Kolmogorov - Sminorv test results

[Dan Smith]

I have the following results of a Kolgomorov - Smirnov test:

ksmirnov tprice, by(snipe10) exact

Two-sample Kolmogorov-Smirnov test for equality of distribution functions

Smaller group________D________P-value____ Exact
----------------------------------------------
0: _______________0.0022 ______0.997
1: _______________-0.2161 _____ 0.000
Combined K-S: ____0.2161_______0.000______ .

Note: ties exist in combined dataset;
there are 749 unique values out of 1375 observations.

What does the point in the last line under Exact mean?

[Dan Smith]

anyone?

[Nick Cox]

You are being warned about the large number of ties. K-S is for comparing continuous distribution functions and the P-value calculations may not be so reliable with a strong flavour of discreteness. Or so I surmise.

Personally I just plot the distributions.

[Debarun Sengupta]

After calculating the Kolmogrov-Sminorv test, I found a note indicating that "Ties exist in the combined dataset, there are 101 unique values out of 117 observations" - What is the meaning of the line?

[Joro Kolev]

After calculating the Kolmogrov-Sminorv test, I found a note indicating that "Ties exist in the combined dataset, there are 101 unique values out of 117 observations" - What is the meaning of the line?

The meaning of this line is the same as what it was in August 2014 when Nick Cox gave the answer above: the Kolmogorov Smirnov test is in principle devised for continuous distribution, in continuous distributions you should not have ties. Stata is warning you that ties were encountered. I do not think this is a problem.

[Nick Cox]

If it helps, unique means here distinct.