How U4S verifies tuition, and why it usually publishes no number

University pages follow the same two-tier logic as countries and cities, with one addition. The prose on a university page is orientation: it describes what a student has to establish and which office holds the answer, and it contains no claim about the institution — no size, no age, no character, no reputation, and nothing resembling a ranking. Alongside it sit facts that were actually retrieved: the official website, the source list, and the date each source was read.

Tuition is treated more strictly than anything else on this site, because it is the only thing here that can cost a reader money if it is wrong. A figure is published only when the institution’s own page states all of it: the amount, the currency, the academic year or intake it applies to, the study level, the student category, the period it covers, and the conditions attached. Miss any one of those and the figure is not published. In its place the page says that tuition has not been reviewed and sends the reader to the institution.

That rule is enforced rather than intended. Prose and data are separate files, and prose cannot introduce a number, a year or a source at all; a build check rejects any university page whose narrative contains a monetary figure, a year, ranking language or a stated admission requirement. A fee item that cites a source the site could not retrieve is rejected too, because an unreachable citation supports nothing.

Institutions in the United States publish the money side in a shape the rule here cannot accept, and it is worth explaining rather than leaving as an apparent gap. What is commonly published is a total cost of attendance, a net price estimated from an individual student’s circumstances, or an interactive calculator that asks about your situation before showing anything. Each of those is a legitimate and useful document. None of them is a tuition rate.

The distinction matters more than it sounds. A tuition rate is the charge for instruction, and it only means something beside the labels it carries: the academic year, the study level, and the residency category, since public institutions commonly charge differently by residency and rates differ by school and programme within one university. A cost of attendance total adds an allowance for accommodation, food, materials and everyday spending, and exists for budgeting rather than billing. A net price is an estimate after aid, particular to a household. Publishing any of the second three under the name of the first would misdescribe it by a wide margin.

So every United States profile records tuition as not reviewed, and each carries a next step naming what to look for on that institution’s own student financial services pages. That is a statement about what U4S has verified, not about the institution: the figures exist and are published, they simply are not published in a form that satisfies the field set on the editorial policy page.

Fee pages across these three systems are, on the whole, straightforward to find and plainly written. Very little from them still clears the rule above, and the obstacle is different in each case. It is worth naming all three, because a reader who finds an amount elsewhere and cannot find it here deserves to know that it was set aside deliberately rather than overlooked.

In the United Kingdom and in Ireland, what a student pays turns first on fee status, and fee status is settled by residence history and previous study rather than by nationality alone. One institution therefore publishes several charges for the same course — one per category, and frequently more that vary by programme and by study level. An amount lifted out of that structure without its category and its level attached is not a simplified fact but an unattributable one, and it may well belong to a category the reader is not in. Irish pages add a further step, since what a student owes is commonly assembled from more than one component rather than presented as a single charge.

Swiss institutions commonly publish a plain per-semester amount, often with an additional charge for students whose previous qualification was completed elsewhere, and with mandatory contributions listed separately. What is frequently absent is the academic year the amount belongs to. An amount without its year cannot be checked, because a reader has no way of telling whether it is current or several years out of date, so it stays unpublished on the same grounds as any other incomplete figure. Fee structures in this system are also set at institution and cantonal level rather than nationally, which means one institution’s published amount carries no implication whatever about another’s.

The practical result is identical in all three cases, and it is the result the rule was written to produce: the profile records tuition as not reviewed and names what to look for on the institution’s own fee pages. A reader doing that work themselves will find the same checks set out in the guide to checking official tuition fees, and the reason categories differ at all in the guide to domestic and international fee categories.

The largest group of institutions on this site now sits inside the European Union, and their fee pages are, as a rule, public, plainly written and easy to find. Almost none of them produces a figure this site can publish. The reasons are worth setting out concretely, because a reader who finds an amount elsewhere and cannot find it here is entitled to know it was set aside on purpose.

The first obstacle is that the amount often depends on a category before it depends on anything else. Institutions in several of these systems publish more than one charge for the same programme, and which one applies to a given student is settled by residence history, by previous study, or by a status the institution assigns during admission — not by nationality, and not by anything a reader can work out from a web page. An amount lifted out of that structure is not a simplified fact. It is an unattributable one, and it may belong to a category the reader is not in.

The second is that the shape of the amount varies. It may be quoted per credit rather than per year, which changes entirely what a total means. It may cover a semester rather than a year. It may be assessed on a household’s circumstances rather than fixed, in which case no single published number exists at all. It may be set by an authority above the institution, or, at an institution that sets its own, by the institution alone. Programmes taught in different languages are sometimes priced separately from one another. None of that is a complication to be smoothed over; each of those distinctions is the difference between a number that helps and one that misleads.

