Scott, it's Ken again. And I would say, it's generally what we had expected. So let me give you the reasoning behind it. But I would say the Connecticut information has not been shared formally though. They'll be sharing their membership files with us in November, and it is preliminary, but we've had conversations and tied into it. So I'd first say that, it's pretty well documented, the time and the process used to go through the application process and enrollment. And therefore, the individuals that sign on early are going to have a higher acuity or be older and looking for coverage if they were unable to get affordable coverage previously. We expect to receive the actual enrollment files from the state in November, and we'll be able to analyze the business more in detail at that point. But I would say, the Connecticut numbers, if you look at them as reported, it was about 3,800 enrollees that were reported on. Half went to Medicaid, half went to the exchanges and 2/3 of the enrollment went to us or about 1,200 members. And that's just too small a membership number to draw any conclusion on. But based on our simulations, and we did over 58,000 simulations, to prepare and understand consumer behavior. We did see that the older would choose sooner, they're more interested, they're more likely to purchase gold in part because they're more risk averse and in part because of the higher actuarial value they're seeking. So there's no surprise in the numbers. What it does indicate, and what Wayne stated earlier is, to get the right risk pool, we need to do marketing, we need to pull people through the exchanges, we have planned for that. Our business and implementation control centers are ready to dial that up when appropriate, but we're going to get a better risk pool and the right risk pool as we market through a variety of channels, which include telesales, our broker support, direct mail, Internet purchasing and our search tools. So we have a number of activities that we'll be able to pull the membership in.