Economics of Education Review 29 (2010) 138–146 Contents lists available at ScienceDirect Economics of Education Review journal homepage: www.elsevier.com/locate/econedurev The impact of high-stakes testing on student proficiency in low-stakes subjects: Evidence from Florida’s elementary science exam Marcus A. Winters a,∗ , Julie R. Trivitt b , Jay P. Greene c a Manhattan Institute for Policy Research, 52 Vanderbilt Ave, 3rd Floor, New York, NY 10017, USA b Arkansas Tech University, USA c Department of Education Reform, University of Arkansas, USA article info Article history: Received 11 September 2008 Received in revised form 30 June 2009 Accepted 9 July 2009 JEL classification: H11 I28 J24 Keywords: Educational economics Human capital Productivity High-stakes testing Accountability abstract An important criticism of high-stakes testing policies – policies that reward or sanction schools based on their students’ performance on standardized tests – is that they provide schools with an incentive to focus on those subjects that play a role in the accountabil- ity system while decreasing attention to those subjects that are not part of the program. This paper utilizes a regression discontinuity design to evaluate the impact of Florida’s high-stakes testing policy on student proficiency in the low-stakes subject of science. We confirm prior results that students in schools facing more immediate sanctions under the policy made substantial gains in the high-stakes subjects of math and reading. Contrary to the crowding-out hypothesis, we find that students in these schools made substantial achievement gains in the low-stakes subject of science as well. © 2010 Elsevier Ltd. All rights reserved. 1. Introduction School systems across the nation have adopted account- ability policies that reward or sanction schools based on their students’ performance on standardized tests. Such high-stakes testing has been a significant force in education policy since at least the 1990s. More than half the states had already implemented some form of high-stakes test before the No Child Left Behind (NCLB) Act made it universal in 2002. One of the most frequently raised concerns regarding high-stakes testing policies is that they provide schools with an incentive to focus on those subjects that play a role in the accountability system and ignore the rest (Groves, ∗ Corresponding author. Tel.: +1 646 839 3354. E-mail addresses: mwinters@manhattan-institute.org (M.A. Winters), jtrivitt@atu.edu (J.R. Trivitt), jpg@uark.edu (J.P. Greene). 2002; Gunzenhauser, 2003; Holmstrom & Milgrom, 1991; Jones, Jones, Hardin, & Chapman, 1999; McNeil, 2000; Murillo & Flores, 2002; Nichols & Berliner, 2007; Patterson, 2002). The vast majority of these policies base their rewards or sanctions exclusively on the results of reading and math tests. Though some policies are more comprehensive than others, few hold meaningful consequences for students’ performances in other subjects such as science, history, or the arts. Schools have clear incentives to respond under account- ability policies and there is a considerable and growing body of research indicating they do respond (Greene, 2001; Reback, 2008; Rouse, Hannaway, Goldhaber, & Figlio, 2007; Springer, 2008; West & Peterson, 2006). Accountability sys- tems that reward or sanction schools based on the number of students who meet certain thresholds implicitly encour- age schools to focus on low performing students. Springer (2008) and Reback (2008) both consider changes in the dis- tribution of student achievement and find no evidence that 0272-7757/$ – see front matter © 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.econedurev.2009.07.004