My long-form article ("Pre-K Decay") in the current print version of National Review follows a series of shorter pieces on preschool that I've written in the last few months.
"Does Government Preschool Add Value? Probably Not" makes the point that a government preschool program should offer "value added" rather than just substitute for the early education that children are already receiving through private providers or their parents. A randomized experiment properly measures value added, while regression discontinuity design (a popular non-experimental method) does not.
"Public Pre-K Defenders’ Latest Excuse: If Only Elementary Schools Weren’t So Bad..." takes on a common defense of government preschool. Preschool "works," according to this view, but the gains fade out due to the poor performance of elementary schools that children go on to attend. The trouble is that no one knows how to reliably and substantially improve K-12 schooling, and--even if we did--it would not necessarily make preschool any more valuable.
"Can de Blasio Look to New Jersey’s Pre-K Success? The Evidence Is Weak" details the problem of self-selection bias that plagues non-experimental preschool studies. New Jersey's preschool program for its "Abbot districts" is hailed as a great success, but the evidence is based on a simple matching exercise that cannot adequately deal with selection bias.