The false dichotomy between "fixing poverty" and "fixing schools"

See all posts
Even the most die-hard ed reformers will admit to the occasional twinge of self-doubt over the ability of teachers, however talented, to overcome the adversity that fills so many students' lives.  Sure, great teaching matters, but is it spitting in the wind to try and place a legion of great teachers in classrooms filled with many kids whose families and neighborhoods can so quickly undermine any progress? As Diane Ravitch put it, "Teachers can have a profound effect on students, but it would be foolish to believe that teachers alone can undo the damage caused by poverty and its associated burdens."

In a policy brief written with his usual clarity and elegance, UC Santa Barbara economist Richard Startz and author of Profit of Education (one of our favorite books) puts forth a reassuring argument. The fact that school-based ed reforms (such as improved teacher quality or reduced class size in the primary grades) explains only a small fraction of variation in student outcomes does not mean that those reforms aren't likely to have an important and cost-effective benefit.  Using two real-world examples and then a quantitative model of the "decomposition of variance," Startz makes the case that the important thing on which ed reformers should focus is impact, pure and simple.  He cites the well-known Tennessee STAR study of small class size as one of his two examples: the impact of class size reduction was sufficiently large that the fact that background variables "out-explain" performance effects of the intervention by thirty to one did not change the calculation that it was still a viable reform strategy.    

To paraphrase Startz, the bottom line on teacher quality is this: Because the marginal impact of having a better teacher is huge, the fact that teacher inputs explain very little of student outcomes is immaterial to assessing the promise of policy interventions related to teacher inputs. 

Whew! We feel better already!

Julie Greenberg