Blog
Original analysis of the federal files, and field notes from IR offices. 63 posts, 13 of them with the queries published so you can re-run every number.
63 posts
- Retrieval-Augmented Generation for Higher Ed Data, Explained for IR OfficesThe architecture behind every "AI that answers from your data" pitch, what it fixes, and the five ways it still breaksChoosing IR tools · 8 min · Choosing IR tools8 min ·
- What is a CIP code? The federal taxonomy behind every program number in higher edSix digits, three levels, one maintenance agency. How the Classification of Instructional Programs works, where it touches your job, and which level to use for which analysis.Federal data · 7 min · Federal data7 min ·
- How to Say No to Data Requests Without Burning BridgesA triage rubric, four scripts for the asks you actually get, and the redirect menu that turns a no into something better than a yesRunning an IR office · 7 min · Running an IR office7 min ·
- The Enrollment Cliff by State: Where the Freshman Supply Is Already FallingOriginal dataIPEDS residence data shows the cliff is regional, not national: Texas added 34,000 first-time freshmen a year since 2016 while New York lost 15,700Federal data · 10 min · Federal data10 min ·
- A third of colleges cut a team, and the team was small for years firstOriginal data549 of 1,715 continuously filing colleges dropped at least one team between AY 2014-15 and AY 2024-25, inside a sector that added 1,992 teams. A doomed team sits at the 9.4th percentile of roster size for years before it goes, and the variable that sorts cutters from non-cutters is enrollment rather than money.Federal data · 12 min · Federal data12 min ·
- One college sport is vanishing, and the fastest-growing one has no nameOriginal dataTennis lost 124 institutions in twenty years on a fixed set of colleges. The four sports that look like they are dying are all growing. And the largest gain in the file belongs to a survey bucket nothing in the data can describe.Federal data · 11 min · Federal data11 min ·
- 55 percent of the students, 42 percent of the athletes: the Title IX proportionality gap, measuredOriginal dataThe national gap is 13.05 points in AY 2024-25, wider than in 2014-15. Football is about half of it. Both sides of the ratio sit on the same EADA row, so any institution can compute its own number in one query.Federal data · 12 min · Federal data12 min ·
- Athletics revenue is not revenue: 40,829 EADA filings, zero deficitsOriginal dataThe revenue column is an allocation balanced to the expense line, not an income statement. Where the balancing happens, and what the file can honestly be used for insteadFederal data · 13 min · Federal data13 min ·
- 95,584 fewer clerical jobs, 80,571 more in finance: IPEDS staffing by occupation on 4,902 colleges, 2012 to 2024Original dataNet headcount change over twelve years at the 4,902 colleges that filed IPEDS Human Resources every year. Instruction barely moved; the office around it was rebuilt.Federal data · 14 min · Federal data14 min ·
- Graduates per opening: matching IPEDS completions against BLS demand across 17 CIP to SOC pairsOriginal dataRegistered nursing produces 1.39 graduates for every projected annual opening. Paralegal studies produces 0.25. Twelve of the 17 pairs we could honestly map fall below 1.00, and the state rankings underneath are more fragile than the national ones.Federal data · 14 min · Federal data14 min ·
- Why your program has no earnings number: 76.2 percent of PPD:2026 is dark, and what to put in the memo insteadOriginal dataWe measured the suppression instead of describing it: the reason ladder, the observed reporting floors, and the second source that nearly doubles coverageFederal accountability · 16 min · Federal accountability16 min ·
- Programs die small: which programs colleges are quietly closing, and which fields churn fastestOriginal dataThe median discontinued program awarded 2 credentials in its final year. The median survivor awarded 16. A completions-based churn benchmark for program review.Federal data · 14 min · Federal data14 min ·
