Table 4.

MR-CAUSE analysis displaying that the sharing model has a significantly worse fit than the causal model for the association between ADHD and the outcomes education, income, TDI, as well as intelligence, respectively

Model 1a	Model 2a	Δ ELPDb	s.e. Δ ELPD	z-score	p-value
Education (SNPs=1430)
Null	Sharing	−6.9	2.2	−3.2	0.00072
Null	Causal	−13.0	4	−3.3	0.00041
Sharing	Causal	−6.4	1.8	−3.5	0.00023
Income (SNPs=1430)
Null	Sharing	−3.1	1.5	−2.1	0.018
Null	Causal	−8.2	3.6	−2.3	0.011
Sharing	Causal	−5.1	2.1	−2.4	0.008
TDI (SNPs=1430)
Null	Sharing	−5.2	2.2	−2.4	0.0092
Null	Causal	−10.0	4.1	−2.5	0.007
Sharing	Causal	−4.8	1.9	−2.5	0.0063
Intelligence (SNPs=1430)
Null	Sharing	−0.91	0.77	−1.2	0.12
Null	Causal	−4.60	2.7	−1.7	0.043
Sharing	Causal	−3.70	1.9	−1.9	0.027

^aModel 1 and Model 2 refer to the models being compared (null, sharing, or causal)

^bModel fit is measured by Δ Expected Log Pointwise Posterior Density (ELPD); Negative values indicate that model 2 is a better fit