Comparison and pooling of independent correlation coefficients

Fisher-z pooled estimate with interval (eqs 7.401, 7.407-7.408), the homogeneity chi-square sum (n_i - 3)(z_i - zbar)^2 on k - 1 df (7.409) or against rho0 on k df (7.406), the two-sample z (7.400), and the weighted r_gem = sum (n_i - 1) r_i / sum (n_i - 1) with its t on n - k - 1 df (7.395-7.396).

Usage

corcmp(r, n, rho0 = NULL, confidence = 0.95)

Arguments

r

Correlations (at least two).

n

Their sample sizes.

rho0

Optional hypothesised common correlation.

confidence

Interval level.

Value

Named list: r_pooled, ci, z_pooled, se_pooled, chi2, df, p_value, chi2_rho0, p_rho0, z_two, p_two, r_gem, t_gem, p_gem (the rho0 and two-sample entries only when they apply).

References

Hedderich, Sachs & Reynarowych (2023), eqs (7.395)-(7.409).

Examples

corcmp(c(0.6, 0.7, 0.8), c(28, 33, 23))$chi2
#> [1] 1.827348