Dispatches

Each year, Allegheny County publishes an update to the County crash dataset, joining together data published by PennDOT (the Pennsylvania Department of Transportation) after filtering it down to vehicular crashes that took place in Allegheny County in the previous calendar year. There are individual tables for each year, with record-level data going back to 2004. This dataset has now been updated with the crash data for 2024.

There have been some changes to the PennDOT data format over the years, and the dataset has accommodated this by standardizing the published set of fields from 2017 onward. All the records are compiled into a single cumulative table for ease of analysis.

It's interesting to study the total number of reported crashes in Allegheny County by year from 2004 to 2024.

Aside from an abrupt dip in 2020 (a decrease in crashes which started in March of 2020 and could be attributed to decreased travel during the first year of the pandemic), countywide crash counts have decreased year over year since 2016. (The statewide variation in crashes by year (not shown here) exhibits a similar pattern to the countywide trend.)

These tables have almost 200 columns, which provide details about each crash, including how many vehicles/bikes/pedestrians were involved; the weather, condition of the road, and the lighting at the crash scene; the latitude and longitude of the crash site (though note that they are in the nonstandardly named `DEC_LAT` and `DEC_LON` fields); and whether any driver was using a cell phone, had been drinking, ran a stop sign, or ran a red light. With this data, you can ask and begin to answer questions like "What is the relationship between the speed limit of the road (or whether any speeding was involved in the accident) and the severity of the accident?".

The Allegheny County Crash Data dataset also links to a PennDOT page where you can download the raw data, and a primer from PennDOT on the crash data and how it is assembled.

Yesterday, we officially launched the Parcels N’At Property mapping tool in a media event at the City County Building. We were joined by Mayor Gainey and other friends from the City, County, and Turtle Creek Valley Council of Governments, who shared how they have been using the beta-version of the tool. See more about it on the press release, and let us know what you think about the Parcels N’At tool itself

The City of Pittsburgh recently opened applications for its Food Justice Fund. Proposals are due April 14. Grants of up to $75,000 are available to increase food security and improve food systems. https://engage.pittsburghpa.gov/food-justice-fund

A question on the application addresses community needs, so I wanted to provide a few links to places where relevant data can be found. Here’s the question that is asked on the application: 

“What communities does this project serve? What community needs will this project address? For example, low-income seniors with food insecurity or a neighborhood without access to nutritious food markets.”

Data from the Census Bureau can describe the community in terms of size, demographics and needs. Colleagues here at the University Center for Social and Urban Research recently updated its community-level reports containing data from the American Community Survey. 

https://ucsur.pitt.edu/reports/census-reports

Allegheny County's Department of Human Services has developed an indicator of need in all of the County's communities by bringing together different datasets that show communities where residents face comparatively high socioeconomic barriers. The Community Need Index is a composite indicator that is based on several indicators, including poverty, single parent households, unemployment, low levels of internet access, homicide rates, low levels of educational attainment, and fatal overdoses.  

One place to find data about food access is the USDA’s Food Access Research Atlas. It uses data on income from the American Community Survey and data on the locations of grocery stores and supermarkets to highlight areas of the country where people have low incomes and low access to places that sell healthy food nearby. Please note that the data on the map has not been updated since 2019, and data about smaller retailers, farmers markets, community gardens/refrigerators, and food pantries are not used to calculate access to food. Also be aware that the tool may not take our rivers and hillsides into account when making calculations about food access. 

https://www.ers.usda.gov/data-products/food-access-research-atlas

A similar tool dating back to 2020 is our community asset map. https://assets.wprdc.org/ It shows data on food retailers, restaurants, etc, but is outdated. 

To produce it, we pulled data from a number of different data sources, including data we receive from Allegheny County. This updated data is available on our open data portal. 

