Virginia workers have filed more than 1,450 “valid complaints” that employers failed to protect them against COVID-19, government data shows. Only seven states have logged more complaints than Virginia. But workplace health and safety inspectors have cited only about 70 Virginia employers for coronavirus-related violations – and as of Aug. 13, only 25 of them have paid fines. In 24 cases, including instances in which workers died from COVID-19, inspectors did not impose a financial penalty or dropped the fine during negotiations with employers. In the vast majority of complaints, employers were told to correct problems voluntarily, without an inspection.
The Virginia Department of Labor and Industry stepped up enforcement of the state’s rules for protecting workers against COVID-19 after the story was published on Aug. 30.
Before publication, DOLI was citing about four employers a month for coronavirus-related violations. Since publication, the agency has been citing almost six employers per month – even though complaints tapered off as vaccines became available.
Between Aug. 30 and Dec. 30, DOLI posted 23 coronavirus-related citations. The Walmart store in Yorktown, Virginia, for example, was cited for three “serious” violations of COVID-19 standards and fined $22,845. The story had noted that Walmart operations in Virginia had drawn more complaints than any other employer but that none had resulted in a citation.
Workers’ rights organizations praised the investigative project and thanked me especially for writing a sidebar about how to file a complaint with government agencies about health and safety issues in the workplace.
Since the project’s publication, more employees have filed complaints with DOLI and OSHA alleging violations of COVID-19 rules. OSHA has posted more than 115 such complaints against Virginia employers since Aug. 30.
This project required multiple tools and techniques to execute various data-related tasks:
Tabula — To extract text and data from PDFs of workplace citations I received from the Virginia Department of Labor and Industry (DOLI) under the state Freedom of Information Act.
Open Refine, Notepad++ and Microsoft Excel — To clean datasets, including OSHA’s downloadable list of “closed and valid” COVID-19 complaints against employers nationwide (about 60,000 records); a list of employers cited by OSHA for COVID-related violations (obtained under the federal FOIA); and a list of employers cited by DOLI for COVID-related violations (obtained under Virginia’s FOIA). I used these tools to standardize spellings, correct misspellings and split and reformat fields. I also used Microsoft Access to add each employer’s industry sector to the complaints data.
Microsoft Excel and Access — To conduct most of the analysis. For example, I summarized the complaints data, counting the number of complaints filed in each state. I also extracted all complaints against Virginia employers and summarized those records (counting the number of complaints by employer, by industry sector and by date). I also used Excel and TextBlob (a Python library) to conduct a word analysis of the complaints — to see how many involved masks or face coverings, for example. Using Microsoft Access, I joined the complaints dataset and the citations dataset — to identify complaints that led to citations.
Google Sheets and Google Drive — To post all of my data and analysis. I also posted full-text PDFs of citations and inspection reports.
Caspio — To create a tool for the public to search the complaints data.
StoryMapJS — To create a map of “19 Employers Cited for COVID-19 Violations.”
Datawrapper — To create data visualizations, including a chart of employers drawing the most complaints.
What was the hardest part of this project?
I faced three main challenges in reporting this project:
1. I had to submit FOIA requests for the data and documents I needed. OSHA eventually agreed to post the COVID-19 workplace complaints dataset online. However, OSHA and the Virginia Department of Labor and Industry (DOLI) would provide information about employers cited for COVID-19 violations only in response to formal FOIA requests. It took persistence and cajoling to get FOIA officers to respond to requests promptly. When I asked DOLI for the full text of citations and inspection reports, the agency said its policy was to provide paper copies only. It took me four months, and a $21 fee, to get the documents.
2. Data cleaning was a significant challenge. OSHA’s file of COVID-19 complaints contained numerous misspellings and inconsistencies in the employer name field, and I had to split a field that contained both the number of alleged hazards and the number of employees affected. Moreover, the file included only a code for each employer’s industry classification. The citation files also had inconsistencies: In many cases, they listed an employer’s DBA or parent company — not its common name. As a result, joining the complaints file and the citations data required creativity (“fuzzy” joins on names or joins on addresses, for example).
3. It took persistence and time to fact-check and contextualize COVID-related complaints and citations. By email and phone, I contacted every employer mentioned in the story, and I visited three of the cited businesses to seek comment. I received responses from many of the employers and wrote a sidebar about a restaurant that received DOLI’s steepest fine for violating COVID workplace standards. But in several cases, employers refused to comment on COVID-19 complaints and citations filed against them.
What can others learn from this project?
Keep track of your FOIA requests with free online tools like iFOIA (from the Reporters Committee for Freedom of the Press) and MuckRock’s FOIA letter generator. My project required filing 21 FOIA requests (for COVID-19 workplace complaints, for constantly updated lists of employers cited by state and federal agencies, and for the full-text of citations). These tools can monitor the status of each request and remind me when to send a follow-up letter. Make your FOIA requests polite (“I understand that the COVID-19 pandemic has disrupted work schedules …”) but firm (“The law requires agencies to respond within …”).
Document and automate your data clean-up and analysis routines. This was important because I was constantly getting fresh data: The COVID-19 complaints dataset, for instance, was updated about 50 times while I was working on this story. (And the updates didn’t simply include appended records. OSHA often added or revised records in the middle of the dataset.) I kept meticulous notes on how I scrubbed or manipulated each field so that I could repeat each step on new data. I used consistent file naming conventions and audit trails to track how the data had changed. I also used a tool called Workbench to replicate my analysis each time datasets were updated. (Workbench has since been shut down.)
Let readers drill down to the data details that most interest them. People reading my article would want to know about complaints and citations involving their employers, their industry sector or their locality. So I put all of my data online: COVID-19 complaints nationwide; COVID-19 complaints in Virginia; OSHA citations; and citations by the Virginia Department of Labor and Industry. And I created a tool for easily searching the data. I also posted the full text of citations and workplace inspection reports.