I don't have any definitive explanation, but I would point out a few things:

1) There is a lot more data for the Camry hybrid, so it is much less likely to be skewed by a handful of (possibly atypical) data points - in either direction.

2) The Camry hybrid has been available for more than a year, whereas for most of the Altima hybrids in the database, this is the first winter. That means that seasonal effects have had time to average out in the Camry's case, but not in the Altima's data - and this is compounded by the small sample size (see #1).

3) I am skeptical about the EPA numbers, in that the Altima has a slightly larger engine; other things being equal, I'd expect the Camry to have a small mileage advantage (though probably more like 0.5mpg, not 2-3mpg). The new EPA numbers are a lot closer to typical driver experience than the old ones, but they are still only a rough guide.