Skip to contents

Tests two correlation coefficients (given the sample-sizes) to determine if they are statistically similar.

Usage

compare.correlations(corr1, corr2, n1, n2, method);

Arguments

corr1

First Spearman's Rho (raw)

corr2

Second Spearman's Rho (raw)

n1

First sample-size

n2

Second sample-size

method

What correlation method should be used? Currently only supports Spearman's rho

Value

Returns a vector of length nine containing: corr1, corr2, sd1, sd2, delta.corr, pooled SD, p-value, n1, n2

References

For Spearman's rho we use the formulae from: Fieller et al. "Tests for Rank Correlation Coefficients. I," Biometrica, Vol. 44, No. 3/4 (Dec. 1957)

Author

Mehrdad Shamsi

Examples

# this small differential correlation is actually borderline significant (p = 0.10) 
compare.correlations(
  corr1 = 0.15,
  corr2 = -0.15,
  n1 = 55,
  n2 = 80,
  method = 'spearman'
  );
#> [1]  0.15000000 -0.15000000  0.02038462  0.01376623 -0.30228087  0.18479948
#> [7]  0.10189745 55.00000000 80.00000000