p value of 0.00 in a KS test what to do?

[Vesa Dike]

Hello,

I am conducting a KS test to check for normality on the my data set. the data set is made up of 380 observations and the score is reported as w/ties:

One-sample Kolmogorov-Smirnov test against theoretical distribution

Smaller group D P-value Corrected
----------------------------------------------
CDDmeanbl: 0.7794 0.000
Cumulative: -1.4973 0.000
Combined K-S: 1.4973 0.000 0.000

Note: ties exist in dataset;
there are 14 unique values out of 371 observations.


Does this imply normal distribution or are the ties influencing the data set influencing the normality of my data?

thanks
Vesa

[Maarten Buis]

The null hypothesis is (probably, you did not show us the command you used) that the variable is normally distributed. A p-value less than .05 means that you reject that hypothesis. The discreteness of your data may be a reason, but it does not have to be the main reason.

Why do you think you need normality/Gaussianity?

[Marcos Almeida]

Hi Vesa,

Welcome to the Forum!

Unfortunately, you didn't display the "exp" you selected for the default command:

. ksmirnov variable_name = exp

Apart from the insightful comments in #2, please note what the Stata manual particularly says about one-sample KS test:

[...] test whether x in the example above is distributed normally. Kolmogorov –Smirnov is not a particularly powerful test in testing for normality, and we do not endorse such use of it;

Best,

Marcos

[Carlo Lazzaro]

Vesa:-
you may also want to visually inspect your data by -qnorm-

[Vesa Dike]

Thank you all! I worked it out eventually I didnt need to test for normality