Reference
The College Rankings Glossary
A college rankings glossary is only useful if it tells you who defines a term and what it is worth. Every entry below gives one plain definition, the publishers that score it, and the exact weight each one assigns, read from the five methodology pages in September 2026.
How do I tell a real ranking term from one the press invented?
A publisher-defined term has three things attached to it: a written definition on the publisher's own methodology page, a number of percentage points, and a stated data source with a data year. Peer assessment has all three. So does per capita academic performance. Terms like "top-tier", "elite", "hidden gem" and "Ivy-caliber" have none of the three, and no formula anywhere produces them.
That distinction matters because the vocabulary is doing arithmetic. In the 2026 U.S. News Best Colleges formula, the opinions of other administrators are worth 20 points out of 100, and everything a school spends per student is worth 8. The bar below shows where the whole 100 goes for National Universities.
Five steps to date a claim about a ranking
- Open the publisher's own methodology page and note the edition year printed in the first paragraph.
- Read the data years inside, not the edition year. The 2026 U.S. News graduation rate averages the classes that entered in autumn 2015 through autumn 2018.
- Look for the paragraph that lists changes from the previous edition. Every publisher writes one, and it is where removals are announced.
- Check whether the term you are chasing appears on that page at all. If a factor is discussed in advice articles but is missing from the methodology, it has left the formula.
- Compare against an archived copy of the same page from two years earlier. That is how the dropped terms table below was built.
If you want the weights side by side rather than term by term, the comparison of ten ranking methodologies lays out every published weight in one matrix, and the shorter explainer on how college rankings work covers the mechanics of normalisation and scoring.
What do the terms on a college rankings methodology page actually mean?
Search the 40 entries, or filter by the publisher whose page you are reading. The badge on each card is the highest weight that term carries in any of the five rankings covered here. Filter by a publisher and the badge switches to that publisher's own weight. Terms with no badge are definitional: they shape the score without being scored themselves.
No terms match that search.
A to C
Academic reputation surveyreputation survey
A questionnaire asking academics to name or rate the institutions they consider strongest, converted into a score with no measured output behind it.
- QS academic reputation 30%
- THE teaching and research reputation 33%
Why it can mislead: QS lets each institution submit up to 400 contacts for each survey, and the THE figure combines two waves of one survey, so the number moves slowly and partly reflects who was asked. Peer assessment Employer reputation
AlumniARWU indicator
The count of an institution's bachelor's, master's and doctoral degree holders who have won a Nobel Prize or a Fields Medal, discounted by the decade the degree was awarded.
- ARWU 10%
Why it can mislead: A degree awarded in 1921 to 1930 still counts at 10 percent of full value, and only degrees awarded after 2011 count in full, so a century of history sits inside a 2026 score. Award
ARWUShanghaiRanking
ShanghaiRanking's Academic Ranking of World Universities, built from six indicators covering prizes, highly cited researchers and papers, with no survey and no teaching measure in the formula.
- Definitional, no weight of its own
Why it can mislead: ShanghaiRanking says more than 2,500 universities are actually ranked and only the best 1,000 are published, so absence from the table usually means the position was not printed. Rank Not Published
AwardARWU indicator
The count of staff who won a Nobel Prize in physics, chemistry, medicine or economics, or a Fields Medal in mathematics, while working at the institution, discounted by the decade of the win.
- ARWU 20%
Why it can mislead: A winner affiliated with several institutions gives each the reciprocal of that count, and a shared Nobel is split by each winner's share, so one prize can be worth a fraction of a point in several places at once.
Bibliometrics
The family of measures built from publication and citation counts rather than from surveys or figures a college reports about itself.
- Definitional, no weight of its own
Why it can mislead: The databases and windows differ. ARWU counts Web of Science articles from a single year, THE counts Scopus publications from 2020 to 2024 with citations collected to 2025, and U.S. News uses Elsevier data over 2020 to 2024. Two bibliometric scores for one university are not comparable. Normalisation
Carnegie Classification
The framework that sorts United States institutions by award level, academic program mix and size, and so decides which schools are compared with each other.
