Some projects teach you a method. This one taught me a profession.
Early in my career, I became part of PennDOT's Act 44 transit data review effort, a statewide oversight initiative focused on verifying the accuracy of the data used to allocate operating subsidy funding. The work centered on the formula variables that mattered most to transit agencies and to the Commonwealth: revenue vehicle miles, revenue vehicle hours, total ridership, senior ridership, and ADA service statistics.
These were not abstract compliance exercises. PennDOT needed stronger confidence that the data behind transit subsidy reporting could be trusted, especially after statewide concerns about whether reported service and ridership were adequately supported.
Before the reviews became a repeated field assignment, I helped develop the Act 44 Data Quality Control Plan template, a standardized framework intended to help Pennsylvania transit agencies verify the data used in Act 44 Section 1513 funding allocation. That template later became the guiding framework used to conduct the Act 44 Data Verification Reviews.
A Repeatable Process, Never the Same Process
The work was part audit, part field investigation, part operations research, and part public-sector trust-building. We did not simply accept certified numbers at face value. We traced reported values from dotGrants submissions back to source data, agency procedures, and field practices, then tested whether the reported values could be independently reproduced.
The process repeated across agencies, but it was never the same process. Each transit system had its own operating environment, tools, constraints, and habits. At one agency, the key records might be farebox exports and video logs. At another, they might be driver sheets, dispatch spreadsheets, odometer readings, and payroll comparisons. And at another agency, they might be Ecolane reports, RouteMatch outputs, Avail data, paper tally sheets, fare media, or contractor invoices.
That variety is what made the work such an adventure. I had the rare opportunity to do the same kind of job repeatedly, but in different places, with different people, different systems, and different problems to solve. The reviews took me across Pennsylvania, including places such as Indiana, Hazleton, Warren, Altoona, Johnstown, DuBois, Meadville, State College, and New Castle.
The Heart of the Work Was Traceability
I worked with PennDOT's Bureau of Public Transportation, the consultant team, and staff at each transit agency who understood the day-to-day reality behind the numbers. The onsite and remote reviews were designed to understand how agencies collected, compiled, checked, and reported their Act 44 data, then to confirm whether those procedures were accurate, adequate, and actually being followed.
For revenue vehicle miles and revenue vehicle hours, I reviewed how agencies derived service statistics, worked with staff to reproduce calculations for fixed route and demand response service, compared reported numbers against source records, and tested whether exceptions such as detours, missed trips, weather delays, route deviations, deadhead mileage, and service disruptions were properly handled.
For ridership, I looked at the data and technology used to derive passenger statistics, then analyzed trips by fare category, route, driver, day, hour, and sometimes minute. When the raw data was not ready for analysis, the work involved reconstructing and standardizing datasets, parsing date and time fields, interpreting farebox key definitions, converting proprietary outputs, and sometimes normalizing paper logs or scanned records into usable review files.
Senior ridership became one of the clearest examples of why data quality required judgment. Because senior trips were free, they did not have a clean payment record tied to that trip. That made verification more complicated.
The goal was not just to find errors. The goal was to determine whether an agency's reporting process was traceable, reproducible, defensible, and supported by adequate primary and secondary verification procedures.
Where Technical Work Met Human Systems
The reviews required judgment. Statistical anomalies, operational implausibilities, weak documentation, and unexplained patterns could lead to targeted verification. Where no obvious anomalies appeared, random vehicle days or records could be selected for independent validation.
Video reviews were one of the clearest examples of that judgment in practice. A selected vehicle day could turn into a slow reconstruction of what actually happened: matching camera footage against driver records, farebox events, schedules, timestamps, trip patterns, and the reported numbers. The point was not that video alone answered the question. The point was that it helped test whether the story told by the spreadsheet matched the story unfolding in the field, minute by minute.
That meant the work moved constantly between the technical and the human. A spreadsheet could tell one story; a dispatcher, driver log, farebox key, route schedule, or maintenance record could tell another. The job was to reconcile those stories into something PennDOT and the agency could rely on.
The work also taught me the difference between suspicion and verification. The point was not to assume that agencies were wrong. The point was to make sure every reported number had a defensible path back to a source record, a procedure, and a person or system responsible for producing it. That distinction was important because good oversight had to protect PennDOT, the agencies, and the credibility of the funding process.
I learned much of that mindset from Wade White, my mentor on these projects. He taught me to question every number, not as a reflexive distrust of people, but as a discipline for understanding how a number was made. He guided me with just enough information to force me to figure things out, but stayed close enough that I was never left to fail alone. He had the relationships with people across the Commonwealth, the patience to listen carefully, and the curiosity to look under a bus when that was what it took to understand something.
One of the most valuable lessons was that data quality is not only a technical issue. It is an operating discipline. Good reporting depends on clear roles, consistent procedures, usable source records, secondary checks, exception documentation, and management review.
The Value Created
The work created value in several ways. It helped PennDOT test whether subsidy allocations were supported by data that could be traced, tested, and defended. It helped agencies identify where management-dependent practices needed to become documented, auditable quality-control procedures. And it turned technical findings into practical recommendations, implementation steps, and clearer documentation for future reporting cycles.
It also helped me grow. Repetition gave me pattern recognition. Travel gave me context. Fieldwork gave me respect for the people behind the numbers. And the combination of data analysis, interviews, documentation, and public-sector oversight helped shape how I still approach complex assignments: start with the source record, understand the workflow, test the assumptions, document the evidence, and make the recommendation usable.
Looking back, the Act 44 reviews were more than a series of assignments. They were a proving ground. I got to learn by doing, help by listening, and create value by making complicated operating data clearer, more reliable, and more accountable.