0

How can I (in R) test whether the given data set is normally distributed? I read related questions suggesting shapiro.test. Unfortunately, according to wikipedia, the null-hypothesis is "data are normal".

I expect my data to be normal, hence I need a test whose null-hypothesis states "data are not normal" and with some level of certainty (e.g. 0.05) states that null-hypothesis doesn't hold.

What test should I use?

Community
  • 1
petrbel
  • 431
  • 1
  • 4
  • 10
  • 1
    This is a question of statistical understanding, not programming, so I've voted to migrate it to CrossValidated. – Thomas Mar 31 '15 at 13:32
  • You are using the correct test. You put in the null what you think is true. – user603 Mar 31 '15 at 14:17
  • 1
    See http://stats.stackexchange.com/questions/2492 – whuber Mar 31 '15 at 14:19
  • @user603 ok but test may not be strong enough... how can I decide that? – petrbel Mar 31 '15 at 14:52
  • @petrbel: I'm sorry I don't know where to start from. Have you read the links suggested by whuber? Maybe try to reformulate a new question in light of them – user603 Mar 31 '15 at 15:28
  • @user603 here it is, I hope it is more clear now - http://stats.stackexchange.com/questions/144236/reversed-shapiro-wilk – petrbel Mar 31 '15 at 15:50

0 Answers0