University Of Maryland Salary Database Reveals Shattering Truths About Campus Paychecks

What You Actually Get When You Pull the UMD Salary Data

The University of Maryland salary database is a public records dump that contains employee compensation information for the institution. It is not a single clean spreadsheet. It is a collection of files that the university releases, usually through their human resources or financial services portal, sometimes in response to public records requests. The data covers base salary, gross pay, benefits, and sometimes supplemental compensation for faculty and staff across multiple campuses within the system. I spent probably six months working with this dataset when I was doing compensation analysis for a faculty senate committee. We were trying to understand pay equity across departments. What I learned was that the raw data is mostly useless without a serious amount of cleaning and contextual knowledge about how the university structures its payroll.

How to Access the University Of Maryland Salary Database Reveals Shattering Truths About Campus Paychecks

Start at the University of Maryland's central human resources page. They typically publish annual compensation data there. You may also find it through the Office of the Comptroller or the business services division. The data usually comes in CSV or Excel format, sometimes as PDF tables that you have to scrape out of, which is annoying. If the published version does not have what you need, you can file a public information act request through the university's administrative office. That process takes about two to four weeks and they will send you whatever they have on file. The files you get will contain columns like employee name, department, position title, job classification, base salary, taxable wages, retirement contributions, health insurance premiums, and sometimes individual benefit line items. The exact columns vary year to year, which makes longitudinal comparison a pain.

The Problems Nobody Warns You About

The first issue is that job titles are inconsistent. A professor in the computer science department might be listed under five different title variations across the files. Some are "Assistant Professor," some are "Asst. Prof.," some have research appointments layered on top. If you are doing cross-departmental analysis, you need a mapping table or you will end up comparing apples and oranges. I built an Excel sheet that normalized roughly forty different title variants into twelve broad categories. It took me about three weeks to get it right, and I still had to manually review edge cases. The second problem is that salary figures are often reported differently depending on the fiscal year structure. UMD operates on a fiscal year that may not align with the calendar year, and some positions are twelve-month while others are nine-month academic appointments. A ninety-thousand-dollar salary on a nine-month contract is not the same as a ninety-thousand-dollar salary on a twelve-month contract. The database will list both numbers identically. You have to know which column indicates appointment length and adjust accordingly. I learned this the hard way when I calculated that the history department seemed to pay its faculty twenty percent more than the engineering department, which was obviously wrong. A specific edge-case I ran into involved dual appointment employees. Someone who holds a joint position between the graduate school and an academic department shows up in the data twice, once under each unit's budget code. If you simply sum all salaries by department to get total payroll, you will double-count these people and your numbers will be inflated. The workaround is to create a unique identifier from the employee name and a partial SSN or employee ID if it is included in the data. UMD sometimes redacts the full employee ID, but a combination of name plus department plus hire date is usually enough to deduplicate. I wrote a Python script using pandas to handle this, and it cut my reconciliation time from a full day down to about twenty minutes.

What the Data Actually Shows

When you clean it properly, the numbers tend to confirm what people in higher education already suspect. Administrative compensation has grown faster than faculty compensation over the past decade. There are significant pay gaps between colleges within the university, with business and engineering typically at the top and humanities and education toward the bottom. adjunct and part-time faculty compensation is extremely variable and often not captured consistently in the main database. One counter-intuitive finding is that the largest salary discrepancies are not always between departments. They are often within departments, between people with the same title but different years of experience or different funding sources. A professor funded through a grant may have a different base salary structure than one funded through departmental allocations, even at the same rank. The database will show them as identical job titles with different pay, which looks like inequity on the surface but may reflect legitimate funding differences.

Practical Steps for Working With This Data

Download the most recent file and any historical files you can find. Load everything into a single spreadsheet or database. Create a master column that standardizes the fiscal year and academic year for each record. Build a department hierarchy that maps sub-departments to their parent colleges. Flag any employee who appears multiple times. Normalize job titles using a lookup table rather than relying on the raw text. When calculating average salaries, report both the mean and the median. Salary distributions are heavily right-skewed because executive compensation drags the mean upward. The median tells you what a typical employee in that category actually makes. Also calculate salary ranges at the twenty-fifth and seventy-fifth percentiles, because the spread matters more than the average for understanding equity issues. Do not trust the data for anything involving benefits unless the file explicitly includes benefit line items with dollar values. Many years of the UMD salary data omit detailed benefit breakdowns or include them in a separate file that is not clearly linked. If you need benefits data, you will likely have to file a separate request or pull it from a different source entirely, such as IRS Form 990 filings which the university is required to publish.

Limitations and When to Stop

The database does not include contractor or vendor payments, so any analysis of total campus spending on labor will be incomplete. It does not reliably capture postdoctoral researchers in many years, which distorts STEM department analyses. Graduate student stipends often appear under a different classification code than regular employees, and sometimes they are excluded entirely from the published salary files. If you are studying compensation equity in STEM fields, you need to account for this gap or your results will be systematically biased against those departments. The data is also lagged. What you are looking at is usually from the previous fiscal year or earlier. If there have been significant hiring freezes, promotions, or restructuring events recently, the published numbers will not reflect them yet. I have seen people cite UMD salary data that was eighteen months old as if it were current, which led to incorrect conclusions about ongoing disputes that had already been resolved. If you need real-time salary information or want to track individual employee changes over time, this database will not serve you well. It is a snapshot tool, not a monitoring tool. For longitudinal analysis, you are better off requesting the same dataset annually and building your own tracking system, though you will need to deal with the column variation issue each year.

The biggest practical takeaway is that the data reveals more through careful cleaning and contextual understanding than it does through a simple download and summary. The shattering truths are there, but you have to do the work to surface them. Most people who look at this data casually come away with superficial observations that repeat talking points rather than insights grounded in the actual numbers.