KTAU procedure
Calculates Kendall's rank correlation coefficient τ (R.W. Payne & D.B. Baird).
Options
Parameter
Description
KTAU calculates Kendall's rank correlation coefficient (known as τ i.e. tau) between pairs of samples. The samples can be stored in different variates and supplied in a list in the DATA parameter. Alternatively, they can all be placed in a single variate, and the GROUPS option set to a factor to indicate the sample to which each unit belongs.
The PRINT option controls the printed output, with settings:
By default these are both printed.
The CORRELATIONS option allows the correlations to be saved, in a scalar if there are only two samples or in a symmetric matrix if there are three or more. Similarly, the probabilities can be saved using the PROBABILITIES option. Also, you can use the NORMAL option to save a transformation of τ that approximately follows a Normal distribution with mean zero and variance; this provides reasonably accurate probabilities when the number of units N is no smaller than 8 (see Kendall 1948).
Options: PRINT, GROUPS, CORRELATIONS, PROBABILITIES, NORMAL.
Parameter: DATA.
Method
Kendall's rank correlation coefficient τ is a measure of association between the rankings of two variables measured on N individuals. It is calculated as
τ = S / √(NC1 × NC2)
S is defined as the sum of
SIGN(xi - xj) × SIGN(yi - yj)
over all pair of distinct units i and j. NC1 and NC2 are the number of valid comparisons (removing ties and missing values) that can be made amongst the first and second set of samples, respectively. (See Siegel 1956, pages 213-223.)
The transformation of τ into a Normal random variable is given by
τ / √( (2 × (2 × N + 5)) / (9 × N × (N - 1)) )
The probabilities are calculated using procedure PRKTAU.
Action with
RESTRICT
If any of the variates in DATA is restricted, the statistic is calculated only for the set of units not excluded by the restriction.
References
Kendall, M.G. (1948). Rank Correlation Methods. Griffin, London.
Siegel, S. (1956). Nonparametric Statistics for the Behavioural Sciences. McGraw-Hill, New York.