- Same 3,361 programs, old GE test vs new OBBBA test: 569 walk free and 19 get caughtOriginal dataPPD:2026 carries the old Gainful Employment flags and the new OBBBA flags on the same rows. Running both shows how far the regime change moves the target.Federal accountability · 13 min · Federal accountability13 min ·
- Completion gaps by field: the higher a bachelor's field pays, the smaller the women's share of its graduatesOriginal dataA field-level equity baseline from IPEDS Completions 2024-25 and PPD 2026, for teams whose dashboards stop at the institution lineFederal data · 14 min · Federal data14 min ·
- Computing up 56.9 percent, nursing flat: bachelor's degree production by field, 2018-19 to 2024-25Original dataIPEDS Completions for the 2024-25 award year, compared with 2018-19. Registered nursing added 113 institutions with the program and 40 graduates.Federal data · 13 min · Federal data13 min ·
- 1,220 programs fail the OBBBA earnings test in PPD:2026. Here is the field and credential breakdown.Original dataThe honest denominator is 44,052 tested programs, not 209,321 rows, and undergraduate and post-baccalaureate certificates sit outside the test entirelyFederal accountability · 13 min · Federal accountability13 min ·
- Clema News and Updates: What's New for IR and IE TeamsA running index of product launches, research releases, and company updates from Clema for institutional research professionalsUsing Clema · 4 min · Using Clema4 min ·
- IR Resources and Further Reading: Everything We Published in One PlaceA curated index of Clema's guides, whitepapers, benchmarks, and reading lists for IR and IE professionalsCareer and community · 3 min · Career and community3 min ·
- The IR Team's Year: When Each Season Hits and What to PreparePeak (Sep, Jan, May), moderate (Oct, Nov, Feb, Mar), and low (Apr, Jun, Jul, Aug, Dec) mapped to the requests, reports, and prep that come with eachRunning an IR office · 6 min · Running an IR office6 min ·
- How a Large IR Team Uses Clema to Manage Breadth Across SchoolsA scenario walkthrough anchored in the whitepaper finding that large IR teams spend up to 1,310 hours per month on ad-hoc requestsUsing Clema · 5 min · Using Clema5 min ·
- How an IR Office Uses Clema When Their Key Analyst ResignsA scenario walkthrough for the institutional intelligence gap, anchored in research that 85% of institutions depend on a single person for institutional knowledgeData definitions · 5 min · Data definitions5 min ·
- How an IR Director Uses Clema When "Retention" Means Five Different ThingsA scenario walkthrough for the definitional drift problem, anchored in the whitepaper finding that 73.5% of requests are vague or unclearData definitions · 5 min · Data definitions5 min ·
- How an IE Director Uses Clema to Bridge the Dashboard-vs-Email GapA scenario walkthrough tied to the whitepaper finding that 64.7% of institutions struggle with self-service dashboardsUsing Clema · 5 min · Using Clema5 min ·
- How an Office of One Uses Clema During IPEDS Peak SeasonA scenario walkthrough for an IR director running everything solo, anchored in whitepaper findings on small-team peak-season utilizationUsing Clema · 5 min · Using Clema5 min ·
- IR Capacity Benchmarks: The Real Cost of Ad Hoc Data RequestsWhat 50+ interviews with IR teams reveal about how much time ad hoc requests actually consume, by team sizeRunning an IR office · 7 min · Running an IR office7 min ·
- What Clema Cannot Do Yet (and When It Is Not the Right Fit)An honest limitations list for IR teams evaluating Clema, and the institutions where a different tool is the better callChoosing IR tools · 6 min · Choosing IR tools6 min ·
- How to Ask an IPEDS Question in ClemaThe prompt patterns that return sourced, defensible IPEDS figures in plain EnglishUsing Clema · 5 min · Using Clema5 min ·
- How to Build a Peer Group in ClemaA step-by-step walkthrough for constructing a defensible IPEDS peer group in plain EnglishUsing Clema · 5 min · Using Clema5 min ·
- How to Set Up SSO With Clema (SAML and OAuth)The exact steps to configure SAML 2.0 or OAuth 2.0 SSO with your institution identity providerUsing Clema · 5 min · Using Clema5 min ·
- Clema Onboarding: A Four-Week Setup ChecklistWhat happens each week from data connection to stakeholder rollout, with the people and artifacts involvedUsing Clema · 6 min · Using Clema6 min ·
