My plans for my (1996-97) year at the Center for the Advanced Study in the Behavioral Sciences (CASBS) were firm: I was definitely doing nothing—other than a final edit of a book and tidying up a textbook manuscript. I had earned a rest, and I was going to take it. California is la-la land, and its 70 to 80 degrees worth of temperature were to be savored. My job was to hike, play tennis, and (maybe) attend a few seminars at which others gave presentations.
Unfortunately, it turned out to be the most hectic year of my life, one that transformed the rest of my career. On the eve of my departure for the West Coast, one of my students, Jay Greene (now chair of the Department of Education Reform, University of Arkansas), and I had released a paper to be presented at the annual meetings of the American Political Science Association, which found that students receiving school vouchers scored higher in math and reading (after three and four years) than did students remaining in the Milwaukee Public School system. Our data came from a randomized experiment embedded in a larger survey of student performance. An earlier observational study had found the Milwaukee voucher program to have had no impact on student academic prowess.
Our study was released the very week Senate Majority Leader Robert Dole (unbeknownst to us) announced support for school vouchers in his speech accepting the Republican party’s nomination for the presidency. That was the first time any presidential candidate of a major party had ever embraced this controversial approach to school reform. As a consequence, the results from our study hit the front page of the New York Times, hundreds of other newspapers ran an Associated Press story, the Wall Street Journal ran the op-ed piece we sent to them, and dozens of radio and TV shows wanted interviews. No less important, the author of the earlier observational of the Milwaukee program called me a “snake” in a feature story that ran a few weeks later in the upper right-hand column of the front page of the Wall Street Journal. Nor was he the only person to publicly question my character in the heated political atmosphere stirred up by the voucher debate.
None of this would be worth mentioning were it not for the fact that the hullabaloo caught the eye of a group of Wall Street philanthropists who had set up a foundation called the School Choice Scholarships Fund (SCSF) in order to fund a school voucher program in New York City in early 1997. The impetus for their initiative was an invitation issued by Cardinal John J. O’Connor, Archbishop of New York, to Rudy Crew, Chancellor of the New York City public school system, to “send the city’s most troubled public school students to Catholic schools” and he would see that they were given an education. When New York City Mayor Rudolph Giuliani attempted to raise the funds that would allow Catholic schools to fulfill the offer made by the Cardinal, his proposal encountered strong opposition from those who saw it as a violation of the First Amendment’s establishment clause. As the controversy raged, SCSF announced that it would cover a portion of the costs of the private education of 1,000 low-income students. SCSF gave students a choice of any participating private school in New York City. It offered a chance to win a scholarship to all elementary students from low-income families who were currently attending public schools in grades 1 through 4 or about to enter first grade.
When I heard about SCSF’s plans, I urged the foundation to allow me to make use of the lottery feature of their plan to conduct a randomized experiment. The philanthropists who founded SCSF had (I later learned) little time for Harvard professors, but after seeing I had been called a “snake” in public deemed me worthy of their respect. I could run the experiment and report without restriction anything I found, but they simply refused to waste a dime of their precious scholarship money on an academic exercise: All their money would go to help students. Yet the cost of the experiment, if conducted by MPR, a firm well known for its conduct of randomized trials, would cost millions.
Fortunately, I was playing tennis, and on the Stanford courts one day I learned that David Packard, a trustee of the Packard Foundation, had become interested in supporting research on school vouchers. Within a couple of days I made the case to Mr. Packard that only via a randomized trial would we ever know whether school vouchers really worked. He understood the logic of that argument, but he had a major objection: If a lottery is used to award the vouchers, then in most cases only one child in a family would go to the private school while siblings would attend other schools. Families should not be divided in these ways.
That was a reasonable objection, I thought, but I was unable to persuade SCSF to cough up more money so that siblings could attend private schools, too. They did not have the resources to go beyond the 1,000 students to which they were already committed.
At that point I told Packard that he was dead right but the problem was something he could solve. The Packard Foundation could pay for both the siblings and the research. Of course that doubled the costs. Still, he agreed, and the project moved forward.
But by now 1996 was all but history, and the vouchers were to be handed out in early 1997. Over the next six weeks we had to put together a plan that would allow for the baseline testing of students as well as the surveying of parents about family background characteristics and their opinions about their children’s education. Then even worse news broke: 26,000 applications had been received in February 1997. Our administrative capacities were overwhelmed.
MPR devised the solution: They decided on two step lotteries that after the first random draw excluded from consideration most of the applicants, reducing the number of students tested to the few thousand necessary to have enough observations to cover those who would win the lottery and an equivalent number who would form the control group. After these students were tested, the second lottery was held to determine the winners.
Still worse news. Voucher lotteries had become so popular I was asked to mount two more experiments in Dayton, Ohio and Washington, D. C., the latter becoming the crucial forerunner of what today is known as the District of Columbia Opportunity Scholarship Program, a highly successful, if still controversial, voucher-style program in the nation’s capital, which is funded by the U. S. federal government.
To help handle the burgeoning undertaking, I turned to William Howell, a quantitatively qualified Stanford student (now the Sidney Stein Professor at the Harris School, University of Chicago) who I had met (where else?) on the Stanford tennis courts, asking him if he would help me struggle through the morass. When he agreed, I knew my problems could be solved.
The two of us had to orchestrate all the projects from CASBS via phones run by a switchboard operator, a FAX machine, and a clunky email system which did not allow for attachments (as was standard in those bygone days). At the end of the year I was given prizes for being the CASBS fellow who had talked the most on the phone, sent the most emails, and sent the most FAXES. My scholarly credentials remained dubious, however. CASBS documentary records very probably can prove the accuracy of these assertions!
Nothing before in my career had been this exciting, and everything since has been conditioned by this experience. Just this year (2015) the Journal of Public Economics published the latest research piece to emerge from the New York City study. In that article my former student, Matthew Chingos (now a Senior Fellow, Brookings), and I investigate the impact of the NYC voucher program on college graduation rates nearly 20 years later, the first such study ever undertaken. The finding: Minority students who had the opportunity to use a voucher were 35% more likely to graduate from a four-year college than were the members of the control group.
Were we able to obtain these results because of CASBS? Or would the randomized experiments have taken placed even if I had delayed my time at CASBS for another year or two? In the words of experimental research: Did CASBS have a causal effect?
The case for a causal effect is handsome. Time was critical in 1996 and 1997 as the timing of the voucher lotteries were being set by others, and only at CASBS would I have had the time and complete freedom to devote myself fully and entirely to this complex undertaking. Furthermore, outdoor tennis courts close to large foundations with trustees interested in school vouchers are an extremely rare type of species in New England.
In the CASBS library can be found a book by William G. Howell and Paul E. Peterson entitled The Education Gap: School Vouchers and Urban Schools (Brookings, 2002). I am told the book should rest on these shelves only if inspired by experiences at CASBS. This book belongs.
And my tennis game, honed to perfection by steady practice under Howell’s tutelage, continues to prosper–at least when I can see the ball and a backhand is not required. What a year!
Leave a Reply