We currently maintain data on:

• Food facilities inspected by Allegheny County’s Health Department

  https://data.wprdc.org/dataset/allegheny-county-restaurant-food-facility-inspection-violations

• Weights and Measures inspections by the County (businesses with scales and gas pumps are captured, among other things)

     https://data.wprdc.org/dataset/allegheny-county-weights-and-measures-inspections

• Farmers Markets

     https://data.wprdc.org/dataset/allegheny-county-farmers-markets-locations

• An older but still useful dataset is Grow Pittsburgh food gardens (we are in need of a data update), but they may have more up-to-date data available on their website.

     https://data.wprdc.org/dataset/grow-pittsburgh-food-gardens

• The Greater Pittsburgh Food Bank also provides data on food pantries and other places on their website.

      https://pittsburghfoodbank.org/find-food/get-help/

• The United Way’s 211 system is a source for data about needs. Administrative data about helpline requests is available by Zip code, month, and request type on their dashboard - select “food” as the taxonomy group to learn more about how many people called for help finding food. 

      https://embed.domo.com/embed/pages/yol6W

Keep in mind that you and people in your community may be the best source of information. If you’re already operating a food assistance program, reporting the number of people and meals you serve, who is served, the kinds of help you provide, where it is provided, and the monetary value of the support are ways to quantify your efforts. You can also collect and share stories from the people you serve about how your work has made a difference.

Hopefully, these sources of information can be helpful in your application for funding. Please reach out if we can be of assistance, at wprdc@pitt.edu.

We just updated our project page to include details on work we've been involved with over the past several years. Our search for a new Center Director here at UCSUR is moving forward, and wanted to make sure the candidates had a chance to learn more about us. 

Fish Fry season is upon us and once again saying a big thank you to all of the organizers and volunteers that do the thankless work of updating the Pittsburgh Fish Fry Map every season. 

Open Letter to the City Paper

The Western Pennsylvania Regional Data Center (WPRDC) was referenced in an article yesterday (January 29, 2025) about the Pittsburgh City Planning Commission’s inclusionary zoning meeting in the City Paper. I’m writing to provide additional context about WPRDC, the permit data cited in the article, and share some of the challenges of measuring housing units through administrative data. 

The WPRDC is a partnership between the University, Allegheny County, and the City of Pittsburgh designed to make civic data available and useful. Managed by a team at the University of Pittsburgh’s University Center for Social and Urban Research since its launch in 2015, it has provided access to hundreds of datasets shared by the City, County, and several other publishers as open data on its website. Staff with the project build tools enabling people to access data, and provide training and technical assistance to people seeking to learn more about their community.  

We disagree with the way that the WPRDC was characterized as a source of data in the City Paper article. 

Had we been consulted by people mentioned in the article, we would have provided general advice when it comes to working with administrative data. For example:

• Administrative data isn’t created for your purpose, unless your purpose is administering the program that created the data. You’re likely using data for something other than its intended purpose.
• Understand the context of the data you’re working with in order to use it appropriately. As a consumer of data, it also is fundamental that you pay attention to  context so that you can evaluate how others are using data. To capture context, we like to document the motivation for collecting the data, its composition, collection processes, transformations made to the data, who uses it, how it's shared, and maintenance of data and systems used to create it.
• Show your work. In order for people to trust your numbers, you need to talk honestly about data contexts, and be transparent about the choices you have made when working with data.
• Data is not the full truth. No matter how much we try, you just can’t quantify human experience or the experience of living in a city.
• Data is not neutral. Humans are biased and therefore data is biased. Data or its absence also reflect power structures in place at the time it was collected.
• Data is not always accurate. There can be many different reasons for a lack of accuracy or precision, including inadequate resources or errors in technology, reporting burdens, data entry errors, and even power dynamics (where people with power may benefit from inaccurate data to control the narrative).

There is no dataset of new housing in Pittsburgh. Permit data alone isn’t going to provide a complete picture of development activity. We would have suggested that anyone looking to create a dataset of new housing start by combining several of the following administrative datasets:

