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 1 of 5
An Interval for an Average Is Not an Interval for Your Own Outcome
A report can estimate a group average quite precisely while individual outcomes remain widely spread. The distinction matters when a student reads an earnings model, a cost projection or a travel-time estimate: an interval describing uncertainty about a mean is not automatically a range that predicts the individual student’s result.
Worked analysisOne Distant Data Point Can Tilt a College Comparison
A line drawn through a scatterplot can make a set of colleges appear to follow a simple rule. But a single observation far from the rest on the horizontal axis may have a large effect on that line. Before using the slope to guide a decision, ask how much it depends on that observation and whether it is comparable with the other institutions.
Worked analysisAn Improvement After an Extreme Result May Have More Than One Explanation
Use a fictional repeated-measure example to examine selection on an unusually low baseline.
Baseline-selection checkFinding One Significant Difference Among Many Needs a Wider Check
Calculate a fictional multiple-testing example and distinguish a preplanned comparison from a selected result.
Multiple-comparison worksheetA Smaller Standard Error Does Not Mean Students Became More Similar
Separate individual variation from the sampling variability of a mean in a fictional study.
Spread-and-error worksheetTwo Possible Offers Do Not Automatically Make a 75% Chance
Compare fictional joint outcomes with the same individual probabilities under different dependence assumptions.
Joint-outcome worksheetThe Share of Correct Flags Depends on How Common the Condition Is
Use fictional eligibility-screening counts to show why a flag does not establish an individual decision.
Screening-count worksheetA Rank Can Change Because the Comparison Field Changed
Track a fictional institution and its comparison field before interpreting movement in rank.
Field-change auditA Histogram Bin Can Hide a Gap Between Students
Re-bin an invented set of scores to show what grouped data can and cannot reveal.
Histogram bin worksheetTwo Identical Trends Can Look Different When Chart Axes Change
Read a fictional pair of college charts by comparing values, axis ranges and units.
Axis-range checkA Service Visit Rate Needs Both Events and Time at Risk
Use a fictional student-month denominator to compare service contact counts without treating repeat visits as unique students.
Student-time worksheetA Percentile Rank Is Not the Percentage of Questions Correct
Use a fictional score distribution to separate a score, its reference group and a percentile rank.
Percentile worksheet