This first-of-its-kind analysis uncovered widespread use of suspensions for attendance violations in Arizona and a potential civil rights violation.
Our three-part series offers the most comprehensive look to date at this controversial discipline practice, even beyond journalism and into academia. As a result of our reporting, the Arizona Department of Education acknowledged that it may be time to revisit its discipline policies. The U.S. Department of Education’s assistant secretary for civil rights made one of the clearest public statements yet about her office’s commitment to investigating racial disparities in school discipline under the Biden administration. A researcher said he would like to use our database for an academic paper about school discipline. And a statewide nonprofit, Read On Arizona, discussed our findings in a special convening to combat chronic absenteeism. And a reader successfully challenged her fourth-grader’s suspension, armed with information from our story.
We used student enrollment data available through the Arizona Department of Education’s website.
The only way to find out how often students were being suspended for attendance reasons was to request discipline data directly from school districts. We submitted more than 400 public records requests to Arizona’s public school districts and charter networks. From there, we were able to request chronic absenteeism data by district from the state department of education, which helped us better compare district responses to absenteeism and direct our focus on individual districts.
Using the data we obtained from 89 school districts that suspended students for attendance violations from 2017-18 to 2021-22, we constructed our own database of all suspensions over that time period. Districts provided data in a variety of different ways, all of which had to be standardized and cleaned. Our final database included the violation, the number of suspensions and/or the number of days students were suspended (based on what was available) and the demographic information of the student suspended, namely race, ethnicity and special education status. We also hand-coded whether or not the violation was attendance related, sometimes reading through long narratives of the disciplinary incident to determine that. Our database ultimately included 13,582 rows, logging about half a million total suspensions. The cost to collect the data itself was negligible as very few districts asked for payment to produce the records.
Context about the project:
School districts were deeply reluctant to provide the information we sought. Hundreds flouted Arizona Public Records law and ignored our repeated requests or refused to share the documents, in some cases on the advice of their attorneys. One attorney published an open letter giving school administrators a list of ideas to delay and evade our request, several of which contradicted state open records laws. Among the districts that provided their data, some presented it in messy and unstandardized ways, sometimes with heavy redactions. Many pushed back against providing the documents in an Excel-compatible format.
We circulated a legal memo of our own, pushing back on the lawyer’s public advice and highlighting relevant elements of Arizona’s public records law. We went back to hundreds of districts multiple times seeking the original records and haggled with the vast majority of the approximately 150 school districts that did provide data to get everything we requested, in an Excel-compatible format, which sometimes required citing case law. Ultimately we couldn’t get better than PDF scans for a handful of districts and more than 250 districts never provided complete data.
What can other journalists learn from this project?
The data was very different across districts. Almost every district had its own taxonomy for behavior violations, with a long list of categories specific to that given district. Some had fewer than a dozen general categories (i.e., attendance violation) and some had more than 50 more narrow ones for a given year. To address this, we added a manually coded column to the database that identified whether or not that violation was attendance-related.
Beyond that, districts gave us their data in one of four general formats: 1., (most common) as Excel-compatible spreadsheets where each row described a single disciplinary incident, 2., as Excel-compatible spreadsheets where each row aggregated a given violation type, 3., as PDF copies of spreadsheets organized in one of those two ways and 4., as PDF copies of scanned incident reports. For the first bucket, we could create simple pivot tables to aggregate violation types and put them into our database. For the second bucket, we could put the totals directly into our database. For the third, we had to convert, sometimes manually, the data into Excel for analysis and standardization. And for the fourth, we had to read through the incident reports to create spreadsheets for each district.