- U.S. News, sets the ranking category
Why it can mislead: The 2025 Institutional Classification replaced the Basic Classification and produces 31 groups, but the 2026 U.S. News edition still sorted schools with the 2021 Basic Classification, so two systems are in circulation at once. See what changed in the 2025 Carnegie methodology. Research activity designation
Citations per facultyCpF
The QS research impact indicator, which divides an institution's citation count by the size of its faculty so that research is judged per head rather than in total.
- QS 20%
Why it can mislead: Dividing by faculty rewards small research-intensive institutes and penalises large teaching-heavy universities for reasons that have nothing to do with research quality. Per capita academic performance
College Scorecard
The United States Department of Education portal that publishes cost, completion, debt and earnings figures for institutions that take federal student aid.
- U.S. News, source for two indicators worth 10 percent together
Why it can mislead: For the third year running the 2026 edition read debt from releases pooling fiscal years 2019 to 2020 and 2020 to 2021, so both money indicators describe students who left college years ago. Median federal loan debt
Common Data SetCDS
A set of standard item definitions agreed between colleges and the publishers behind it, the College Board, Peterson's and U.S. News, so one survey question means the same thing everywhere.
- U.S. News, survey definitions
Why it can mislead: The CDS is a set of standards and definitions, not a survey instrument or a database, so a college's published CDS file is its own answer sheet and no central body audits it. IPEDS
D to G
Employer reputation
The QS survey of employers, asking which institutions produce the graduates they consider best prepared for work.
- QS 15%
Why it can mislead: Employers can only rate institutions they have heard of, and QS says the survey measures the reputation of institutions worldwide, not only the ones that nominated the respondent, so brand reach counts twice.
Faculty resources
The U.S. News group of three indicators about who teaches: average full-time faculty salaries, the share of instructional faculty employed full time, and the student to faculty ratio.
- U.S. News National Universities 11%
- U.S. News other categories 15%
Why it can mislead: The group is worth four points more outside the National Universities list, because those categories have no faculty research indicators to absorb that weight. Salaries alone are 6 percent for National Universities and 8 percent elsewhere. Student-faculty ratio
Field of study data
College Scorecard figures reported per program rather than per institution, keyed to the first four digits of a Classification of Instructional Programs code plus a credential level.
- Definitional, no weight of its own
Why it can mislead: A program median can rest on very few graduates, and the credential level changes what is counted: undergraduate loans are excluded from the debt figure shown for a master's program. If you are choosing between fields first, the college major quiz is a better starting point than a program earnings table.
Financial resources per student
Spending on instruction, research, public service, academic support, student services and institutional support, divided by full-time equivalent enrolment, with housing and athletics left out.
- U.S. News 8%
Why it can mislead: It measures money spent, not money spent well, and the 2026 edition used fiscal year 2023 expenditure against 2022 to 2023 enrolment. Research and public service spending is first discounted by the undergraduate share of enrolment.
First-year retention rate
The share of a full-time entering class that came back the following autumn, averaged by U.S. News across the classes entering in 2020 through 2023.
- U.S. News 5%
Why it can mislead: A cohort year only counts when at least 25 first-year students are in it, and the average is taken across whichever years have usable data, so a small college can be scored on one year.
Graduation rate (six-year)
The share of full-time, first-time bachelor's-seeking students who finished a bachelor's degree within six years, which is 150 percent of the normal four.
- U.S. News 16%
Why it can mislead: The 2026 edition averaged the classes entering in autumn 2015 through autumn 2018, so it describes students who started college about a decade ago. Transfers in and part-time students are outside the cohort entirely. Graduation rate performance
Graduation rate performance
The gap between a school's actual six-year graduation rate and the rate a regression predicted for its intake, scored as actual divided by predicted and averaged over four cohorts.
- U.S. News 10%
Why it can mislead: The prediction controls for spending per student, so a school that spends more is expected to graduate more and earns no credit for doing so. Worked numbers are in the section below. Value added
H to N
Highly Cited ResearchersHiCi
The count of an institution's researchers who appear on the Clarivate Highly Cited Researchers list, counted once for each field in which they are recognised.
- ARWU 20%
Why it can mislead: ARWU 2026 used the list issued in December 2025 and counts only primary affiliations, so a researcher with a second appointment adds nothing to the second university. Hiring one such researcher can move a mid-table score.
