The third component of the school choice decision is usage, i.e. actually changing school, conditional on the availability of new choices. Perhaps the clearest evidence comes from the randomized field trials reported by Howell and Peterson (2002). In the first year of the field trial, a significant proportion of children offered the voucher did not use it and attended public school (18% in New York; 20% in Dayton; and 32% in Washington, DC; see Howell and Peterson, 2002, Table 2-2). In subsequent years, these rates increased, such that at least one-third of all students offered a voucher were not using it (30%, 33%, and 63%, respectively). In addition, a non-trivial proportion of children who were not offered the voucher attended private school regardless (6%, 17%, and 11%, respectively, in the first year). And, these rates were conditional on eligibility, not application status. A second example is the evidence from the CPS lottery programs: Cullen et al. (2005) report that winning the lottery only raises attendance rates at preferred schools by 36%. Finally, recent evidence from the Washington DC School Choice Incentive Act shows the complexity of school choice with vouchers. Of the 41,000 eligible students within Washington Public Schools (DCPS), 1,848 were actually eligible when criteria related to student characteristics were applied. Across the characteristics of grade, income, and initial status, there were very different probabilities of scholarship receipt. Of the 1,366 scholarship winners, 24% did not complete the school search process and therefore were not allocated vouchers. Depending on how students were classified, the usage rate for the voucher was 62% to 75% for the first year alone.