- Get Started With IPEDS in Plain EnglishFive IPEDS questions to ask Clema in your first session, and what each one shows about the federal datasetUsing Clema · 5 min · Using Clema5 min ·
- First Five Questions to Ask ClemaA starter list of plain-English questions that show what Clema does on federal and institutional dataUsing Clema · 5 min · Using Clema5 min ·
- Clema vs HelioCampus: A Cost Comparison for IR TeamsPublished pricing that is free to start, with plans from $99 per month, versus unpublished enterprise pricing that requires a quoteChoosing IR tools · 6 min · Choosing IR tools6 min ·
- Clema vs EAB Edify: A Cost Comparison for IR TeamsPublished pricing that is free to start, with plans from $99 per month, versus custom enterprise pricing that ran $36,000 to $216,000 per year on a published fee scheduleChoosing IR tools · 6 min · Choosing IR tools6 min ·
- Why Higher Ed Needs Its Own AI Assistant, Not ChatGPTGeneral AI reads the public web and guesses. Higher ed needs AI that reads its data, knows IPEDS, and shows its sources.Choosing IR tools · 5 min · Choosing IR tools5 min ·
- Why IR Capacity Has Not Kept Up With Data DemandRequests keep climbing, headcount stays flat, and the math no longer works for an IR team of one to three peopleRunning an IR office · 6 min · Running an IR office6 min ·
- A Calendar of Federal Data Releases Higher Ed IR Teams Should Track in 2026Twelve federal datasets, when they drop, what is current, and what is coming next so your IR team is never caught off guardFederal data · 11 min · Federal data11 min ·
- Why 'enrolled student' has five definitionsThe data is rarely the problem. The missing layer of meaning on top of it is, and one ordinary term can carry five valid definitions at once.Data definitions · 11 min · Data definitions11 min ·
- Is your institutional knowledge one resignation away from gone?In 85% of the institutions we studied, the definitions that run the place live in one person's head. Here is a way to diagnose how fragile yours are.Data definitions · 12 min · Data definitions12 min ·
- From person-dependent to system-supported: a framework for governing data definitionsA six-step framework for moving institutional definitions out of one analyst's head and into governed, survivable infrastructureData definitions · 12 min · Data definitions12 min ·
- How to learn institutional research: certificates, courses, and free trainingMost people land in institutional research sideways, from the registrar, the provost office, or a faculty data role. This is the directory of where to actually learn the craft, sorted by what it costs and what you get.Career and community · 12 min · Career and community12 min ·
- Data storytelling for institutional research: turning numbers into narrativesA graduation rate is a sentence with the ending cut off. Data-informed advocacy is the skill of finishing it for the person in the room.Career and community · 10 min · Career and community10 min ·
- From reactive reporting to proactive prediction: a playbook for IR teamsMost institutional data is used to describe the past. Here is how IR leaders flip it forward.Predictive analytics · 10 min · Predictive analytics10 min ·
- Why gradient boosting beats LLMs for student retention predictionA general-purpose chatbot cannot tell you which students will not return next term. Here is why, and what does.Predictive analytics · 11 min · Predictive analytics11 min ·
- Predictive analytics in higher ed: the $40-a-month stack vs the $200K platformA cost and vendor reality check for IR leaders weighing build, buy, or partner.Predictive analytics · 10 min · Predictive analytics10 min ·
- Predicting term-to-term retention: what a working model actually looks atInside a retention model evaluated on 276,519 unseen student-term records.Predictive analytics · 11 min · Predictive analytics11 min ·
- Predicting graduation within 150% time: course combinations, confidence, and committee-ready explanationsA graduation model evaluated on 36,000 unseen students reached ~0.911 AUC. Here is what it sees that cohort reports do not.Predictive analytics · 11 min · Predictive analytics11 min ·
