I don't know from the nortest package, but it should ***always*** be the case that you test hypotheses H_0: The data have a normal distribution. vs. H_a: The data do not have a normal distribution. So if you get a p-value < 0.05 you can say that ***there is evidence*** (at the 0.05 significance level) that the data are not from a normal distribution. If the nortest package does it differently (and I don't really see how it possibly could!) then it is confusingly designed. I rather suspect that its design is just fine, and that it does what it should do. cheers, Rolf Turner rolf at math.unb.ca Original message:> I have a question regarding normality testing with the nortest > package. I have to admit, my question is so general that it might more be > suited a newsgroup such as sci.math. However, just in case it is > implemented differently, I was hoping someone who has used it can help me > out. > > I just want to double check that for all of the tests, the null > hypothesis is whether or not the input distribution *differs* with the > normal distribution. So, if you get a p-value less than (say) 0.05, you > can reject the null hypothesis at a 95% confidence level and say that the > input distribution is *not* normal. > > So these tests check/confirm whether a distribution is not > normal...as opposed to confirming that it is normal. Does this sound > about right? > > Ray
Hi Rolf, On Sun, 21 May 2006, Rolf Turner wrote:> I don't know from the nortest package, but it should ***always*** > be the case that you test hypotheses...> If the nortest package does it differently (and I don't really see > how it possibly could!) then it is confusingly designed. I rather > suspect that its design is just fine, and that it does what it should > do.Thank you for your reply. As I had expected, it was my knowledge of the test which was inadequate and it had nothing to do with the package. No doubt it works fine! Your explanation sounds very reasonable to me...thanks again! Ray
Rolf Turner wrote:> I don't know from the nortest package, but it should ***always*** > be the case that you test hypotheses > > H_0: The data have a normal distribution. > vs. > H_a: The data do not have a normal distribution. > > So if you get a p-value < 0.05 you can say that > > ***there is evidence*** > > (at the 0.05 significance level) that the data are not from a > normal distribution. > > If the nortest package does it differently (and I don't really see > how it possibly could!) then it is confusingly designed. I rather > suspect that its design is just fine, and that it does what it should > do. >I suspect so as well. If you think something is wrong, please contact the package maintainer (CCing; he's not reading R-help posts). Uwe Ligges> cheers, > > Rolf Turner > rolf at math.unb.ca > > Original message: > > >> I have a question regarding normality testing with the nortest >>package. I have to admit, my question is so general that it might more be >>suited a newsgroup such as sci.math. However, just in case it is >>implemented differently, I was hoping someone who has used it can help me >>out. >> >> I just want to double check that for all of the tests, the null >>hypothesis is whether or not the input distribution *differs* with the >>normal distribution. So, if you get a p-value less than (say) 0.05, you >>can reject the null hypothesis at a 95% confidence level and say that the >>input distribution is *not* normal. >> >> So these tests check/confirm whether a distribution is not >>normal...as opposed to confirming that it is normal. Does this sound >>about right? >> >>Ray > > > ______________________________________________ > R-help at stat.math.ethz.ch mailing list > https://stat.ethz.ch/mailman/listinfo/r-help > PLEASE do read the posting guide! http://www.R-project.org/posting-guide.html
rwan at kuicr.kyoto-u.ac.jp wrote:> One thing that threw me off (and this is not really specific to > Nortest as it seems to be correct, but just my understanding), but the > p-value seems quite unstable. For example:.... It is worth noting that if the null hypothesis is true, then the p-value is uniformly distributed on [0,1]. This should be kept in mind when assessing the ``instability'' of p-values. cheers, Rolf Turner rolf at math.unb.ca