• City of Pittsburgh Permit data are created by the City of Pittsburgh to regulate construction activity. Permit records can be used to identify where at least some new housing is built, but would like to raise awareness of these potential issues: 1. Not every permit record for housing development may include a count of units; 2. Not everyone who is required to file a permit actually obtains one. Without proactive enforcement, it’s very likely that permit data under-reports construction activity; 3. Aggregating data might be required to present a full picture of housing development project by project. Developments may span multiple parcels, and people may file permit applications for each parcel separately. Phased development projects may also result in several different permit applications.
• Allegheny County Property Assessment data includes data on the year a building on a parcel was built, but this data may be missing for larger residential buildings and parcels with more than one building. Older buildings converted to a residential use may also not be captured, as the “year built” field wouldn’t change. Measuring changes in parcel land use over time can be one way of capturing data on new construction lacking data on the year it was built. On the dataset link above, you’ll also see an example of how we documented the context about this dataset in a user guide.
• Other permits can also be used to identify construction or rehabilitation activity: Allegheny County Asbestos Permits are required for asbestos abatement and Plumbing Inspections data are available for properties meeting the Health Department’s requirements for a plumbing inspection. The City of Pittsburgh also shared data on requests to close streets, and these are often obtained as part of the construction process for new housing development.
• Affordable housing subsidies: Much of the data about existing affordable housing can be tracked through project subsidies provided by funders. In Allegheny County, we have developed the HouseCat tool with partners in the Housing Preservation Working Group to assemble data from HUD, the Pennsylvania Housing Finance Authority, and other sources to provide a near real-time look at subsidized property. Note: You must apply for access to this data tool through the Housing Preservation Working Group.
• Construction bids: Private-sector data brokers track construction bidding opportunities. Construction companies have relied on this data to find project opportunities for decades.
• USPS Address Counts. In partnership with HUD, the USPS shares data about vacant addresses at a Census tract level every quarter. Mail carriers collect data on properties not receiving mail. In addition to vacant data, the dataset also includes the total number of residential mailing addresses in a tract. While not at a parcel level, growth in the number of addresses in a tract over time can provide an indication of where housing development may be taking place. 

Some housing may not be captured by counting units. Beds in shelters and other group quarters facilities (such as college dormitories) would have to be counted in other ways. The Census counts or estimates group quarters, but other approaches for using administrative data could be applied here. 

After this data is assembled, it’s also vital to “ground truth” it. Local knowledge provided through on the ground assessments or interactions with community leaders can ensure that you’re using local knowledge to accurately capture data on as many developments as possible. 

We are always happy to spend time talking with anyone that is looking for help working with data about our community in our role as a data intermediary. Please send us an email (wprdc@pitt.edu) if you ever need help or just want to “talk data.” 

Reflections on my POSE I-Corps training experience

Last year, we were awarded a Phase II grant from the U.S. National Science Foundation’s Pathways to Enable Open Source Ecosystems (POSE) program. In this two-year project, we will be working to strengthen the CKAN open source software ecosystem by broadening community participation, increasing adoption, updating the business model, and creating performance measures that can enable future conversations about the ecosystem. We use CKAN to power the WPRDC open data portal, and it’s used by several hundred open data and open society initiatives around the world. You can learn more about CKAN on the website, and our Phase I ecosystem discovery work can be viewed on the project webpage. We are grateful to be working in partnership with the dedicated community members that help to build and maintain CKAN, along with our project partners at datHere and Eleanor Mattern from the University of Pittsburgh’s School of Computing and Information.

The NSF required our project team to participate in a month-long training through their I-Corps program in September and October. We participated in several days of workshops and meetings during the first week, followed by weekly workshops and check-ins with our assigned mentor over the next three weeks of the program. In addition to attending the workshops, we also were required to set-up and conduct 60! interviews with current and prospective users and others in the ecosystem before the conclusion of the training. 

Thanks to generous friends and colleagues, I was fortunate to be able to quickly schedule over 35 interviews in response to one desperately-worded email sent to the listserv of the National Neighborhood Indicators Partnership, a community of civic data intermediaries in the U.S. I leveraged personal and organizational relationships with many in the network that I’ve been a part of since 2008, and wanted to thank all of the long-time and new friends based largely in academic and other nonprofit organizations in this community of practice that found a few minutes for a conversation with me. These conversations largely focused on what it would take for them to adopt a product like CKAN in their work. 

Here are my key takeaways from those conversations:

• Interviewees were aligned with the values and value proposition of open source.​
• Users in academic and nonprofit sectors often had flexibility to choose and install software of their choice (unlike many working in public-sector organizations).​
• Word of mouth is vital to their software discovery processes.​ Many people scheduled an interview with me so that they could learn more about CKAN from a current user in their network.
• Potential adopters would find it less-daunting to install and use CKAN if they could do it as part of a community. ​Going forward with POSE, it makes sense for us to leverage relationships in existing learning communities and peer networks (like NNIP) to encourage adoption.​
• Some of the people we spoke with were contributors to open source ecosystems and reinforced the importance of feeling welcome and encouraged.​

Our project partner Joel Natividad at datHere spoke to many public-sector organizations and others already in the CKAN ecosystem, and published a blog with his experiences late last year. 

