swapping—in which those who hold confidential data run a specific analysis for other researchers and then strip out identifying information—is another approach.

In Currie’s view, a great deal of valuable information is locked up in administrative data sets that are not currently accessible—and making use of them could be a cost-effective way to answer important questions. Many participants supported the idea, noting, for example, that “we are not going to be reproducing the Perry Preschool study any time soon, and we don’t want to wait around for 40 years [but] we are going to be implementing these programs.”

Looking at the back-of-the-envelope estimates Magnuson had described as well as Currie’s linkage approach, a participant noted that they are “useful—if you know what the cost is. If even a rough estimate that you think is an underestimate is still higher than the cost of the program that you are thinking about,” you have enough information to go forward. Moreover, these kinds of approaches make it possible to look at far larger samples: “We can break it down for different types of children so we can look at whether there are differences in these patterns by children with different backgrounds or different ethnicities—true data may be best, but we are never going to have large enough samples given the cost of collecting it.”