International outlook
The pillar that scores how much of a university's staff, students and co-authored papers come from outside its own country.
- THE international outlook 7.5%
- QS global engagement 15%
Why it can mislead: THE now adjusts these measures for country population, because crossing a border is easier from a small country. A study abroad measure sits in the pillar at zero weight and does not affect any score.
IPEDSIntegrated Postsecondary Education Data System
The federal survey system that every United States institution taking federal student aid must report to, and the fallback source publishers use when a school ignores their own survey.
- U.S. News, secondary data source
Why it can mislead: IPEDS instructs schools to report on a broader group of faculty than the Common Data Set does, so a school scored from IPEDS is measured on a different population than the school beside it. Common Data Set
Median federal loan debtborrower debt
The midpoint of accumulated federal loan debt among students who borrowed and then graduated, with PLUS loans, Perkins loans and post-separation disbursements excluded.
- U.S. News 5%
Why it can mislead: Graduates who borrowed nothing are not in the figure at all, and it does not control for how many students borrowed, so a school where few families need loans posts a debt number that describes almost nobody. College Scorecard
N&SNature and Science indicator
The ARWU count of research articles published in Nature and Science between 2021 and 2025, weighted by where the institution sits in the author list.
- ARWU 20%
Why it can mislead: A corresponding-author address counts in full, a first-author address at 50 percent and any other contributing address at 10 percent. For institutions specialised in humanities and social sciences the indicator is dropped and its weight moved elsewhere.
Net price
Published tuition, fees, books, supplies and living costs minus grant and scholarship aid, averaged over full-time first-time students who received federal financial aid.
- WSJ, half of years to pay off net price 17%
Why it can mislead: It is one average across every aided student at one school, so it can sit far from what your family would pay. Read how net price is calculated before comparing two figures.
Normalisation
The step that puts unlike measures on one scale before they are added, usually by z-score, by position in a cumulative distribution, or by setting the top institution to 100.
- Definitional, no weight of its own
Why it can mislead: Normalisation decides what a raw difference is worth. THE adds an exponential component for reputation, research excellence, research influence and patents, so a small raw lead at the top of those skewed distributions moves the score a long way. Percentile
O to R
Peer assessmentexpert opinion
The U.S. News survey in which presidents, provosts and deans of admissions rate the undergraduate academic quality of other schools in their own category on a five-point scale.
- U.S. News 20%
Why it can mislead: It is a two-year rolling average, and respondents who do not know a school are told to skip it, so the score is slow and thin at the same time. Full numbers in the section below. Academic reputation survey
Pell graduation rate
The six-year graduation rate of students holding a federal need-based grant, summed with the share of the entering class that held one, capped at 0.5.
- U.S. News 5.5%
Why it can mislead: The cap means a school gains nothing for enrolling past half its class from low-income households, because U.S. News found the relationship stopped holding there.
Pell graduation performance
The ratio of a school's six-year completion rate for grant holders to its rate for everyone else, capped at 1 and then topped up by the share of the class holding grants.
- U.S. News 5.5%
Why it can mislead: Institutions that do not accept these grants, such as the service academies, are assigned a value one standard deviation above the mean rather than being left out, on the assumption their affordability already delivers access.
Percentile
A rank position expressed as the share of institutions scoring below a value, used to flatten outliers before a measure is normalised.
- Definitional, no weight of its own
Why it can mislead: U.S. News converts borrower debt and low-income completion values to percentiles first, so a school can move several places while its underlying dollar figure barely changes. Test scores are converted to national percentile distributions too. Normalisation
Per capita academic performancePCP
The ARWU size correction: the weighted scores of its other five indicators divided by the number of full-time equivalent academic staff.
- ARWU 10%
Why it can mislead: When staff numbers for a country cannot be obtained, ShanghaiRanking substitutes the average staff count of the world top 1,000 for every institution in that country, so the denominator is sometimes an estimate rather than a measurement.
Publications indicatorPUB
The ARWU count of papers indexed in the Science Citation Index-Expanded and the Social Science Citation Index in a single year, with social science papers counted twice.
- ARWU 20%
Why it can mislead: Only documents typed as articles count, and only one year does: the 2026 edition used 2025 alone, so a strong year and a weak year can move a university several places for reasons no student would notice.
