The Washington Post asked a number of experts what aspects of American life they'd get rid of—sort of a cultural spring cleaning. Economist Peter Orszag, a former member of Obama's cabinet, chose something that will sound familiar here in Chicago: the 3 PM school day (emphasis mine).

A longer school day, especially if combined with other steps such as intensive tutoring and frequent feedback to teachers from administrators, could improve academic achievement. For example, Roland Fryer and Will Dobbie of Harvard University have found that a longer day is a key aspect of high-performing charter schools in New York. Fryer found similar results for traditional middle and high schools in Houston: Implementing reforms there, including a longer school day, produced large gains in math performance (albeit modest to no gains in reading). More time at task improves student achievement.

There's a lot contained in "other steps" and "intensive tutoring." The Houston example is instructive; the reforms were, as Fryer describes (PDF), not just a longer school day, but "increased instructional time, a more rigorous approach to building human capital, more student-level differentiation, frequent use of data to inform instruction, and a culture of high expectations." With regards to instructional time, it's more complicated still. First, they ran into a familiar roadblock:

In an ideal world, we would have lengthened the school day two hours and used the additional time to provide two on one tutoring in both math and reading. This is the model developed by Michael Goldstein at the MATCH school in Boston.

Due to budget constraints, we were able to lengthen the school day one hour and tutor in one grade only. We chose sixth and ninth grades in an effort to get students up to grade level when they entered middle and high school, and we chose math over reading because of the availability of a solid curriculum and knowledge map that is easily communicated to first time tutors.

Why did they run into budget constraints? Hiring tutors, even at the comparatively low wages paid for the new hires, gets expensive when the tutoring is as intensive as that applied to the nine "treatment" schools:

We hired 250 tutors…. Tutors were paid $20,000 with the possibility of earning an average bonus of $3,500 based on tutor attendance and student performance.

At the base salary of $20,000, that's $5 million for one subject in two grades over nine schools for two-on-one tutoring, at a cost of $2,500 per student; Chicago Public Schools spent between $13,000 and $13,500 per student in 2009-2010.

Non-tutored students got "double-dose" classes if they were below grade level in math or reading, for an extra 189-215 hours of instruction. How'd it go? For the tutored students, it went well:

As one would expect given the results already presented, the effects of tutoring are positive and quite large: 0.309 [standard deviation] in sixth grade, 0.392 in ninth grade.

For double-dose students, not as good:

In eighth grade math we show a positive and statistically significant effect of 0.235. This is an anamoly [sic] relative to the other subject-grade pairs. All other results are small and statistically insignificant. Pooling across all four grades, the estimated effects are 0.072 in math and -0.014 in reading.

Reading was worse, if not (statistically) significantly worse. But the authors acknowledge that's pretty typical:

The difference in achievement effects between math and reading, while striking, is consistent with previous work on the efficacy of charter schools and other educational interventions. Abdulkadirogluet al. (2009) and Angrist et al. (2010) find that the treatment effect of attending a Boston “No Excuses” charter school is four times as large for math as ELA. Dobbie and Fryer (2011a) demonstrate effects that are almost 5 times as large in middle school and 1.6 times as large in elementary school, in favor of math. In larger samples, Hoxby (2009) reports an effect size 2.5 times as large in New York City charters, and Gleason et al. (2010) show that an average urban charter school increases math scores by .16 with statistically zero effect on reading.

The authors give a few reasons for this, including the theories that it's easier to catch students up in math than reading for developmental reasons.

Houston did what Chicago intends to do: add an hour (in Chicago, it would be an hour and a quarter) to the school day. There were gains, but they were strongest where that was combined with an intensive, expensive two-on-one tutoring session that necessitated new hires for that purpose—which shouldn't come as a surprise given how much longer the school day can be for teachers than students.

Photograph: ecastro (CC by 2.0)