Hai, i have sample about 300. If i want to check data distribution, and skewness and kurtosis say the data normally distributed, but kolmogrov smirnov says otherwise, what is the best conclusion for this data distribution?
Thankyou
Greetings,
You may have a look at to this paper;
Article TO DETERMINE SKEWNESS, MEAN AND DEVIATION WITH A NEW APPROAC...
Hi Harrison,
Since different tests calculate deviation from Gaussian distribution using different methods, it is not surprising to observe different results. According to RB D'Agostino "The Kolmogorov-Smirnov test is only a historical curiosity. It should never be used". In general, D'Agostino-Pearson omnibus test is a more reliable method.
Hope this helps,
Parham
Greetings,
As we are dealing with behavioral sciences, in which hundred percent precision is not possible. There are various methods to check nature or distribution of data among which one of the easy and friendly method is to check the range of z value of skewness and kurtosis which you can get by dividing the value of skewness by the standard error of skewness so as with kurtosis. According to Huck, Cross and Clark (1986) the range of z -value for the normality of data must be within -2.58 to +2.58 which indicates that z -score would fall in a normal curve under the range of -3 to +3 whereas Doane and Seward (2011) found a more stringent criterion for the normal distribution and suggested the range of z-value between -1.96 to +1.96. Other than this it is also recommended to go through the graphical representation of data. Thus the distribution of data should also be checked with the assistance of Normal Probability Curve (NPC), Quantile-Quantile plot (Q-Q plot) .
- Badges
- Science topic
- Similar topics
- Medicine
- Physiology
- Biosignals
- Biomedical Signal Processing
- Statistical Analysis