I don't know if it's true but it should be. Either way, it's believable and hysterical.
Joe Martino tells TOF that "One of the most horrible examples I ever encountered was the use of a Cobb-Douglas Production Function to predict the effectiveness of bombing the Ho Chi Minh trail. When I first saw the model it fairly screamed 'wrong!' But the people who put it together saw nothing wrong with it."
A Cobb-Douglas Production Function, a quick wiki search discovers, is a model designed to predict input-output relationships in manufacturing. I'll assume that my readers understand my amusement.
The article from which the quote came, a general overview of bad statistical modeling:
https://tofspot.blogspot.com/2014/03/americas-next-top-model-part-ii.html
The article which led me to the aforementioned article, a review of a study in which 70 teams given the same data and same hypothesis came up with over 1,000 models, none of which produced the same result:
https://www.wmbriggs.com/post/43052/
And finally, the article I'll be reading after I finish the first listed above, a review of the same study by a different author:
https://wattsupwiththat.com/2022/10/17/a-hidden-universe-of-uncertainty/
And in case you haven't been following along, the synopsis of the case against modeling:
- Models say what they are designed to say
- Models make significant numbers of assumptions, all of which change the results of the model
- People assume that a small p-value means the model is accurate (hint: all the studies in the example discussed by Briggs had wee p-values, even the ones that contradicted each other in results)