The third is that a great many pages simply do not carry an amount. Retrieval during this pass repeatedly ended in a chain of navigation pages, each linking onward to another, with the actual figures living in a downloadable regulation document at the end of it. Others print a charge that is not tuition at all — an application charge, an enrolment or insurance charge, a student body contribution, an administrative fee — sometimes on a page headed with the word fees. Publishing any of those under the name of tuition would misdescribe them by a wide margin, and one of them appearing beside the word tuition is exactly the kind of near-miss that survives a careless reading.

The fourth is the one that disqualifies the most otherwise-complete figures: a clear amount published without the academic year it belongs to. That is not a small gap. Without its year a reader cannot tell whether they are looking at something current or something several revisions old, and the page gives them no way to find out. The rule is applied the same way here as it was for institutions elsewhere on this site, and it produces the same result for the same reason.

So EU profiles record tuition as not reviewed and name what to look for on the institution’s own fee pages. That is a statement about what U4S has verified rather than a comment on any institution: the figures exist and are published, they are simply not published in a form that satisfies the field set. The guide to checking official tuition fees sets out the same checks for a reader doing this work themselves, and the category question is the one to settle before any amount means anything.

Institutions in Canada and Australia generally publish fees at programme level rather than for the institution as a whole, which suits the publishing rule better than a headline figure would — but only if the programme, the student category and the year travel with the number. A rate lifted out of a programme table and presented as the institution’s fee describes a course the reader may never take. The unit matters just as much: an amount per unit or per subject is not an amount per year, and turning one into the other means assuming a study load the fee page never stated, an assumption that would be invisible in the result.

Pages in both systems also list student services and amenities charges beside tuition. Those are real and worth knowing about, and they are not tuition. U4S records them separately where a page states them and never folds them into a tuition figure, because a total assembled here would be a number no institution published.

In the Nordic, wider European Economic Area and Baltic systems the most common error available is a different one: assuming, from something read elsewhere, that study is free or that it is not, and treating that assumption as a fact about a particular institution and a particular student. U4S publishes no country-level statement about whether tuition applies. Whether a given student is charged can turn on the category they fall into, the language their programme is taught in and their level of study, and only the institution and the relevant official authority can settle it.

Two further traps recur across those systems. Several publish a compulsory semester registration or student body charge on the same page as tuition, sometimes under the same heading; it is not a tuition rate. And an amount may cover a semester rather than a year, stated once in a heading and not again beside the figure. Where the period is not given beside the amount, U4S treats the field as missing rather than inferring it from context, because doubling a semester figure into a year is an assumption rather than a reading.

Programmes taught in different languages are frequently priced separately in these systems, which means a single institutional figure may not exist at all. Where that is what a page shows, the honest form is a range with the reason attached, or no figure — never a representative number chosen from among several. The checks a reader can run themselves are the same ones set out earlier on this page.

Across the education hubs of East and Southeast Asia the same rule meets two further shapes. Many institutions publish an official page in English alongside an official page in another language, and the two are not always the same document; where only one of them was read, U4S records which, rather than implying the whole site was checked. And an entrance or admission charge is very commonly listed directly beside tuition, sometimes in the same table, while an amount that looks annual may cover a semester. Neither of those can be settled by looking at the number, only by reading what the page says beside it.

Publishing a fee means opening the document the institution published it in, and the documents come in a small number of recognisable shapes. An HTML fee page states amounts in prose. A fee table states them in rows and columns. A downloadable schedule or regulation carries them as an annex among other provisions. An interactive lookup asks about the reader before showing anything at all. And a navigation hub, which is the most common shape of all, is a page about fees that states none, linking onward to another page that does the same.

Each shape fails differently, so U4S records which one it met rather than recording only that it found nothing. A retrieval that stops at a navigation hub needs someone to follow the chain; one that stops at an interactive lookup needs a published schedule to be found instead, because driving the lookup with invented inputs produces a figure that describes nobody. Recording the obstacle is what stops the next attempt repeating the last one.

Downloadable documents get a further step, and it is the step that turns out to matter most. A regulation retrieved during this work gave up almost all of its content streams and tens of thousands of characters of readable text, including the annex listing rates by category and level for a clearly named year. Every field the publishing rule asks for was present, and it was still not publishable. The next section sets out why, and corrects the first explanation this page gave for it.

