4

I sometimes encounter a view that only perfect forecasting is really forecasting.

For example, if I claim that I have a model which forecasts election results, people will think I'm making the absurd claim that I can forecast election results with perfect accuracy. When I explain that my forecasts have errors but the errors are, say, 10% smaller than chance would suggest, I am told that I'm "not really" forecasting.

Is there a formal name for this fallacy? I know that this is a special case of a fallacy of equivocation, assuming the word forecast is being used colloquially when it is being used technically. But is there a more specific name?

Andre Silva
  • 3,080
user4945913
  • 81
  • 4
    I wouldn't call not understanding statistics a "fallacy"... – Tim Jun 28 '15 at 20:00
  • I agree, but I'm asking because of my experience that people are more willing to accept that they are in error when they are told that they are committing a named fallacy, rather than being told that they do not understand something. – user4945913 Jun 30 '15 at 02:13
  • 1
    So make-up some name based on beliving-named-fallacies fallacy you observed... Related: http://dilbert.com/strip/2008-05-08 – Tim Jun 30 '15 at 07:11
  • 2
    And seriously: your clients/audience would learn more if you described them your method and results, rather then throwing some "technical" name that "explains" their lack of understanding. Names do not really solve things. – Tim Jun 30 '15 at 07:21
  • I don't think this question is on topic, as this is an issue of communication, not specific to statistics. For instance, in the context of law, many people misunderstand the meaning of "anonymity" as defined by the GDPR. The issue is similar. Perhaps this is a question for https://philosophy.stackexchange.com? – J-J-J Mar 05 '24 at 13:27
  • Maybe a question like "How to explain the idea of forecasting to a lay audience?" would be more on-topic, but it doesn't look like that's what OP was after. – J-J-J Mar 05 '24 at 13:32
  • 1
    More generally, I have even encountered PhDs (in the physical sciences!) who apparently believe (according to sworn testimony) that "random" means "completely arbitrary, without any discernible pattern or regularity." This seems to be a layperson's impression that has not been resolved even by 20+ years of formal education -- and therefore is something it behooves us to be aware of when we do try to communicate statistical ideas. – whuber Mar 05 '24 at 18:31

1 Answers1

2

It's hard to prove a negative. But I have been forecasting for over 18 years now, and have not encountered a specific name for this effect. Perhaps "the perfect is the enemy of the good" fits what you are looking for.

In the meantime, you or your interlocutors may find our resources on forecasting helpful: Resources/books for project on forecasting models

Stephan Kolassa
  • 123,354