Table 2.

Weekly activity attributable to kynect and insurance company advertising, Kentucky, 2014 and 2015 open enrollment periods.

Outcome		Actual, n	Model-based predictions
			Prediction (with observed advertising), na	Counterfactual (no kynect advertising), na	Change attributable to kynect advertising, n (%)b	Counterfactual (no insurance company advertising), na	Change attributable to insurance company advertising, n (%)c
Information-seeking behavior
	Page views	1,978,749	1,925,863	1,474,730	451,133 (23)	1,804,879	120,984 (6)
	Visits	105,618	101,616	79,538	22,078 (22)	98,061	3555 (3)
	Unique visitors	61,838	58,964	37,037	21,927 (37)	55,242	3722 (6)
Enrollment activity
	Web-based applications	9643	9050	5576	3474 (38)	11,759	–2709 (–30)
	Total applications	11,832	11,232	8072	3160 (28)	12,516	−1283 (−11)

^aRepresent model-based predictions (using the model results shown in Multimedia Appendix 3 and Multimedia Appendix 4) of the weekly outcomes. These comparisons use the 2014 and 2015 open enrollment periods to estimate how much activity was attributable to advertising. Calls and Marketplace enrollment are not included because neither kynect nor insurance company advertising during open enrollment had a significant effect on these outcomes.

^bDifference (%) between "Prediction (with observed advertising)" and "Counterfactual (no kynect advertising)".

^cDifference (%) between "Prediction (with observed advertising)" and "Counterfactual (no insurance company advertising)".