In the mundane world of statistical social science, everything is related to everything else, all else being equal. Rarely do honest researchers report unequivocal findings.
But that is what I saw at a recent talk at the University of Arkansas by Howard Bloom, a leading expert on regression discontinuity (look that up in your advanced statistics text), an evaluator of dozens of programs including pre-school options, and a fellow not given to overstatement.
In summarizing his methodologically rigorous Journal of Policy Analysis and Management article evaluating the impacts of small high schools on graduation rates and test scores for New York City public school children, Bloom declared that "in 35 years of conducting evaluations I have never seen impacts like these."
That got my attention. Here is what happened, and what Bloom and his team found. (Other researchers using different methods reached the same conclusions.)
When Joel Klein took over as chancellor of New York City public schools back in 2002, he fretted over his giant school district's horrendous record serving low-income and minority children, which is to say the vast majority of New York City students. Klein understood disadvantaged kids need more than instruction. They need mentoring by adults who know their names and step in when they flounder.
But how can you get that kind of personalization in a school with thousands of kids? In the real world, with real rather than superhuman educators, you can't. Traditionally, public schools responded with ever finer divisions of labor. Keep schools huge but add certified guidance counselors, psychologists, academic coaches, deans, social workers and school resource officers (cops). From 1950 to 2006 the ratio of kids to school employees fell from 19.3-1 to 8-1.
On paper it all made sense: If a teenager has psychological problems, send him to a psychologist; for scheduling, get a guidance counselor, etc. Yet in real-world schools the division of labor fed grownup turf wars, without necessarily helping kids. Non-teaching professionals see kids in occasional meetings without the fullness of daily contact. Too many kids get lost in the shuffle.
In response Klein left intact most of New York City's big high schools, but under the guidance of researchers he also built more than 200 small high schools, started and run by carefully vetted educators and social service professionals. Klein believed that in high schools of under 500 students and 50 educators, fewer teens would be overlooked and grownups could better work together to solve kids' problems. Klein also instituted a school choice scheme in which each student could request up to a dozen public high schools of any size, with admission to oversubscribed schools determined by lottery. In effect, the lotteries randomly assigned teens to treatment and control groups, enabling researchers to measure whether small schools help kids.
The impacts were dramatic. Small schools cost the same per child, somewhat improved test scores, substantially increased the percentage of college-ready graduates and surged graduation rates. Controlling for everything else, attending a small high school increased mean graduation rates from 60.9 percent to 70.4 percent, a 15.7 percent raise from the base. Though everyone did better, the biggest gains came for African-American males, those faring the worst in big high schools. Attending a small high school cut the gap between white and black graduation rates in half.
Does this mean we in Northwest Arkansas should break up our huge high schools into small ones? Not necessarily. Klein's strategy worked not just because he was a man with a plan, but also because he held power for a decade. Klein stuck with small schools when early backer Bill Gates, for reasons known only to him, pulled support and buried a favorable evaluation of the program. Klein had the time and clout to get things right rather than leaving small schools as just another half-implemented initiative torpedoed by grownup egos.
Second, New York City kids need the nurturing small schools can provide. In contrast, middle-class kids tend to get their mentoring at home. Most have the skills to excel in large, impersonal settings, and many enjoy the options a big school can offer.
The key takeaway is not to kill off big schools, but to offer small school options for parents who want them. In education, like almost everything else, the best rule of thumb is different strokes for different folks.
ROBERT MARANTO IS THE 21ST CENTURY CHAIR IN LEADERSHIP IN THE DEPARTMENT OF EDUCATION REFORM AT THE UNIVERSITY OF ARKANSAS AND HAS DONE FIELDWORK IN ROUGHLY 200 SCHOOLS.
Commentary on 11/16/2014