Rank Not Published
A position the publisher calculated but chose not to print, because differences at that end of the table are too small to report as separate numbers.
- U.S. News, bottom decile ranges
- THE, bands below rank 200
Why it can mislead: Most ranked institutions have no single number at all. THE prints exact ranks only for the top 200, and ShanghaiRanking publishes only the best 1,000 of the 2,500 or more it ranks. Tie Unranked
Research activity designationR1, R2, RCU
The 2025 Carnegie labels for research volume: Research 1 needs 50 million dollars of research and development spending and 70 research doctorates, Research 2 needs 5 million dollars and 20 doctorates, and Research Colleges and Universities needs more than 2.5 million dollars.
- Definitional, no weight of its own
Why it can mislead: The thresholds count money and doctorates, not quality, and there is no cap on how many institutions qualify. None of the five methodologies here scores the designation itself. Carnegie Classification
S to Z
Selectivity
The old grouping of admissions inputs: entrance exam scores, high school class standing and acceptance rate. Only the exam half survives, as the standardized tests indicator.
- U.S. News standardized tests 5%
Why it can mislead: High school class standing last counted in the 2023 edition and acceptance rate was already gone from the nine broad areas listed in 2022, so any page still calling selectivity a ranking factor is out of date. See what counts as a good SAT score for the score side.
Social mobility indicator
A measure of how well a school enrols and graduates students from low-income households, published by U.S. News both inside the main formula and as a separate Top Performers list.
- U.S. News, two factors worth 11%
Why it can mislead: Only schools whose values were computed without substituted figures qualify for the separate list, so some genuine large enrolers of low-income students are excluded on data grounds rather than on performance.
Student-faculty ratio
Full-time equivalent students divided by full-time equivalent faculty, counting a part-time instructor as one third of a full-time one.
- QS faculty student ratio 10%
- THE staff-to-student ratio 4.5%
- U.S. News 3% or 4%
Why it can mislead: The three publishers count different people. U.S. News follows Common Data Set instructions and excludes stand-alone graduate and professional faculty, while THE counts academic staff involved in teaching or research. The same campus can produce three different ratios. Faculty resources
Sustainability score
The QS indicator on institutional sustainability performance, and the only sustainability measure carrying weight in any of the five general rankings covered here.
- QS 5%
Why it can mislead: Five percent is the same weight QS gives to employment outcomes and to each of its three global engagement measures, so a strong sustainability result moves very few positions.
Tie
Two or more institutions sharing one position, either because their overall scores are equal after rounding or because the publisher grouped them into a band.
- THE, banded ranks
Why it can mislead: THE lists banded institutions alphabetically, so a university that appears first inside a 201 to 250 band is not ranked first in it. Alphabetical order is not a result.
Unrankedreporter status at THE
An institution listed in a directory but given no position at all, because it failed the publisher's eligibility rules rather than because it scored badly.
- U.S. News, eligibility rules
- THE, reporter status
Why it can mislead: U.S. News requires at least 100 undergraduates and a six-year graduation rate for a cohort of at least 25, so unranked is often a size test. THE excludes universities with fewer than 1,000 relevant publications over five years. Rank Not Published
Value addedsalary value impact
Any measure that compares a school's result against a statistical estimate of what its own intake would have produced elsewhere, then scores only the difference.
- WSJ salary value impact 33%
- WSJ graduation rate impact 20%
- U.S. News graduation rate performance 10%
Why it can mislead: The estimate is the whole measure. WSJ builds its salary estimate from students' prior exam results and the cost of living in the college's state, so a college in a low-cost state starts from a lower expectation and is easier to beat. Graduation rate performance
Weight
The share of the final score an indicator controls, applied after every measure has been put on a common scale.
- Definitional, no weight of its own
Why it can mislead: Weights move when data is missing. A U.S. News school with no usable entrance exam scores for two years has that 5 percent added to graduation rates, taking them from 16 percent to 21 percent, so two schools on the same list are not scored on the same formula.
Yield
The share of admitted students who enrol, once treated as a measure of how much students wanted a place.
- Definitional, no weight of its own
Why it can mislead: None of the five methodologies in this glossary lists yield as an indicator, so a claim that a ranking rewards yield describes a formula that is no longer in use. Admissions offices still track it for their own planning.