I have some additional thoughts about my I-Corps POSE experience:

The training program was designed to help grantees identify commercialization opportunities for researchers, so applying it to the POSE program created some challenges for many of the teams, as the focus of open source software usually doesn’t involve commercializing new innovations. The training sessions and materials helped us to think about markets, competitors, barriers to adoption, and value propositions through the business model canvas activities, but materials related to monetizing the product missed the mark with POSE grantees like us. Had more time been spent adapting the I-Corps materials to the specific challenges in strengthening existing open source software communities, we would have found the training more-relevant to our context. 

The high interview requirement and short time frame made it impossible for our team to synthesize what we were hearing to our satisfaction. We were also unable to be strategic in scheduling and sequencing the interviews. As a result, we didn’t get as much value as we could have from the conversations. We also lacked the time to speak with people outside of our familiar networks, and I especially wish I had time to connect with people outside the U.S. The experience felt very transactional, and the focus on the number of interviews completed was reinforced by “leader board” visualizations showing the number of scheduled and completed interviews by each grantee team. I became focused more on how many interviews we had completed relative to our peers rather than whom we were talking with or what people were saying. I wonder if a dashboard showing the amount of insight or knowledge gained through our interviews would have changed my mindset in a more-beneficial direction. 

From the start, it felt as if we were being strongly encouraged to focus all of our interviews on the theme of product adoption, even though a large part of our strategy under POSE involves encouraging CKAN users to become contributors to the ecosystem. It wasn’t until after several other teams pointed out that they might also like to focus on the existing user communities for their open source product that one of the instructors encouraged us to think about “the ecosystem as a product.” I’d encourage the team organizing the next round of POSE training to embrace this framing so that participants would be encouraged to schedule interviews with people already involved in their ecosystem from the start. We found the 40+ conversations with people in the ecosystem in our Phase I work incredibly valuable, and can enthusiastically make the case that future training participants should be encouraged to find time to talk with more people in their existing ecosystems. 

Overall, participating in I-Corps was still a valuable experience, but I question if the experience was “worth it” in its current form. I would have found the training far more-useful if the materials were tailored for open source software projects like some of our Phase I training workshops were. We especially found the workshops designed around the Community Participation Model of the Center for Scientific Collaboration and Community Engagement to have been extremely valuable in designing our Phase II work. I hope that future I-Corps training series are adapted to the needs of people working to strengthen open source ecosystems so that participants can maximize the value from the time they invest in the training and interviews.

We’ve been thinking about how to reinvigorate our communications. Prior to the pandemic, we were using an email newsletter, blog posts, and twitter to help people stay connected to our work. During the pandemic, the demands of the response to the crisis caused us to drop the newsletter, which took several hours per issue to write, format, and send, and instead invested that time in building authentic relationships and solidarity with people and communities working to respond.  Once twitter became something different from what it was, and disinvested in many of the systems that worked to provide for a safe and harassment-free space, we stopped using that as a platform. We still infrequently write blog posts, but reserve these to specific topics, such as the release of a product or event recaps. 

We have found value in a “weeknotes” format that provides a place for personal thought and reflection. There’s no set format and no set length, and the process of writing them seems to benefit both the writer and the reader. If you’d like to learn more, we found this post that effectively summarizes the format. https://medium.com/deloitte-digital-connect/how-to-write-weeknotes-for-your-digital-project-26a6bfef38d6

We want to manage expectations around frequency, so will be calling our version “dispatches” as we try and be better about sharing our work

Our dispatches could be a place where:

• We write about what we’ve been up to;
• Share things we’ve found interesting;
• Make announcements about data and data products, such as new tools, datasets or temporarily broken data feeds;
• Promote events and activities;
• Publish reflections related to our work or the broader contexts in which we operate. 

For an example of weeknotes that we value and appreciate, please see the Connected by Data project. In their weeknotes, at least one team member publishes them weekly. https://connectedbydata.org/blog/#weeknotes

We’re also interested in exploring how we might use other platforms like newsletters and social media in concert with weeknotes/dispatches to reach a broader audience. For example, Connected by Data uses social media to point to new weeknotes as they’re published, and their email newsletter serves as an index of the weeknotes published in the past month. 

We’ll give this a try. We hope you find them valuable, and look forward to your feedback.