{
  "id": "2000-april-impact-of-choice",
  "slug": "2000-april-impact-of-choice",
  "url": "https://dwillis.github.io/uplink/article/2000-april-impact-of-choice.html",
  "year": 2000,
  "month": "April",
  "monthNum": 4,
  "headline": "Impact of choice",
  "kicker": "EDUCATION",
  "author_name": "Griff Palmer",
  "author_title": "",
  "affiliation": "The Daily Oklahoman",
  "summary": "An investigation into Oklahoma City's choice school movement reveals that these schools disproportionately serve white, wealthy students from educated neighborhoods, while neighborhood schools become increasingly concentrated with minority and poor students, coinciding with declining test scores.",
  "keywords": [
    "choice",
    "schools",
    "segregation"
  ],
  "topics": [
    "Education",
    "Data & Databases",
    "Investigative Methods"
  ],
  "technologies": [
    "FoxPro",
    "Excel",
    "SPSS",
    "Atlas GIS"
  ],
  "full_text": "Impact of choice\n\nIn the past five years, the Oklahoma City school district has, with growing enthusiasm, embraced the \"choice school movement.\" The growth of the movement has coincided with the dismantling of the district's court-ordered busing system and a return to neighborhood schools, many of them with student bodies that are 90 percent or more black.\n\nOklahoma City's choice schools are widely viewed as unqualified successes. But dissenting voices have argued that the establishment of choice schools, catering to elite students, imposes a cost on other students.\n\nEducation reporter Bobby Ross and I decided to put a yardstick to the debate: Are choice schools having an impact on other schools in Oklahoma City? How much of an impact? What are the implications? Our measurement tools included FoxPro, Excel, SPSS and Atlas GIS.\n\nWe found that students attending Oklahoma City's choice schools are disproportionately white, and are drawn disproportionately from the city's wealthiest, best-educated neighborhoods. We also found that neighborhood schools have greater concentrations of minority students and poor students than live within those schools' attendance boundaries. This bleeding off of talent from the wealthiest neighborhoods has coincided with a marked decline in standardized test scores among students in neighborhood schools.\n\nData used\n\nAt the heart of our analysis was a data file containing the home addresses and school assignments of Oklahoma City's 40,000 students.\n\nSeveral fellow journalists expressed surprise that we were able to get the school district to give us this data. There is ample room, though, under the Family Educational Rights and Privacy Act of 1974, for a school district to release student directory information.\n\nThe school district also provided us boundary files for neighborhood school attendance zones. We geocoded students' addresses and then used Atlas GIS to assign geocoded addresses to corresponding school attendance boundaries.\n\nThen we were able to measure how many students were leaving their neighborhood school zones to go to choice schools. But we also wanted to know something about the economic status of those leaving and those staying.\n\nEconomic status\n\nWe first tried to link student addresses to county property tax data, intending to use property values as proxy measures of student economic status. We discovered problems with this approach, though. We felt that we would have to develop separate measurement scales for those who lived in rental property and those who lived in their own homes. However, the tax rolls didn't identify rental properties or provide rent cost data.\n\nWe abandoned this approach, swallowed hard, and turned to our marketing department for a piece from its collection of Claritas data: estimated 1999 median household income and 1990 educational attainment by census block group. (See the Jan./Feb. 2000 issue of Uplink for a more detailed discussion on measuring student economic status).\n\nWe used Atlas GIS to assign each student's address to the census block group within which it fell. Then, we linked the student address data to the Claritas income and education data, joining on the block group field.\n\nBlock group-level income and education data served as rough indicators of individual students' socioeconomic status. This method allowed us to make observations based on the socioeconomic status of the block groups in which students lived, but not on individual students.\n\nWe used SPSS to calculate income quartiles for all middle and high school students, based on the block group in which they lived.\n\nWe assigned an income quartile ranking to each student in the data set. Then, we used FoxPro to create three tables:\n\n1.) One grouping the data by middle school/high school income quartile within each school.\n2.) One grouping by middle school/high school income quartile within each attendance zone.\n3.) One grouping by income quartile among students living in each attendance zone but enrolling in a choice school.\n\nWe joined the three tables on school name. The result was a table that contrasted the number of students in each quartile living within a school attendance zone with the number of students actually attending the school and the number going to choice schools.\n\nOpening the table into Excel, we calculated percentage of each school's enrollment, attendance zone and choice school out-migration falling into each quartile.\n\nWe found that in 11 of Oklahoma City's 13 neighborhood middle and high school attendance zones, a disproportionately low number of students from the lowest income quartile, and disproportionately high number of students from the highest income quartile, had left neighborhood schools for choice schools.\n\nSchool lunch data reinforced this observation. We found that the opposite was happening in neighborhood schools.\n\nOther measures\n\nVirtually all of the income disproportionality between neighborhood schools and their attendance zones was attributable to the out-migration of students to choice schools. A tiny portion was attributable to out-migration to other neighborhood schools.\n\nHowever, we were unable with our data to measure how many students were leaving neighborhood school attendance zones for suburban schools and private schools. While we weren't able to distinguish between students recaptured from private schools and suburban public schools, we were able to identify those students enrolling in Oklahoma City schools from addresses outside the district. (A disproportionate number of high-income students from out of district were getting into choice schools.)\n\nWe used similar methods to analyze household education levels within neighborhood schools' attendance zones, and to contrast them with the household education levels of the block groups in which out-migrating students lived.\n\nTo get an idea of racial proportionality, we compared the racial makeup of each choice school with the racial makeup of the nearest neighborhood school serving the same grades. In virtually every instance, we found that the choice schools, while having diverse populations, had disproportionately more white students.\n\nWe also analyzed district admission statistics and found that a greater percentage of minority applicants than of white applicants were being denied admission to those choice schools that used selective admission criteria.\n\nThese measurements provided a framework on which Bobby hung anecdotal evidence gathered through countless hours spent inside choice schools and neighborhood schools, and through innumerable interviews and focus-group sessions.\n\nDid we prove that neighborhood-school students are being harmed by the choice school system?\n\nNo. But we did create a replicable set of measurements against which to discuss studies showing that students in schools with greater concentrations of poverty tend to perform more poorly on standardized tests.\n\nGriff Palmer can be reached by e-mail at gpalmer@oklahoman.com",
  "provenance": {
    "extracted_by": "anthropic/claude-sonnet-4-6",
    "extracted_at": "2026-07-05",
    "verified": true
  },
  "similar": [
    {
      "slug": "2005-novemberdecember-destination-failing-schools",
      "score": 0.25
    },
    {
      "slug": "1998-august-car-cooking-raw-numbers",
      "score": 0.246
    },
    {
      "slug": "1995-april-whats-affecting-test-scores",
      "score": 0.207
    },
    {
      "slug": "2000-september-school-test-scores",
      "score": 0.206
    },
    {
      "slug": "2001-novemberdecember-deciding-boundaries",
      "score": 0.195
    }
  ],
  "html_url": "https://dwillis.github.io/uplink/article/2000-april-impact-of-choice.html",
  "json_url": "https://dwillis.github.io/uplink/article/2000-april-impact-of-choice.json"
}