What does peer assessment score mean, and who fills the survey in?
Peer assessment is the single largest indicator in the 2026 U.S. News Best Colleges formula at 20 percent. Presidents, provosts and deans of admissions rate the undergraduate academic quality of other schools in their own ranking category on a five-point scale running from marginal to outstanding, and anyone who does not know a school well enough is told to mark "don't know" or leave it blank. Blanks do not count for or against.
The sample is smaller than the weight suggests. U.S. News administered the most recent surveys in spring and summer 2025 to 4,591 academics, and 30.8 percent responded. That is roughly 1,410 people, spread across every category. Response was higher where the categories are small and famous: 39 percent for National Universities and 48 percent for National Liberal Arts Colleges.
Two consequences follow. First, the score is a two-year rolling average, so a school that changed sharply this year will see the change diluted by half. Second, at 20 percent, other administrators' opinions are worth two and a half times everything a school spends per student, which carries 8 percent. If a campus press release attributes a rank move to new facilities, check whether the peer score moved instead.
What is graduation rate performance, and how is it worked out?
Graduation rate performance is worth 10 percent of the U.S. News score. It does not ask how many students graduate. It asks how many graduated compared with the number a regression predicted from the intake, and scores the ratio of actual to predicted, averaged over four entering cohorts.
The prediction is built from spending per student, the share of undergraduates receiving federal need-based grants, entering test scores and high school class standing. For National Universities only, the share of degrees awarded in fields on the Department of Homeland Security STEM list is added as a fifth variable. High school class standing comes from old U.S. News surveys because it is not collected anywhere else.
Worked example (illustrative figures, not real schools)
School A graduates 78 percent of an entering class within six years. The model predicted 74 percent from its spending, intake and test profile. Its ratio is 78 divided by 74, which is 1.054, so it overperforms.
School B graduates 92 percent against a prediction of 95 percent. Its ratio is 92 divided by 95, which is 0.968, so it underperforms. School B graduates 14 percentage points more of its students than School A and still scores lower on this indicator.
The trap in the prediction
Spending per student is one of the inputs to the prediction, and it is also a separate 8 percent indicator. A wealthy school is therefore expected to graduate more students, and gets no credit on this indicator for meeting that expectation. Reading the two indicators as if they both reward money double counts what is actually a control variable.
What does value added earnings mean for a college?
Value added earnings is an estimate of how much a college lifts its graduates' pay above what the same students would have earned somewhere else. In the WSJ and College Pulse 2026 ranking, salary value impact is worth 33 percent of the total score and is estimated from students' exam results before college and the cost of living in the college's state. A second measure, years to pay off net price, is worth 17 percent.
The published value added to graduate salary figure is a simpler thing: the difference between the median earnings of the school's graduates and the median earnings of high school graduates, blended between the college's own state and the whole country in proportion to its in-state and out-of-state mix.
Checking the payback formula against the publisher's own numbers
Years to pay off net price is average net price multiplied by four, divided by value added to graduate salary. Two figures published by WSJ and College Pulse for its 2026 top two make the arithmetic checkable.
| School | Average net price | Value added to salary | Four-year cost | Payback we compute | WSJ prints |
|---|---|---|---|---|---|
| Stanford University | $12,136 | $94,725 | $48,544 | 0.51 years | 6 months |
| Babson College | $38,876 | $96,561 | $155,504 | 1.61 years | 1 year, 7 months |
Both reproduce. That is worth knowing because the payback number is extremely sensitive to net price: Babson's salary uplift is about 2 percent higher than Stanford's, yet its payback is more than three times longer, purely because its net price is 3.2 times higher. If you want to run the same comparison on your own two shortlisted schools, the college cost comparison calculator takes four-year costs directly.
Figures above are quoted from the WSJ and College Pulse 2026 Best Colleges ranking for the two top-ranked schools, linked in Sources.
What does PCP mean in ARWU, and why does it move small universities?
PCP is per capita academic performance, worth 10 percent of the ARWU score. ShanghaiRanking takes the weighted scores of the other five indicators, Alumni, Award, HiCi, N&S and PUB, and divides that total by the number of full-time equivalent academic staff. It is the only place in the ARWU formula where institutional size is corrected for.
