8.3.2.6.3 Comparison of KS and
ChiSquare GoodnessofFit Tests
The KS test is superior to c2 in the following ways:
(1) 

The KS Test can be used to test for deviations in a
given direction, while ChiSquare Test can by used only for a
twosided test.

(2) 

The KS Test uses ungrouped data so that
every observation represents a point of comparison, while the
ChiSquare Test requires the data to be grouped into cells with
arbitrary choice of interval, size, and selection in starting point.
Minimum expected frequency values are required.

(3) 

The KS Test can be used in a sequential
test where data become available from smallest to largest,
computations being continued only up to the point at which rejection
occurs. 
The ChiSquare Test is superior to the KS Test in the following
ways:
(1) Chisquare can be partitioned and added
(2) Chisquare can be applied to discrete
populations