When the distribution of variables is not Normal, the degree of relationship between the variables can be determined using Rank correlation. Instead of using the precise values of the variables, the data are ranked in order of size, and calculations are based on the differences between the ranks of corresponding values X and Y.
- Variable Y - Variable X: select the 2 variables of interest.
- Filter: (optionally) enter a data filter in order to include only a selected subgroup of cases in the statistical analysis.
- Correlation coefficients: select Spearman's rho and/or Kendall's tau. The confidence interval for Kendall's tau is estimated using the bias-corrected and accelerated (BCa) bootstrap (Efron, 1987; Efron & Tibshirani, 1993). Click bootstrapping options such as number of replications and random-number seed. for
Click OK or press the Enter key to obtain the following statistics in the results window.
In this example the Spearman's coefficient of rank correlation rho is 0.114. The 95% confidence interval ranges from -0.084 to 0.304. The associated P-value is 0.255 and the conclusion therefore is that there is not a significant relationship between the two variables.
When you want to print these results, click Print on the File menu, or press Ctrl P.
- Altman DG (1991) Practical statistics for medical research. London: Chapman & Hall.
- Armitage P, Berry G, Matthews JNS (2002) Statistical methods in medical research. 4th ed. Blackwell Science.
- Bland M (2000) An introduction to medical statistics, 3rd ed. Oxford: Oxford University Press.
- Efron B (1987) Better Bootstrap Confidence Intervals. Journal of the American Statistical Association 82:171-185.
- Efron B, Tibshirani RJ (1993) An introduction to the Bootstrap. Chapman & Hall/CRC.
Statistical Methods in Medical Research
Peter Armitage, Geoffrey Berry, J. N. S. Matthews
Buy from Amazon