While RIMS II, IMPLAN, and other methods regularly deviate substantially from each other in the answers they yield to analytical questions, is there a "band" within results should be considered "more accurate".

For example, if RIMS and IMPLAN give a quantitative result on the order of x, and a third method gives a result 2,3, or more orders of magnitude higher, should the third method be disregarded?

I ask this because I've submitted analysis based off industry-specific methods and had it roundly rejected not because there was a flaw in my analysis, but because it didn't agree with general case tools, or differed to markedly.

Is there a best practice within econometrics that says "throw out all results farther than Y from some accepted average"?

  • $\begingroup$ I am unfamiliar with what RIMS II and IMPLAN are. Could you please add a line of explanation or a link. $\endgroup$ Nov 19, 2014 at 2:17
  • $\begingroup$ Will do when not on mobile. $\endgroup$ Nov 19, 2014 at 2:17

1 Answer 1


In econometrics and statistics, you sometimes will truncate outliers. Otherwise, maybe it's worth mentioning that in the business of predicting, analysts are punished for either being wrong, but their punished even more if they were boldly wrong. This is a form of herding behavior. Deviating from the consensus can be disproportionately risky.

Here's a paper that talks about it: "Security analysts’ career concerns and herding of earnings forecasts" in the RAND Journal of Economics.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.