D study: project a G study onto a proposed facet size

A D study holds the estimated variance components fixed and asks what the coefficients would be for a different number of conditions. Both coefficients increase monotonically in n and tend to 1, which is what makes the D study a decision tool.

Usage

Genvdm(G_components, n_proposed, target = 0.8)

Arguments

G_components: Length-3 vector (var_p, var_i, var_pi).
n_proposed: One or more candidate numbers of conditions.
target: Generalizability coefficient the decision looks for.

Value

List with estimate, e_rho2, phi, meets_target, n_required, n, method.

Details

Formula: E rho^2(n) = var_p / (var_p + var_pi/n); Phi(n) = var_p / (var_p + (var_i + var_pi)/n).

References

Brennan (2001), Generalizability Theory, Springer, ch. 3.

Examples

Genvdm(c(0.5, 0.2, 0.3), n_proposed = 20)
#> $estimate
#> [1] 0.9708738
#> 
#> $e_rho2
#> [1] 0.9708738
#> 
#> $phi
#> [1] 0.952381
#> 
#> $meets_target
#> [1] 1
#> 
#> $n_required
#> [1] 20
#> 
#> $n
#> [1] 1
#> 
#> $method
#> [1] "D study projection of fixed variance components, Brennan (2001) ch. 3"
#> 
#> attr(,"class")
#> [1] "morie_rich_result" "list"