Rushly
Sign in
Methodology

How rankings work

Most Greek rating sites just average the stars. That lets three glowing reviews beat two hundred honest ones, and it never forgets a review from six years ago. We do it differently, and we show our work.

01

A Bayesian average, not a raw one

Every chapter starts from a neutral prior of 3.4/5, worth about 8 ratings. Real ratings pull the score away from that prior. A chapter with a handful of reviews stays close to neutral until enough people weigh in, so a tiny sample can’t rocket to #1.

02

Recent ratings count more

Chapters turn over completely every few years, so old opinions shouldn’t define them. Each rating decays with a 2-year half-life — a two-year-old rating counts half as much as a fresh one, and very old ones fade toward irrelevance.

03

Verified students count a little more

Anyone can rate anonymously — no account needed. But a rating from a verified .edu student carries about 1.5× the weight, because it’s harder to fake at scale.

04

We fight brigading

One device gets one voice: it can rate a chapter once every six months, and only its newest rating counts. When more than 3 ratings for a chapter come from the same network, each is scaled down proportionally (so a dorm can weigh in, but one person can’t flood a page). A sudden burst — 8+ ratings in 48 hours making up most of a chapter’s ratings — is held for review and counts at half weight. Device and network signals are stored only as one-way hashes — never raw — and used solely to keep the ratings honest.

05

Readers can back up or doubt a review

Reviews are anonymous, so anyone who knows the chapter can mark a review helpful or not helpful. A review many readers back counts for up to 1.5× in the score; one many readers doubt counts for as little as 0.4×. Votes never delete a review, and they can be gamed too, so they’re limited: one vote per device, nothing moves until a review has at least 4 votes, anonymous votes from one network count for at most 3 in total on a review, votes from the reviewer’s own network don’t count, and a verified .edu student’s vote counts 2×. Reviews that are widely marked not helpful are sent to a moderator to check.

06

Sample size and confidence are always visible

Chapters are ranked against the others of their kind at the same school: every fraternity a student can rush there in one list, every sorority in another. A chapter’s page and its school’s list always show the same rank. Closed or suspended chapters, and ones the university doesn’t list, keep their scores but aren’t ranked. A chapter needs an effective sample of at least 5 ratings to earn a rank. Below that we still show its score, marked “too few to rank.” Every score sits next to a confidence meter (high ≥ 30, medium ≥ 10) so you know how much to trust it.

The exact prior, half-life, and thresholds can change as we learn — when they do, we’ll update this page. Transparency is the point. See the content policy →