In a dark, disaster-driven version of old-fashioned snow days, California schools are sending kids home in record numbers, slashing weeks of instruction in growing swaths of the nation’s most populous state. The cause? Wildfire and other climate-fueled disasters, largely. But gun violence, the widening we gap and other modern emergencies also contribute. State law doesn’t require the lost class time to be replaced — a policy that is changing because of this series. The findings, unearthed by CalMatters education writer Ricardo Cano, 27, were buried in thousands of pieces of school closure data, which Cano analyzed and reported out.
Armed with Cano’s data, California lawmakers have introduced legislation to create a “disaster relief summer school” program to recoup the burgeoning school time being lost to natural disasters in California. It’s difficult, Cano explains, because of the extensive local control school districts are guaranteed under the state constitution. The series also is informing debate over a statewide school bond measure, placed after the series on the March 2020 ballot, to address acute maintenance issues — and to correct inadequacies that were giving poor communities less access than rich ones to state school bond money. It is also underpinning legislation to shore up air filtration and electrical systems at schools to prevent closures for smoke pollution and public safety power shutoffs. And at a time in which a youth climate activism movement has captured the world’s attention, the reporting has galvanized a campaign by students and teachers to push education leaders and institutions to publicly acknowledge how climate change is harming the next generation in California.
We built an interactive app, map and several databases, including the comprehensive tool featured in the series. The main database was built using 17 years of school closure records obtained through the California Public Records Act from the California Department of Education that came in Excel format. Others included a database, compiled and built by Cano, of local school bond spending throughout California dating back to 1998; a Microsoft Access database of large-scale drops in attendance reported to the state over two decades; and real-time announcements by school districts statewide of closures for wildfire or related issues during the first three months of the 2019-2020 school year. Cano spent $79 on a subscription-based online software (Gridoc.com) that he used to join multiple datasets along the reporting and data-building process. The source of the main data was attendance waivers California school districts submit to the state when they have to temporarily close down schools for emergencies. The information on the forms, which details key information such as which schools closed, why they closed, and how many instructional days they closed, is self-reported by districts. We primarily used Excel and Google Sheets to clean and build our main Disaster Days database. D’Agostino used PHP to write a script to reshape the state’s data and records we received so that we could begin building and analyzing emergency school closures at the school site level. D’Agostino and Al Elew used the Flask Python web framework – as well as an assortment of tools, including QGIS, JQuery and JSON – to build and house the interactive database and visualizations that readers could use to search our data. Part of the cleaning and building process also involved some webscraping.
What was the hardest part of this project?
To our knowledge, no news outlet prior to us had reported or analyzed this data. Climate change is a big story, and wildfires often break news, but the cumulative impact of climate-driven fires on core institutions such as schools and on kids over the long term is less well reported. CalMatters had to work on a shoestring; we are tiny. The whole project was done by one education reporter (working weekends and nights, while covering a beat), a summer data intern (who also lived at the office for months and stayed on for weeks, to finish), one developer and one line editor. The original data for the main database was riddled with typos, duplicate records and inconsistencies. Because the information was self-reported by schools over nearly two decades, and because California is massive, many unique school sites had multiple name configurations (example: Calaveras High, Calaveras High School, Calaveras HS) or had since changed school names altogether. The initial data came at the school district level (with a field that included a comma-separated list of school sites) and had no relational field we could use to join the databases we needed to build our database of school closures. D’Agostino reshaped the data using a PHP script so each row began to reflect a closure per school site. After several stages of cleaning we developed our own system of codes to create a relational field between the data we had on school closures and the 17 school years’ worth of enrollment data we assembled. After doing a direct join using Gridoc, we still had to manually pair enrollment data for more than 1,000 records. To help account for duplicates in the original state data, we paired every record with a “CDS Code,” a unique, 14-digit code designated to every California public school.
What can others learn from this project?
We learned at least two lessons: Sophisticated, informative and groundbreaking reporting doesn’t necessarily require a massive staff, and there’s no replacement for on-the-ground, shoeleather reporting. Disaster Days stemmed from data, but it didn’t stop there. As we built our database and analyzed the numbers, we found gripping, untold stories on the cumulative toll fires and disasters have taken on students, teachers and public schools. Cano and photographer Anne Wernikoff hit the road to report how schools have become stretched to the breaking point by disasters that are nothing like the quaint, traditional snow days. Thousands of California students have lost homes to fires in recent years, we found, and the annual loss of days and weeks of instruction have had significant academic costs for many schools. At a time when natural disasters worsened by climate change have affected millions of schoolchildren worldwide — including in developed nations like the U.S. — we believe our effort provides a framework for reporting and data analysis that could be replicated by other journalists to tell an increasingly important story of human impact. We also learned that there is value in creating accurate and reliable data and analysis that quantifies : The data we published spurred a state senator to propose a new policy framework that would allow schools to make up days lost to now-frequent fires and disasters.