18 2020 at 12:55am
I like this idea. I have a small worry about spurious precision, but I think the value gained through increased clarity far outweighs any problems. I certainly agree with BC’s central complaint: saying that X is possible generally carries zero informational value.
Mark Z
Jan 18 2020 at 7:27pm
While I agree that people should be more precise, I don’t think using probabilities is useful; it would convey, as Phil H notes, a great deal of spurious precision. It also ignores that the probability itself is uncertain. “I am absolutely certain that candidate X has exactly a 50% chance of winning the election, and if we could rerun it over and over would win 50% of the time” is not the same as, “I don’t have enough information to infer how likely he is to win, so my estimation of the probability of winning is a uniform prior (between 0 and 1), the average of which of course is .5.”
It would suffice for people to say “probably,” “almost certainly”, etc. But then one could argue that the colloquial use of “could” or “possible” is understood to convey a meaningful estimation. I would take “it could be so” for example to mean that it’s neither extremely unlikely to be so nor almost certain to be so. It suggests wide uncertainty about the probability of it being so, maybe 10%-90%. And maybe that degree of uncertainty is often warranted.
It’s a point worth repeating: a probability value given is not exhaustively informative. The underlying model may be hierarchical. If I tell you I think there’s a 50% probability of a coin flip being heads, it tells you nothing about what I think the parameter p is determining the frequency of heads vs. tails, i.e. my confidence about whether the coin is a fair coin. I may in fact be certain it’s a rigged coin that will either always be heads or always tails, and just not know which side it’s rigged for. If we want to be full-blown Bayesians, maybe we should demand that people tell us what the hyper-parameters are for the distribution of their probability estimate.
Vivian Darkbloom
Jan 19 2020 at 6:57am
Tax lawyers use modal language to indicate the level of certainty when issuing tax opinions. While modal language is used, behind each modal phrase there is a universally accepted range of confidence (sometimes expressed in percentages). Customarily, there are five levels of confidence (here in ascending range):
1. Reasonable basis
2. Substantial authority
3. More likely than not
4. Should
5. Will
For a more humourous scale, see here: http://www.lawprofessorblogs.com/taxprof/linkdocs/2005-5011-1.pdf
The use of modal language does not need to be “weaselly” or “meaningless”. I find that the communication problems that economists have, even when speaking amongst themselves, frequently arise from the lack of a standard and universally accepted language backed up by some authority (such as Black’s law dictionary, definitions supplied within a Code or established by legal precedent, etc). Without such authorities economics too easily decsends into the realm of rhetorical gamesmanship. And, unlike other professions, there is no legal price (such as liability for malpractice) to prevent such abuse.
Comments are closed.