A Critique of New Jersey's COVID-19 Data Transparency

In this post I explore how New Jersey compares to neighboring states in terms of the granularity, format and accessibility of raw data about COVID-19 cases and deaths made available to the public during the pandemic’s outbreak in 2020. I conclude that New Jersey is failing to adequately publish raw data on COVID-19 cases for use in research and journalistic reporting, falling short compared to others in the tri-state region.

By “raw data” I mean structured data, either in CSV, Excel, an API, or some other machine-readable format published by a state public health authority and not locked to a dashboard or website that cannot be exported for analysis that provides aggregated counts of COVID-19 cases and deaths at the state or sub-regional level.

Making this data available is important because it allows journalists, researchers, and others wishing to study COVID-19 to get a better understanding of both conditions in their communities and the decisions made by government in response to the pandemic.

How New Jersey Compares to Its Neighbors

With that definition in mind, just what does New Jersey have to offer for COVID-19 data?

Not much.

Unlike three neighboring states - including New York, Connecticut and Pennsylvania - the homepage of New Jersey’s state open data portal has not been updated to highlight the state’s collection of COVID-19 data available for download to the same extent as its neighbors have done so. In fact, the New Jersey open data portal really has not changed much at all since the beginning of the year aside from the addition of a small icon. Not updating the state data portal to strategically highlight the availability of raw COVID-19 data was a missed opportunity for state officials to both showcase the value of the state’s open data initiatives as well as to build trust with the public by prioritizing transparency in the state’s response to the pandemic.

Instead, to find out what types of COVID-19 data are available through New Jersey’s portal, one must manually search for it. Even then, the results leave much to be desired.

Links ≠ Data

Using an open data portal - and all of the advanced features that it provides for sharing and visualizing data - as a simple directory of links to other websites that do not provide downloadable data essentially defeats the purpose of adopting an open data portal as a clearinghouse of government data in the first place, since the goal of an open data portal is to facilitate the sharing of data, not a simple directory of links.

Links to external websites that don’t provide data, however helpful they may be, fall short of advancing the wider purpose of open government data.

Disabled Data Downloads

New Jersey’s COVID-19 data dashboard is actually a mishmash of two popular technologies used for publishing online dashboards. While the main dashboard interface is built with ArcGIS StoryMaps, several of the core views of the dashboard are actually built with Tableau.

One of the most frustrating and arbitrary barriers to accessing the state’s COVID-19 data results from attempting to actually download some of the data that is presented on some dashboards. New Jersey’s Tableau-based dashboards provide an option to download the contents of the dashboard to save a local copy to your device. Unfortunately, New Jersey explicitly disabled the option to download copies of the underlying data presented in the state’s COVID-19 dashboard.

Sure, you can download a copy of the dashboard in PDF, PowerPoint or as a JPEG image, but not in the format that it was stored in when used to produce the maps, statistics and visualizations shown by it. This makes it nearly impossible for any substantive analysis of the state’s COVID-19 cases and deaths using the state’s official data.

This is one of the most frustrating barriers to accessing New Jersey’s COVID-19 data due to its seemingly contradictory basis and arbitrary imposition. If the data presented in this dashboard is fit for public release and consumption by way of the dashboard, then surely the same data must also be suitable for the public to download?

Lack of ZIP Code-Level Data

Another frustrating shortcoming of the state’s COVID dashboard lies in the lack of availability of ZIP code level data. In addition to infrequently updating this segment of the dashboard (monthly), many New Jersey towns lack data because the NJDOH refuses to publish these numbers for any ZIP code with less than 20,000 residents, citing privacy concerns.

The barriers to accessing ZIP code-level COVID data in New Jersey significantly limited the public’s ability to study more localized trends in cases, especially in less populated regions of the state.

New Jersey is Behind the Pack

To understand how far behind New Jersey is in publishing open data concerning the pandemic, we need only to contrast the state’s paltry availability of data with the efforts of our regional neighbors in the greater tri-state area.

New York

New York City provides multiple CSV files updated with daily counts of COVID-19 cases and deaths within the city’s five boroughs via GitHub. They’ve done a great job with keeping these files updated and the level of data they provide should be a model for other states and localities to follow. The data files published by NYC include counts of cases and deaths by ZIP code, along with other files tracking rates of COVID-19 in hospitals and broken out by race and ethnicity.

Since the city releases their public dataset via GitHub, this also externalizes any costs for infrastructure and bandwidth associated with making the data available to the public. This simultaneously saves the city health department money while providing a more convenient way for developers and researchers to access the data thanks to the ease of use and ubiquity of the GitHub platform.

Connecticut

Connecticut provides data on COVID-19 via their state open data portal. That state went as far as creating a separate curated collection of COVID-19 data on their website, making this data even easier to access. While Connecticut falls short of providing data at the ZIP code level as New York City does, they do provide a data file with COVID-19 case counts by municipality.

Pennsylvania

Pennsylvania’s official COVID-19 dashboard is also built with some of the same underlying technology that powers the dashboard created by the New Jersey Department of Health, ArcGIS StoryMaps, yet it provides a far greater level of openness because it explicitly facilitates the download of raw data.

The City of Philadelphia provides the most granular COVID-19 data, providing cases and deaths by ZIP code as a flat CSV file or API, but also in geospatial formats including GeoJSON and SHP.

Dashboards Alone Are Not Enough

A dashboard, like the one created by New Jersey or others, should not be the be-all and end-all data strategy for the state’s COVID-19 response. Dashboards can be helpful and have a role to play in informing the public, but publishing open data would better serve the public interest.

At this point in the pandemic, a dashboard should be considered the bare minimum. Dashboards that do not make some form of the underlying data that they present available for analysis fall short of providing full transparency regarding the state’s response to COVID-19.

Fixing the lack of COVID-19 data in New Jersey does not solely have to be an effort from the public sector. Volunteer efforts like the COVID Tracking Project and efforts by media outlets such as the New York Times to compile data obtained by their journalists in machine readable format have helped to fill in some of the gaps in COVID-19 data availability, but governments like New Jersey can and should increase efforts to publish open data regarding the COVID-19 pandemic.