Worked example: two universities each finish with a weighted total of 60 across the five volume indicators. One employs 1,200 academic staff, the other 6,000. Dividing gives 0.050 and 0.010, a factor of five apart, from identical research output. That gap is then scaled so the best institution scores 100 and everything else is a percentage of it.
Two cautions. ShanghaiRanking obtained staff numbers for a named list of countries; where a country's staff numbers cannot be obtained, the average staff count of the world top 1,000 is substituted for every institution in that country, so the denominator can be an estimate. And 90 percent of the ARWU score is still raw volume, so PCP corrects for size in one indicator out of six.
What is the Common Data Set, and which rankings draw on it?
The Common Data Set is a set of agreed item definitions, produced by colleges working with the three publishers behind the initiative: the College Board, Peterson's and U.S. News. Its purpose is to make one question mean the same thing on every publisher's survey and to cut the reporting burden on colleges. Department of Education definitions are used as a guide when items are developed.
The point students miss is in the standard's own wording: the CDS is a set of standards and definitions of data items, not a survey instrument and not a database. Each publisher's own survey borrows some CDS items and adds proprietary ones, so no two publisher surveys contain the same set. Colleges are urged to follow the definitions and to publish their answers, but publication is a convention, not a rule, and there is no central audit.
Where it shows up in the 2026 U.S. News formula:
- Student-faculty ratio was reported under the 2024 to 2025 CDS instructions, which exclude faculty teaching only in stand-alone graduate and professional programs.
- The full-time faculty share followed the 2023 to 2024 CDS instructions, which exclude preclinical and clinical medicine faculty and administrators such as deans, librarians, registrars and coaches.
- Retention and graduation questions in the U.S. News survey use IPEDS definitions, because those match the CDS definitions.
- Faculty salaries follow American Association of University Professors definitions rather than CDS ones, with twelve-month salaries annualised to a nine-month basis by multiplying by 0.818.
U.S. News says the CDS faculty definitions are more representative of who actually teaches undergraduates than the IPEDS equivalents. That is why a school that skips the U.S. News survey and gets scored from IPEDS is being measured on a wider group of staff than its neighbour on the same page.
Which ranking terms were dropped, and in which edition?
Old advice ages badly because the vocabulary changes faster than the articles quoting it. Each row below was confirmed against the publisher's methodology page for the edition named, using archived copies where the current page no longer mentions the factor.
| Indicator | Publisher | Last edition it counted | What the publisher said |
|---|---|---|---|
| Alumni giving rate | U.S. News | 2023 Best Colleges | One of five factors removed for the 2024 edition to focus on outcomes and on data every school reports |
| Undergraduate class size | U.S. News | 2023 Best Colleges | Removed in the same group of five |
| High school class standing | U.S. News | 2023 Best Colleges | Removed in the same group of five, but still used inside the graduation rate prediction |
| Share of faculty with a terminal degree | U.S. News | 2023 Best Colleges | Removed in the same group of five |
| Share of graduates who borrowed | U.S. News | 2023 Best Colleges | Removed in the same group of five, leaving only the debt amount |
| First-generation graduation rate | U.S. News | 2024 Best Colleges | Introduced for National Universities in 2024, removed for 2025 after the Education Department flagged limits in the source data |
| First-generation graduation rate performance | U.S. News | 2024 Best Colleges | Removed for the same reason |
| Acceptance rate | U.S. News | Before the 2022 edition | Not among the nine broad areas listed for 2022, and named in 2026 as no longer used |
| Study abroad | THE | Never counted | Sits inside International Outlook at 0 percent, pending data quality and consultation |
The practical test: if an article tells you a ranking rewards small classes, high alumni giving or a low acceptance rate, it is describing the 2023 U.S. News formula or older. More of these misreadings are collected in the piece on college ranking myths.
Why do "value", "outcomes" and "best" mean different things on each list?
