{ "id": "2001-novemberdecember-deciding-boundaries", "slug": "2001-novemberdecember-deciding-boundaries", "url": "https://dwillis.github.io/uplink/article/2001-novemberdecember-deciding-boundaries.html", "year": 2001, "month": "November/December", "monthNum": 11, "headline": "Deciding boundaries", "kicker": "CENSUS", "author_name": "John Perry", "author_title": "", "affiliation": "The Daily Oklahoman", "summary": "An Oklahoma newspaper analyzes 2000 census data to explore racial segregation patterns, finding that school district boundaries may be a primary driver of residential segregation, with families choosing neighborhoods based on school quality rather than examining segregation itself.", "keywords": [ "segregation", "census", "schools" ], "topics": [ "Demographics & Census", "Investigative Methods", "Education" ], "technologies": [ "ArcView", "Two Theme Analyst" ], "full_text": "I'll admit up front that our approach to the racial segregation story probably had as much to do with ego as with anything else. With the release of the first detailed race data from the 2000 census in March, we knew just about every newspaper would be writing stories about what the data said about racial segregation. Also, in 1995 we had written a series based on 1980 and 1990 census data showing that Oklahoma City was slowly becoming more racially integrated, but more segregated by income class.\n\nThe map showed an island of white dots in the center, each representing 50 more minority kids, surrounded by a sea of darker dots, each representing 50 more white kids.\n\nMost of our readers won't see the census stories in other newspapers, and probably no more than a handful have even vague memories of what we wrote six years ago. But still, we wanted to do something at least a little different from what we expected to see in other newspapers and from what we had already done.\n\nNo change in data\n\nWhen we looked at the redistricting data released in March, we found that racial segregation hadn't changed since 1990 anyway. The dissimilarity index for blacks and whites was 58.9 in 1990 and 58.2 in 2000. For American Indians, Oklahoma's largest minority group, the index went from 35.9 in 1990 to 34.2 in 2000.\n\nWe also rationalized that there is limited benefit to readers from just describing segregation's existence. We wanted to look beyond that to a possible cause. And the No. 1 suspect was the Oklahoma City school district.\n\nAfter a federal judge ordered busing to desegregate Oklahoma City schools in the 1970s, the district's attendance dropped from more than 70,000 to 35,000 in just a few years. The district's students are now mostly minority and poor, even despite a flurry of inner-city redevelopment in the last decade.\n\nThis suggested that families with children were deciding where to live based on school district boundaries. If that were true, we expected that families with children would more likely reside in racially homogeneous neighborhoods. That should make children more segregated than the general population. The data tended to support this theory.\n\nCalculating block groups\n\nThe actual data crunching for the story wasn't that unusual. For our earlier series on segregation, we had used tract-level data to calculate the dissimilarity index. But when we visited inner-city neighborhoods, we found that a single tract can include separate and unique neighborhoods. So we decided to calculate segregation indexes based on block groups this time around.\n\nWhen we calculated the dissimilarity index - most often described as the percent of a population that would have to move to evenly distribute them throughout an area - we found a small difference between segregation among children and among the total population. But the difference went in the predicted direction every time.\n\nWe also used ArcView 3.2 and the free add-on, Two Theme Analyst, to estimate from block-level data the population changes by school district. The under-18 population inside the Oklahoma City district boundaries grew by 2,610. But that resulted from the loss of 8,309 white (non-Hispanic white alone, in census speak) kids and a gain of 10,919 minority kids. Nearby suburban districts saw similar demographic shifts. Only more distant suburban districts had increases in white kid populations.\n\nMap helps pinpoint\n\nThe most powerful proof turned out to be a metro-area map that used dots to show where minority kid populations and white kid populations had grown. The map showed an island of white dots in the center, each representing 50 more minority kids, surrounded by a sea of darker dots, each representing 50 more white kids.\n\nThe story stayed conservative and said only that, \"Child population shifts in the Oklahoma City area suggest that school district boundaries may be a factor in residential segregation.\"\n\nGiven more time, we would have talked with people who recently moved into or out of different school districts. If the qualitative data from the interviews supported what we saw in the quantitative census data, we would have been more confident saying this is a cause of segregation.\n\nBut this is an issue we hope to revisit next year when we get the long-form census data that will include income and occupation information. And something we noticed makes us eager to do so. Even in the suburban districts where the number of white kids grew, the minority populations grew faster. This suggests that concerns about class, as much or more than about race, may be driving families from the inner city.\n\nJohn Perry can be reached by e-mail at jperry@oklahoman.com", "provenance": { "extracted_by": "anthropic/claude-sonnet-4-6", "extracted_at": "2026-07-05", "verified": true }, "similar": [ { "slug": "2003-may-june-segregation-in-black-white-shades-of-gray", "score": 0.379 }, { "slug": "2005-may-june-the-dividing-line", "score": 0.266 }, { "slug": "1997-october-covering-housing", "score": 0.247 }, { "slug": "1995-april-the-school-of-hard-data", "score": 0.231 }, { "slug": "2001-julyaugust-separate-and-equal", "score": 0.224 } ], "html_url": "https://dwillis.github.io/uplink/article/2001-novemberdecember-deciding-boundaries.html", "json_url": "https://dwillis.github.io/uplink/article/2001-novemberdecember-deciding-boundaries.json" }