- #1

kimberley

- 14

- 0

Wikipedia states:

"The [Jarque-Bera] statistic has an asymptotic

**chi-square distribution with two degrees of freedom and can be used to test the null hypothesis that the data are from a normal distribution**. The null hypothesis is a joint hypothesis of both the skewness and excess kurtosis being 0, since samples from a normal distribution have an expected skewness of 0 and an expected excess kurtosis of 0. As the definition of JB shows, any deviation from this increases the JB statistic."

When I look at the Chi-Square Distribution Table at the .05 confidence interval, it returns the number 5.99. Out of an abundance of caution, does this mean that if my Jarque-Bera test statistic is greater than 5.99, that the null hypothesis of normality is rejected? This would seem to be correct since the JB statistic is usually only greater than 5.99 if the skew and excess kurtosis are relatively far from 0, and the JB statistic tends to be closer to 1 or less than 1 when skew and excess kurtosis are close to 0. Thank you in advance.

Kim