- Forecasting enrollment and program viability: where prediction meets cabinet-level decisionsDifferent question, different model family. A guide to time-series forecasting, yield prediction, and program-margin analysis for IR leadership.Predictive analytics · 10 min · Predictive analytics10 min ·
- Inside 8,796 comments on the STATS rule: what higher ed actually told the DepartmentA 16-section atlas of the "Do No Harm" earnings-rule docket. Who showed up, what they argued, what they fear, what they want, and what is most likely to land before February 2027.Federal accountability · 16 min · Federal accountability16 min ·
- FVT/GE is ending: what the final 2026 reporting year means for your STATS transitionOctober 1, 2026 is the last FVT/GE submission. Four moves to make before the first Earnings Premium results in February 2027.Federal accountability · 10 min · Federal accountability10 min ·
- From OBBB to Earnings Premium: how the new Workforce Pell accountability math actually worksThe clean line from statute to STATS to the 2-of-3 eligibility math, with a worked exampleFederal accountability · 10 min · Federal accountability10 min ·
- Your STATS exposure map: triage your programs before February 2027A program-level playbook for the eighteen months between PPD:2026 and ED's first official resultsFederal accountability · 10 min · Federal accountability10 min ·
- The 50% rule: how a handful of weak certificates can put your whole institution on provisional statusThe institutional-level STATS trigger most administrators haven't priced inFederal accountability · 9 min · Federal accountability9 min ·
- Six files, 209,321 programs, three join keys: turning PPD:2026 into answers your provost actually asksThe IR analyst's working view of the PPD:2026 dataset, and where the time actually goesFederal accountability · 10 min · Federal accountability10 min ·
- STATS vs FVT/GE vs PPD: What's the Difference?Three names, one dataset, and why they keep getting confusedFederal accountability · 8 min · Federal accountability8 min ·
- Conferences Institutional Researchers in Higher Ed Should Attend in 202614 conferences worth your time — from the national flagship to regional gatherings and data-focused eventsCareer and community · 12 min · Career and community12 min ·
- What IR Experts Are Reading and Where They're Finding Community: A Resource Guide Built from 50+ InterviewsNewsletters, listservs, and peer communities that IR professionals actually use to stay sharpCareer and community · 8 min · Career and community8 min ·
- What IR Experts Are Listening to and Learning From: Podcasts and Books, A Resource Guide Built from 50+ InterviewsThe podcasts, books, and training programs IR professionals recommend for going deeperCareer and community · 10 min · Career and community10 min ·
- Is Your University's Data Request Form Actually Optimized? How Better Design Can Reduce 77% of Clarification TimeAn analysis of 18 university data request forms reveals three dominant intake formats, threefold complexity variation, and the design gaps that create hidden work for IR teamsRunning an IR office · 8 min · Running an IR office8 min ·
- Who's Growing Through Dual Enrollment? A Side-by-Side Benchmark of Community Colleges and Four-Year InstitutionsScale vs. Speed: How Dual Enrollment Differs Across Sectors Using IPEDS 2023–2024 DataFederal data · 8 min · Federal data8 min ·
- What IR Teams Really Benchmark: Evidence from 50 InterviewsFour Core Benchmarking Domains and Three Benchmarking Types from 50 IR Director InterviewsRunning an IR office · 10 min · Running an IR office10 min ·
- What Is the Relationship Between Data Request Intake Quality, Request Intelligence, and Workload in Institutional Research Given That 78% of Requests Require 10+ Follow-Up Questions?Why IR Teams Lose Capacity Before Analysis Even BeginsRunning an IR office · 10 min · Running an IR office10 min ·
- Best Practices for Institutional Researchers to Optimize Data Requests and Reclaim the 60% of Team's CapacityData Request Guidelines for Streamlining Self-Service DashboardsRunning an IR office · 12 min · Running an IR office12 min ·