Methodology & editorial standards

How we source data — and what we refuse to fake.

College decisions are high-stakes, so the numbers behind them have to be honest. This page lays out where our data comes from, how our models work, and the standards the QuantAdmit editorial team holds every figure to.

Our editorial standards

QuantAdmit Research maintains a simple discipline: cite the source, and never manufacture precision we don’t have.

We cite our sources. Every published figure — acceptance rate, earnings, net price, test scores — traces to a named public or primary dataset, listed below.
We never fabricate precision. When a figure is unknown we show “—” rather than invent one. We publish “<1%” instead of a false “0%,” and we never claim a 100% chance of anything.
We label estimates as estimates. Model outputs are presented as probabilities and ranges, not verdicts, with their limits stated plainly.

Where our data comes from

The facts on QuantAdmit are compiled from public and primary datasets. We don’t buy or resell proprietary lists; these are the real sources behind the product.

U.S. Dept. of Education — College Scorecard: Graduate earnings, average net price (after aid), and SAT/ACT score ranges.; U.S. Department of Education
IPEDS: Enrollment, and the urban-centric locale we map to campus setting (city / suburban / college town).; National Center for Education Statistics
Opportunity Insights (Chetty et al.): Economic-mobility measures — the rate at which a college moves students up the income distribution.; Chetty, Friedman, Saez, Turner & Yagan / Opportunity Insights
Clery Act disclosures: Campus-safety figures from federally mandated crime and security reporting.; U.S. Department of Education
OpenAlex: Research output and scholarly activity associated with each institution.; OurResearch
FEMA National Risk Index: Natural-hazard exposure for a campus's location.; Federal Emergency Management Agency
Open-Meteo: Climate and pleasant-days estimates for each campus's location.; Open-Meteo

Federal datasets lag reality by roughly two years; we surface the most recent figures each source publishes and note when a value is missing.

How our models work

Our tools apply established decision science to admissions. Short, honest descriptions follow; the full derivations and origins live in the science.

Acceptance-rate model

We start from each school's published base acceptance rate and shrink a personalized estimate toward it using empirical-Bayes shrinkage — more aggressively at the most selective schools, where individual stats are least decisive. Your GPA, SAT/ACT, and course rigor move the estimate, but the school's own admit rate anchors it, so a strong profile reads as a realistic chance rather than a coin flip.

Conjoint analysis

Instead of asking what you say you want, we ask you to choose between hypothetical colleges and infer your part-worth utilities — the weight you actually place on cost, selectivity, outcomes, and fit — from the trade-offs you make. These revealed preferences drive your personalized value scoring across schools.

Monte Carlo simulation

We simulate thousands of full admissions seasons across your list. Rather than a single 'chance,' you see the distribution of outcomes — the probability of at least one admit, the expected number, and the shutout risk — with results linked by an applicant-strength correlation so your outcomes move together the way they would in reality.

Portfolio optimization (Chade–Smith)

Our 'Choose your 20' optimizer implements the Chade–Smith (2006) application-portfolio value function: a list is worth the expected utility of the best school that admits you. It combines your conjoint-measured preferences and your correlated admit odds, then adds real-world constraints — the Common App's 20-school cap, one binding Early Decision, and a penalty on piling up correlated reaches.

What we don’t do

We don’t guarantee outcomes. Admissions is inherently uncertain. Every output is a model estimate, never a promise of admission.
We don’t sell your data. Your profile is used to power your analysis — not packaged, sold, or handed to third parties.
We anonymize student inputs. Names are stripped from analysis and never sent to external services.

Questions about a specific figure or method? This page is maintained by the QuantAdmit editorial team. For the deeper theory behind each model, see the science.

These outputs are estimates from a baseline model — not guarantees of admission, cost, or outcome.