English COLLEGEVERSE RESEARCH · 49 ARTICLES
Data & College Choices
Understand the population, year and definition behind a number before using it to compare colleges.
Newest publications first · Page 2 of 5
A List of Active Programmes Cannot Measure Programme Survival
Build an entry-cohort ledger before interpreting a college directory as evidence of programme continuity.
Cohort survival auditA Trend Line Beyond Its Data Is a New Assumption
Distinguish interpolation from extrapolation using a fictional college trend and a transparent boundary check.
Prediction-domain checkZero Average Forecast Error Can Still Mean Large Budget Mistakes
Compare signed error with absolute error in two invented study-cost forecasting methods before calling either accurate.
Bias and absolute-error worksheetAn Index of 120 Does Not Mean One College Costs More Than Another
Rebase two fictional cost series and separate relative growth from their actual currency amounts.
Base-period auditA Monthly Rise in Student Job Listings May Still Be Seasonally Weak
Separate raw monthly changes from a clearly labelled illustrative seasonal comparison before drawing conclusions about study-area opportunities.
Seasonal comparison checkMore Matched Programmes Do Not Necessarily Mean Better Matches
Audit a fictional cross-directory linkage with correct links, false links and missed matches instead of relying on the match rate alone.
Precision and recall auditAn Unchanged Outcome Rate Can Hide Many Students Changing Status
Use a matched transition table to distinguish a stable aggregate percentage from stability in individual students’ outcomes.
Transition accountingTwo Strong Percentages Do Not Tell You How Many Students Meet Both
Calculate the possible overlap of two outcomes when only marginal percentages are published, without assuming independence.
Overlap bounds worksheetThe Same Median Can Hide Very Different Lower Outcomes
Compare two fictional distributions with an identical median, then count outcomes below a clearly specified planning threshold.
Lower-outcome threshold checkA Weighted Survey Percentage Is Not a Count of Respondents
Reproduce a simple weighted result, preserve the actual sample size, and identify which published denominator you are reading.
Weighted-denominator auditA Higher Outcome Rate Can Still Sit Further Below Its Benchmark
Compare invented observed rates with their contextual benchmarks without treating the gap as a causal score or personal prediction.
Observed-minus-benchmark worksheetAn Institution-Wide Number May Not Describe Your Campus
Use an invented campus decomposition to separate a reporting unit from the study location relevant to an applicant.
Campus-scope audit