[1] looks a pretty reasonable overview of the research on homework effectiveness. The meta-analysis is quite supportive of homework as an effective way of improving results:
With only rare exceptions, the relationship between the amount of homework students do and their achievement outcomes was found to be positive and statistically significant. Therefore, we think it would not be imprudent, based on the evidence in hand, to conclude that doing homework causes improved academic achievement.
With only rare exceptions, the relationship between the amount of homework students do and their achievement outcomes was found to be positive and statistically significant. Therefore, we think it would not be imprudent, based on the evidence in hand, to conclude that doing homework causes improved academic achievement.
[1] http://www.ascd.org/publications/educational-leadership/mar0...