CODEWITHSUNDEEP
rstatistics
rrr
rrr

Reputation: 371

Shapiro-Wilk test and validity of confidence interval

I have done a Shapiro-Wilk test for a data with 200 values;

shapiro.test(values.txt)

W = 0.89901, p-value = 2.176e-10

The 99% confidence interval for this data is [54.82, 59.02] with mean = 56.92, stdev = 11.49. Does the shapiro-wilk test support the validity of the confidence interval, or argue against it? What does W indicate?

Upvotes: 0

Views: 666

Answers (1)

Chris Ruehlemann
Chris Ruehlemann

Reputation: 21400

The Shapiro-Wilk test tests for normality. That is, it answers the question whether or not the data in a vector come from the normal distribution. The test assumes normality as its H0(null hypothesis). If the test returns a p-value smaller than a commonly accepted significance threshold the data can be assumed to come from a non-normal distribution. That's what the test does. I don't think that the question of the confidence interval plays into that. (BTW: the Wvalue is negligible; it is reported but almost never discussed).

Upvotes: 2

Related Questions