Not sure if this is the appropriate forum, but:

Is there a way to measure the effect of stack exchange on correct usage of the English language?

  • 6
    It's negative. The more questions we answer, the more questions come in. Nov 26, 2014 at 18:31
  • 8
    No. None. Not if one expects any accuracy. First, how would you measure "correct usage of the English language"? Usage by whom, for instance, or correct according to whose standards? Nov 26, 2014 at 18:32
  • Thanks for the responses, they meet my expectations. Should this question remain here (or anywhere for that matter)?
    – Minnow
    Nov 26, 2014 at 18:38
  • 5
    @Dan Bron: Shouldn't that be "most of the regulars here are descriptivists"? We tend to describe what people actually say, rather than prescribing to them what they should say. Nov 26, 2014 at 18:38
  • 1
    @John Lawler: Good point. It's a bit like asking if visiting ELU has helped you to stop beating your wife. Nov 26, 2014 at 18:40
  • @Fumble, yes, that's what I intended to write, but I got autocorrected (autoincorrected?) to the wrong word by my phone because I had a spelling error which confused it. And now it's too late to change it :/
    – Dan Bron
    Nov 26, 2014 at 18:41
  • 2
    Thanks for the responses, they meet my expectations. Do you mean the answers are good or that they were just as bad/eccentric/elitist/obtuse as you expected? If the latter, was this some kind of experiment rather than an actual question? (Is this response as paranoid as you expected?)
    – itsbruce
    Nov 26, 2014 at 19:02
  • 1
    You can measure anything with enough of the right data. 'enough' and 'right' can be difficult though. You could setup an experiment where you get some people to use ELU and some people to not, take an English proficiency test before and one after some time, and do 'stats'. But I suspect that's not easy. One could do this retrospectively (find people who do and do not use ELU, and some common measure of English proficiency (expert coded proficiency level of question text). Again, I expect not so easy. But it's possible!
    – Mitch
    Nov 26, 2014 at 20:59
  • @itsbruce the comments met my expectations that this would be difficult (and possibly posted in the wrong forum). I've accepted the Bayesian suggestion, so in the end have a potential solution.
    – Minnow
    Nov 27, 2014 at 3:58

2 Answers 2


I know that my math ability has improved because of Khan Academy. Now that I've viewed a lecture on Bayes' Theorem, I think you might be able to answer your question with a dash of Bayes.

Roughly, you are looking for the probability that one's English Language usage improves given one has visited EL&U.

That is, you are looking for P(A|B) where A = "one's English Language usage improves" and B="one has visited EL&U"

To calculate, you'd need three other probabilities:

  1. The probability that one has visited EL&U given that one's English Language usage improved (that is, P(B|A)).
  2. The probability that one's English Language usage improved (that is, P(A)).
  3. The probability that one has visited EL&U (that is, P(B)).

I think that 1 and 2 are inherently hard to measure, as you would need a rigorous definition of improvement, such as a better grade on an English essay or a higher score on a standardized test. In regard to probability 1, it's hard to measure because you will have trouble finding teachers who will grade the same essay twice, once before EL&U, and once after EL&U.

You may also have to take into account that some questions are more about satisfying curiosity, for example, What does "ratchet" mean and when was it first used?. I can't imagine that one would improve either a grade or an SAT score by knowing this etymology.

Other answers about usage, such as Mixing single digit and double digit numbers - how to write formally, might have a slight improvement on the way that some users write out numbers or use figures more consistently. This is arguably a dubious improvement that few readers would recognize.

If you're serious about the question, you may want to visit our sister site, https://stats.stackexchange.com/, and ask how to go about designing a test to measure 1 and 2. I think that 3 could be estimated by the number of EL&U visits by distinct visitors divided by the number of distinct sites visited.

  • Great answer and approach. I will consider the advice about a post on stats exchange.
    – Minnow
    Nov 27, 2014 at 3:59
  • You're quite welcome. When someone hands you a really good hammer, it's surprising how many things start to look like nails.
    – rajah9
    Dec 1, 2014 at 14:53

Given the tremendous quantity of absolutely atrocious questions and answers on this site, frequenting this site is likely to lead to chronic frustration and despair at humanity. This will lead to a degradation of quality of life, which will undoubtedly decrease your own ability to speak coherently.

I don't know if there are cases of this occurring here at English, but it has happened on Stack Overflow from time to time.

  • Based on the rating, it would seem that this question is a cobble in that same road to the hell.
    – Minnow
    Nov 28, 2014 at 18:56
  • It may even lead to a confusion between quantity and number, which I am sure we none of us want. Dec 2, 2014 at 20:25

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .