Here's a compelling statistical analysis that in most cases, yes, recruiting ranking do matter, and are accurate. They divide the top 72 teams (AQ teams+AAC+Independents)  into groups that ended up ranking best to worst in recruiting, and show that 5-star teams (Alabama, Florida, Texas, etc) consistently beat 4-star teams (Arkansas, Nebraska, Texas A&M, etc), and so on, with the lowest ranked group, 1-star teams (UCF, U.Conn, Temple, etc), winning least often. They did find a few teams that seem to consistently do better than recruiting rankings alone would predict, such as Kansas State.

Left unanswered,  is to what extent there is any accuracy below the top 72 Teams. I would presume that if they lumped the other 54 teams into a group, and called it 0-star, they would lose to the 1-star teams most of the time. That doesn't tell us, however, whether you could group those 54 teams into 3/4 star, 1/2 star, 1/4 star, and 0 star teams, and find the same relationships. Given that the recruiting services don't put much emphasis in recruits below the 3-star level, I would suspect that the relationships would be a lot more murky.

I agree that the 3-star and above ratings are players that the recruiting services actually looked at and seriously evaluated. The problem is when you get below that. To put this in perspective, this year WMU had a great recruitng class, the highest rated one I have seen in the MAC. Even then, they ranked #69 overall. The 1-star schools were ranked:
Kansas State 49
Syracuse 50
Boston College 52
Iowa State 56
Duke 58
Central Florida 59
Wake Forest 62.
Boise State 67
Temple 71
SMU 78
Memphis 79
Connecticut 105

Thus, WMU would rank at the lower end of the 1-star schools. Ohio has a long way to go to get there. They and the rest of the MAC would be 0-star schools on that ranking system. Most AAC schools are in the 1-star rank, I think. They seem to recruit midway between the MAC (0-star) and the lower AQ levels (2-star).