Type I and Type II errors' definition following this link is as below

A Type I error means rejecting the null hypothesis when it’s actually true. It means concluding that results are statistically significant when, in reality, they came about purely by chance or because of unrelated factors.

A Type II error means not rejecting the null hypothesis when it’s actually false. This is not quite the same as “accepting” the null hypothesis, because hypothesis testing can only tell you whether to reject the null hypothesis. Instead, a Type II error means failing to conclude there was an effect when there actually was. In reality, your study may not have had enough statistical power to detect an effect of a certain size.

From this link, they provide some ambiguous examples in clinical studies. I am wondering if there are some economics or finance examples to explain them intuitively. For example, I got lost regarding the bold phrases.

  • 4
    $\begingroup$ Does this answer your question? If not why? economics.stackexchange.com/questions/27677/… $\endgroup$
    – 1muflon1
    Commented Jul 22, 2021 at 10:08
  • $\begingroup$ @1muflon1: Thank you for your help, I am voting for closing this question too, I understand now. I am wondering if there is any link explaining simply why we fall into type I and type II errors then? It would be a big help. Many thanks. $\endgroup$ Commented Jul 23, 2021 at 11:23
  • 1
    $\begingroup$ what do you mean about why? The errors exists because statistics is based on randomness. You cannot observe the true data generating processes (DGP). The true laws of nature, human action, psychology etc are unobservable. We only get a random sample of point generated by the DGP. Think of it as youtube algorithm from perspective of content creator. You do not know what the algorithm is only youtube knows that but you can observe which videos get recommended, you can test different tags etc, but you never know what the process actually is. So you form hypotheses like youtube prefers short $\endgroup$
    – 1muflon1
    Commented Jul 23, 2021 at 11:28
  • 1
    $\begingroup$ videos, make a lot of short videos and a lot of long videos, and compare the number of views between them and test whether one gets significantly more views. If so you can reject null hypothesis of no preference in favor of preference for short videos but you cant be 100\% certain, that is impossible. So there will always exist type I and II errors in any setting where you cannot directly observe what DGP is (which you never can really even physicists cant do that in their experiments) $\endgroup$
    – 1muflon1
    Commented Jul 23, 2021 at 11:30
  • $\begingroup$ I got it now, thanks @1muflon1 for your dedicated explanation. $\endgroup$ Commented Jul 23, 2021 at 11:31


Browse other questions tagged or ask your own question.