Table 2. Changes in affective tone of recent marathon-related coverage predicts distress, startle reactivity, perceptual sensitivity, and shooting behavior.

Outcome	B	SE	t-ratio	df	p	Cohen’s d
Self-reported Distress
Model Intercept	9.11	0.74	12.25	90	< .001**	2.58
Model Slope	-28.26	12.42	2.28	90	.025*	0.48
Startle Amplitude
Model Intercept	27.21	1.65	16.47	90	< .001**	3.47
Model Slope	-137.02	50.86	2.69	90	.008**	0.56
Perceptual Sensitivity
Model Intercept	0.64	0.03	19.58	90	< .001**	4.12
Model Slope	1.74	0.62	2.79	90	.006**	0.59
Response Bias	0.38	0.08	5.02	90	< .001**	1.06
Threat Response Bias
Model Intercept	0.15	0.03	4.78	90	< .001**	1.01
Model Slope	-0.77	0.58	1.33	90	0.185	0.28
P. Sensitivity	0.27	0.06	4.25	90	< .001**	0.9

Note: Higher affective tone values indicate more positive content while lower affective tone values indicate more negative content. Model uses robust standard errors (SE; i.e., random effects). Model coefficients (B) are unstandardized. Model Slope represents the coefficient estimates for the variable affective tone, which is centered around each participant’s own mean. Slopes can be interpreted as the predicted change in the outcome variable associated with a 1 unit increase in affective tone. For example, a participant’s startle amplitude is predicted to be 137.02 μV lower when there are an even number of positive and negative affective words in recent media coverage related to the Bombings than when all the affective words are negative.

*p < .05

**p < .0125 (Bonferroni-corrected alpha)