Table 1.

Evidence for differences in accuracy, proportion of Heywood cases, and correctness of indicator-to-factor correspondences between implementations

	N = 180			N = 450
Property	Inc	Eq	Diff	Inc	Eq	Diff
Data sets without negative eigenvalues
RMSE	17	0	91	11	1	96
Heywood	0	108	0	0	108	0
Ind-to-Fac Corres	31	32	45	19	46	43
Data sets with negative eigenvalues
RMSE	1	0	9	0	0	9
Heywood	0	10	0	0	9	0
Ind-to-Fac Corres	5	2	3	1	0	8

Tally of type of evidence for the 216 different models (108 population models × 2 sample sizes) derived from Bayesian regression analyses. The row sum for the rows concerning data sets with negative eigenvalues are smaller than 216 because data sets with negative eigenvalues occurred only for some models. RMSE = Root mean squared error. Heywood = Probability of the occurrence of a Heywood case. Ind-to-Fac Corres = Difference in indicator-to-factor correspondence from found solution to population model. Inc = Inconclusive evidence. Eq = Conclusive evidence for no relevant difference between the implementations (equality). Diff = Conclusive evidence for a difference between at least some implementations