TABLE 10.

Regression modeling for possible SIMD interaction with bilingualism in cognitive function. glm() regression with Poisson error distribution was used.

	Dependent variable:
	TEA-1	TEA-2	TEA-3	Backward digit span
	(1)	(2)	(3)	(4)
(Intcpt)/Monolingual	1.820	1.835	1.457	1.654
	t = 8.293	t = 8.430	t = 5.562	t = 6.942
	p = 0.000	p = 0.000	p = 0.00000	p = 0.000
Bilingual	0.104	0.271	0.474	0.174
	t = 0.437	t = 1.166	t = 1.713	t = 0.677
	p = 0.663	p = 0.244	p = 0.087	p = 0.499
SIMD	−0.082	−0.127	−0.335	−0.115
	t = −0.272	t = −0.426	t = −0.943	t = −0.353
	p = 0.786	p = 0.671	p = 0.346	p = 0.725
Bilingual: SIMD	0.081	0.161	0.376	0.120
(Interaction)	t = 0.261	t = 0.527	t = 1.038	t = 0.357
	p = 0.795	p = 0.598	p = 0.300	p = 0.721
Observations	48	48	48	48
Log Likelihood	−91.376	−108.947	−131.520	−93.593
Akaike Inf. Crit.	190.752	225.893	271.040	195.185

Model 1: TEA-1 ∼ Monoling/Biling × SIMD, Poisson error. Model 2: TEA-2 ∼ Monoling/Biling × SIMD, Poisson error. Model 3: TEA-3 ∼ Monoling/Biling × SIMD, Poisson error. Model 4: Backward digit span ∼ Monoling/Biling × SIMD, Poisson error.