The easiest way to observe retention rates is the world population number. Now that the fancy graph version is down it's a little harder. But right now we have 320 people online. I think that's... a little biased due to a secondary key giveaway perhaps though?
Like I said I'd need graphs to see. What we did observe was 100>300 server pop with the first giveaway. And the retention rate over the first week was roughly 200-250 total pop. That means more than half the players who started stayed through at least the noob stages. That seems pretty good to me.
Always remember sample size issues
You have a fairly small number of observations (5 characters). That can lead to randomness having a lot of effect. In my area 3 noob groups just started up, and all were still there after a week. If I took that to indicate general game health, Salem's doing awesome.
It's been neat to see the evolution of a game. Salem has come so far, and still has far to go. Although frustrating, I think it's been an experience worth the effort.