Nonparametric Methods in Statistics with SAS Applications

60 TESTS FOR CATEGORICAL DATA

For

α

= 0.01 and n = 9, a one-sided critical value is equal to 0.783. The value of

r

s

is larger than the critical value, therefore, the null hypothesis is rejected, and the

conclusion is that there is a significant positive correlation present.

We arrive at the same conclusion after running these lines of code in SAS:

proc freq data=glaucoma;

table Tx*Cx;

exact scorr;

run;

The output is:

Spearman Correlation Coefficient

Correlation (r) 0.8333

Exact Test

One-sided Pr >= r 0.0041

Alternatively, the CORR procedure may be run to obtain an approximate P-value.

The code is

proc corr data=glaucoma spearman;

var Tx Cx;

run;

The results are

Spearman Correlation Coefficients, N = 9

Prob > |r| under H0: Rho=0

Tx Cx

Tx 1.00000 0.83333

0.0053

Cx 0.83333 ...