I have 3 groups of data A, B,C, each having 10 number of samples. when I analyzed for their Z-values (both skewness and kurtosis), all three groups of data are normally distributed. In Shapiro-wilk test, only group A and C are normally distributed; while in Kolmogorov- smirnov test, only group A and B are normally distributed.
Now. On the basis of the number of data in each group and all 3 methods of analysis, which one should I have to choose for explanation of the distribution of my data.
the standard test in econometrics is the Jarque-Bera test based on the skew and kurtosis, but you can also use tests based on the frequency domain (Hinich and Patterson)
Dear Kumari
Because of small sample size it is hard to test normality as Dr. Amin mentioed specially using graphical approach, but in this case you might use the normal probability plot if it linear, then your sample is normaly distributed, this link may help you
good luck
http://www.oswego.edu/~srp/stats/npp_small.htm