-2

Quick question. My brain shot circuited after all the new data.

What was the sig that was legit: was it from ,05 or from ,005?

I have sig = ,019 Can I use it or does it have to be smaller?

Tim
  • 108,699
  • 20
  • 212
  • 390
Fred
  • 1
  • 1
  • 0,05 is pretty standard, and lower is the requirement to reject the null hypothesis :) – MikeP Apr 14 '17 at 14:11
  • So when it is written ,05, that is 0,05, correct? So ,019 is more than enough? :D – Fred Apr 14 '17 at 14:15
  • Possible duplicate of [Regarding p-values, why 1% and 5%? Why not 6% or 10%?](https://stats.stackexchange.com/questions/55691/regarding-p-values-why-1-and-5-why-not-6-or-10) – Tim Apr 14 '17 at 14:16
  • @Fred yup, that is enough to reject the null hypothesis and thus "show" the alternative (which is what you typically wish to do with a stats test) – MikeP Apr 14 '17 at 14:22
  • 1
    Your question seems to assume we're aware of some context, but you'll need to provide it. What do you mean? Who says what is legitimate where? Significance levels shouldn't be chosen by some arbitrary standard that pays no attention to context in any case -- suitable significance levels vary by circumstances. There is no situation in which it's reasonable to choose your significance level *after* you calculate a p-value. If you don't have your significance level figured out you shouldn't be looking at data. – Glen_b Apr 14 '17 at 14:54
  • You choose the level whatever suits you best. Do you want a significant result... then choose .05 as the level. Do you want a non-significant... result choose .005 as the level. In the end, the p-values are just a *nuanced* measure of significance and the sharp boundaries for hypothesis tests are subjective. That means there is more, outside of the data, which determines the levels. You can move away from the cynical viewpoint if the decision rules are given some basis (which we can not give to you without further information). – Sextus Empiricus Feb 16 '18 at 11:10

1 Answers1

1

.05 is a pretty common $α$. .005 isn't. The other conventional choices for $α$ are .1 and .01.

Kodiologist
  • 19,063
  • 2
  • 36
  • 68