These three words carry no shared definition. Each publisher attaches its own arithmetic, and the arithmetic is not close. Reading the row across shows why two lists can disagree about the same school without either being wrong.
| Word | U.S. News | QS | THE | WSJ and College Pulse |
|---|---|---|---|---|
| Outcomes | Student outcomes, 52 percent: completion, retention, debt and earnings | Employability and outcomes, 20 percent: employer survey 15 plus employment outcomes 5 | No outcomes pillar at all | Student outcomes, 70 percent: salary impact 50 plus graduation rate impact 20 |
| Value | A separate Best Value list. In the main formula the nearest thing is median federal loan debt at 5 percent | No value measure | No value measure | Years to pay off net price, 17 percent, computed as four years of net price divided by salary uplift |
| Best | 52 percent outcomes plus 20 percent peer opinion | 45 percent surveys plus 20 percent citations per faculty | 59 percent research across two pillars | 70 percent outcomes plus 20 percent student survey |
How to use the row
Pick the definition that matches your question, then use only that list for it. If your question is whether a degree pays for itself, WSJ has an explicit answer at 17 percent of its score and QS and THE have none. If your question is research strength, ARWU and THE answer it and no student survey does.
Common questions about ranking vocabulary
What does peer assessment score mean?
Peer assessment is the U.S. News survey in which presidents, provosts and deans of admissions rate the undergraduate academic quality of other schools in their own ranking category on a five-point scale from marginal to outstanding. It is worth 20 percent of the 2026 Best Colleges score, the single largest indicator, and it is a two-year rolling average. U.S. News sent the most recent questionnaire to 4,591 academics and 30.8 percent responded.
What is graduation rate performance in the U.S. News rankings?
Graduation rate performance compares a school's actual six-year graduation rate with the rate a regression predicted for its intake, then divides actual by predicted. The prediction controls for spending per student, the share of undergraduates on federal need-based grants, entering test scores and high school class standing, plus the share of STEM degrees for National Universities. It is worth 10 percent of the 2026 Best Colleges score.
What does PCP mean in ARWU?
PCP stands for per capita academic performance. ShanghaiRanking takes the weighted scores of its other five ARWU indicators and divides them by the number of full-time equivalent academic staff, which is worth 10 percent of the total. Where staff numbers for a country cannot be obtained, the average staff count of the world top 1,000 universities is used instead, so the denominator is sometimes an estimate.
What is value added earnings for a college?
Value added estimates how much a college lifts its graduates' pay above what the same students would have earned elsewhere. The Wall Street Journal and College Pulse ranking builds that estimate from students' prior exam results and the cost of living in the college's state, and salary value impact is worth 33 percent of the 2026 score. The published value added to graduate salary figure is a simpler comparison against high school graduate earnings in the same state.
Is the Common Data Set the same as IPEDS?
No. IPEDS is the federal survey system that institutions taking federal student aid must report to, and its data is public. The Common Data Set is a set of agreed definitions maintained with the College Board, Peterson's and U.S. News so that one question means the same thing on every publisher's survey. It is a standard, not a database, and publication of a CDS file is voluntary.
Sources
Every weight and definition on this page was read from the publisher's own methodology page in September 2026. Where a current page could not be reached directly, the archived copy of that same official page is linked and the snapshot date is given. Editions change every year, so check the publisher before relying on a figure.
- U.S. News: How the College Rankings Were Calculated (2026 edition weights; read from the Internet Archive snapshot of 3 September 2026)
- U.S. News: A More Detailed Look at the Ranking Factors (indicator definitions; read from the Internet Archive snapshot of 29 August 2026, with the 2022, 2024 and 2025 editions read from snapshots of 28 December 2021, 20 September 2023 and 24 September 2024)
- ShanghaiRanking: Academic Ranking of World Universities 2026 methodology
- QS: A guide to the methodology for the QS World University Rankings, updated 20 May 2026
- Times Higher Education: World University Rankings 2026 methodology
- WSJ and College Pulse: Best Colleges in the U.S. 2026, methodology and published school figures (read from the Internet Archive snapshot of 1 September 2026)
- Common Data Set Initiative: home page and standards statement
- Carnegie Classification: 2025 Institutional Classification methodology
- Carnegie Classification: 2025 Research Activity Designations
- U.S. Department of Education: College Scorecard data glossary (read from the Internet Archive snapshot of 1 June 2026)
CampusBreeze makes free study tools and guides. Nothing here is affiliated with any ranking publisher.