Table 2.

Results from performing causal inference on simulated data sets

		Simulation model
Method	Tested model	a	b	c	d	E	f	g	h	i	j	k	l
SEM	a	−6	689	688	689	689	686	685	282	1,381	1,380	282	283
	b	504	−6	402	503	− 6	399	1,088	148	148	553	148	841
	c	504	400	−6	− 6	505	1,091	398	148	552	148	840	148
	f	1,090	685	1,091	1,600	1,090	−6	684	688	282	687	− 6	686
	g	1,089	1,091	684	1,092	1,601	686	−6	687	686	282	687	− 6
BUF	a	156.01	120.93	120.7	155.97	156.18	120.84	121.06	98.79	98.84	98.95	98.80	98.78
	b	120.78	120.95	83.68	120.94	156.19	83.58	−0.08	83.61	83.78	37.89	83.76	−0.08
	c	120.95	83.64	120.73	155.98	121.13	−0.09	83.84	83.69	37.81	83.90	−0.08	83.71
	m	35.06	37.28	−0.03	−0.01	35.05	120.93	37.22	15.1	61.03	15.05	98.88	15.06
	n	35.23	−0.02	37.03	35.03	−0.01	37.25	121.14	15.18	15.07	61.06	15.04	98.86
DEAL	a	−1,019	−1,359	−1,360	−1,378	−1,379	−1,343	−1,343	−1,697	−2,245	−2,244	−1,689	−1,689
	b	−1,254	−1,003	−1,200	−1,264	− 1,019	−1,263	−1,530	−1,196	−1,618	−1,821	−1,620	−1,954
	c	−1,254	−1,199	−1,004	− 1,019	−1,263	−1,530	−1,196	−1,618	−1,821	−1,625	1,954	−1,619
	d	− 1,016	−1,011	−1,012	−1,025	−1,025	−1,010	−1,010	− 1,551	− 1,560	− 1,560	−1,553	−1,554
	e	− 1,016	−1,011	−1,012	−1,025	−1,025	−1,010	−1,010	− 1,551	− 1,560	− 1,560	−1,553	−1,554
	f	−1,541	−1,339	−1,537	−1,794	−1,550	−1,004	−1,339	−1,880	−1,693	−1,889	− 1,548	−1,884
	g	−1,541	−1,536	−1,341	−1,549	−1,793	1,338	−1,005	−1,880	−1,888	−1,693	−1,883	− 1,548
	m	−1,544	−1,688	−1,886	−2,148	−1,904	−1,338	−1,673	−2,027	−2,377	−2,573	−1,684	−2,019
	n	−1,544	−1,884	−1,690	−1,902	−2,147	−1,671	−1,338	−2,027	−2,573	−2,377	−2,019	−1,683
BNLEARN	a	−976	−1,322	−1,323	−1,323	−1,321	−1,322	−1,320	−1,671	−2,214	−2,215	−1,667	−1,668
	b	−1,230	−973	−1,178	−1,229	− 974	−1,176	−1,516	−1,601	−1,596	−1,799	−1,598	1,945
	c	−1,231	−1,176	−975	− 974	−1,228	−1,522	−1,173	−1,602	−1,799	−1,597	−1,944	−1,599
	d	−985	−984	−985	−984	−984	−984	−982	− 1,536	− 1,530	− 1,531	−1,531	−1,533
	e	−9,85	−984	−985	−984	−984	−984	−982	− 1,536	− 1,530	− 1,531	−1,531	−1,533
	f	− −1,530	−1,325	−1,531	−1,784	−1,528	−979	−1,320	−1,876	−1,669	−1,871	− 1,527	−1,874
	g	−1,532	−1,528	−1,328	−1,529	−1,782	−1,325	−977	−1,878	−1,872	−1,669	−1,873	− 1,528
	m	−1,521	−1,663	−1,869	−2,123	−1,864	−1,317	−1,658	−2,012	−2,353	−2,555	−1,663	−2,009
	n	−1,523	−1,866	−1,665	−1,867	−2,119	−1,663	−1,315	−2,013	−2,556	−2,353	−2,009	−1,663

Cells represent the average (over 1,000 replicates) of the scores describing how well each model fits the data. Columns represent data simulated under the 12 different scenarios and rows describe which model is being tested. Each of the four methods uses a different score to assess model fit. For SEM, low numeric scores indicate better fit. For the other three methods, higher numeric scores indicate better fit. Average score(s) that indicate the preferred model out of those tested are underlined. Cells with bold indicate the correct model choice.