Biserial Correlation. One of the assumptions of Point-Biserial correlation is that there is similar spread between the two groups of the binary variable. 1958 An index of item-criterion relationship.
In such cases the point-biserial correlation generally under-reports the true value of the association. The point biserial correlation computed by biserialcor is defined as follows r frac overline X_1 - overline X_0sqrt pi 1 - pi S_x where overline X_1 and overline X_0 denote the sample means of the X-values corresponding to the first and second level of Y respectively S_x is the sample standard deviation of X and pi is the sample. The rank-biserial correlation is appropriate for non-parametric tests of differences - both for the one sample or paired samples case that would normally be tested with Wilcoxons Signed Rank Test giving the matched-pairs rank-biserial correlation and for two independent samples case that would normally be tested with Mann-Whitneys U Test giving Glass rank-biserial correlation.
Biserial correlations are most often used in social sciences when validated instruments are compared to non-validated instruments.
The biserial correlation can be calculated with XLSTAT. The point biserial correlation computed by biserialcor is defined as follows r frac overline X_1 - overline X_0sqrt pi 1 - pi S_x where overline X_1 and overline X_0 denote the sample means of the X-values corresponding to the first and second level of Y respectively S_x is the sample standard deviation of X and pi is the sample. Biserial correlations are most often used in social sciences when validated instruments are compared to non-validated instruments. You can check for this assumption by plotting your continuous variable in each of your two groups and visually identifying if the spread of the data is similar.
