The problem with football stats is we don't quantify much, at least in terms of public data, and a lot of what we do quantify isn't done accurately (tackles most famously). Most people would agree that offensive line play is very important, yet it isn't directly measured in the box score. A lot of advanced metric sites get around this by inventing new stats, but that requires watching a lot of tape. So, the only way we can is by watching, too, hence my comment. (Though, even in a perfect environment, I'm a fan of combining analytics with traditional scouting, they really serve two purposes.)
When you watch Rutgers, you immediately see why looking at their stats gives a false impression. They just aren't very talented at all. That's hard to tell in the box score playing Eastern Michigan and Morgan State. Even against Washington, I'd say that result is more in Washington's favor if played in week 3 vs week 1. They just took a bit to get going in that game.