Table 2. Regression results predicting TV viewership and Twitter volume.

Dependent Variable	Model	Independent Variable(s)	Model R2	Model Adj. R2	Model p-value	Beta Coefficient (standardized)	Beta p-value
TV viewership	LR	Composite EEG score	0.57	.57	<10−5	0.76	<10−5
TV viewership	SWMR	Full model	0.68	.66	<10−5
		Alpha/theta power				1.07	<10−5
		Alpha/beta asym.				0.41	<10−3
		Theta/gamma power				0.30	0.003
High TV viewershipa	SWMR	Full model	0.72	0.69	<10−5
		Alpha/theta power				1.10	<10−5
		Alpha/beta asym.				0.52	0.002
Low TV viewershipa	SWMR	Alpha/theta power	0.17	0.13	0.048	0.41	0.048
Twitter volume	LR	Composite EEG Score	0.63	0.62	<10−5	0.80	<10−5
Twitter volume	SWMR	Full model	0.63	0.61	<10−5
		Alpha/theta power				0.71	<10−5
		Theta/gamma power				0.50	<10−5
		Alpha/beta asym.				0.39	<10−3
High Twitter volumea	SWMR	Alpha/beta asym.	0.48	0.44	<10−3	0.68	<10−5
Low Twitter volumea	SWMR	-	-	-	-	-	-
TV viewership	LR	Twitter volume	0.51	0.50	<10−3	0.72	<10−5
TV viewership	SWMR	Full model	0.67	0.66	<10−5
		Composite EEG score				0.51	<10−5
		Twitter volume				0.40	0.001
Twitter volume	SWMR	Full model	0.67	0.66	<10−5
		Composite EEG score				0.58	<10−3
		TV viewership				0.29	0.027

Adj.–adjusted, Asym.–assymmetry, LR–linear regression, SWMR–step-wise multiple regression.

^a For these analyses, TV viewership and Twitter volume values were split along the median for each variable and full step-wise multiple regression models were conducted on each median-split subgroup of data.