That is why an extraction is treated as a candidate and never as a finding. The retrieval tool in this repository fetches a document, works out what kind it is, tries to get text out of it, and reports which of the required fields the text contains. It publishes nothing. A person then reads the document and decides, and the record of that decision carries what was observed in plain words, so a later reviewer can tell whether the reading was right without fetching the file again.

Two requirements disqualify more otherwise-complete documents than anything else. The first is the academic year: an amount without the year it belongs to cannot be checked by the person relying on it, and adjacency makes this worse rather than better — one fee page retrieved during this pass carries a navigation label naming last year directly above a table heading naming this one. The second is the student category. A page seen during this pass states a clear annual amount by level and attaches no category to it anywhere, so there is no way to know which students it describes, and a reader who is not in that group would plan against a number that was never theirs.

Programme-level variation is the third recurring reason a page does not yield a single figure. Institutions commonly publish a band, a range by field of study, or a rate that differs per faculty, and a surcharge may sit on top for a named set of programmes. Where that is what the document says, U4S publishes the range and names it as a range, rather than picking a representative figure — a representative figure is an invention wearing the clothes of a source. Working through several dozen of these pages also settled what actually decides an amount: not the fields around it but what the page calls it, and the label that counts is the nearest one. An early version of this check asked only whether the word tuition appeared somewhere nearby, and it was wrong on every real page it met — reading a scholarship total, a proofreading quote, a government grant and a multi-year combined cost as tuition rates, because in each case the word sat a line or two away. A label sits beside the number it labels. Where a disqualifying label sits close to the amount it wins outright, because an amount headed as an estimated tuition is an estimate however adjacent the second word is.

Finding an amount and a category in the same document is not the same as knowing they belong together. Almost every fee schedule is a table, and a table carries meaning in its geometry: this number is a master’s rate because it sits on the master’s row and under the column headed by a student category. Extract the words without the geometry and every one of them survives while the thing they meant does not.

So the extractor used here recovers the position of every piece of text and rebuilds rows and columns from those positions, rather than reading the document as a stream. Where the rows and columns come back, an amount can be tied to the row it sits on. Where they do not, the document is downgraded and no figure is taken from it. The aim is not better reading. It is more honest failure, because the alternative failure mode is a confident wrong number in front of someone deciding where to spend several years and a large amount of money.

This is also a correction. An earlier version of this page said a regulation had been rejected because a text-stream reading attached one student category’s rate to an amount belonging to another. Re-reading the same document with positions recovered shows that was wrong: the rows rebuild, and each label does sit with its own amount. The figures that looked misplaced were simply what the document prints. The conclusion had been drawn from an amount looking implausible rather than from evidence that anything had been misread, and a number looking wrong is not evidence that extraction failed.

The document was still not published, for reasons the corrected reading makes clearer rather than weaker. Its amounts are right-aligned and drawn a glyph at a time, so reading a value means reassembling it from several fragments — recoverable by a person, not safely by a rule. Other pages of the same file carry amounts whose structure does not rebuild at all, and a document is judged by the weakest page that carries money rather than by its best. And the annex is one of at least two tariff tables in the same file whose difference has not been established, which is a question about meaning that no extractor can settle.

That last point is the general one. Extraction produces candidates, never findings. A machine can say that a page holds a year, a category, a level and an amount, and that the amount sits in a readable row. It cannot say whether the table it read is the one that applies to the reader, whether the institution means the figure as a rate or as an estimate, or whether a second table elsewhere supersedes it. Those are read by a person, and the record of that reading is kept beside the figure so it can be checked later without opening the file again.

The country pages for this part of the world are orientation pages about studying and living somewhere, exactly like every other country page on this site. They are grouped for navigation by geography and nothing else. U4S does not use political or historical bloc labels to organise student-life pages, because a label of that kind carries an argument, and a reader looking for how enrolment works should not have to accept one on the way in.

They also leave things out, deliberately and without commentary. A student page has one job: to help somebody work out what living and studying somewhere would involve, and where to get the answers only an institution or an official authority can give. Political context does not help with that, it dates badly, and it reads differently to two people who need the same practical information. Anyone weighing a move should of course inform themselves about where they are going; U4S is not the publisher for that part.

Official source verification varies here more than elsewhere, and the record says how. Some institutions publish a full site in English, some publish in the local language only, and some did not serve a page to an automated request at all. Where only a landing page was read, the profile says so; where nothing was read, the profile says that instead of implying a check that did not happen. A domain resolving is not a page being read.

Tuition follows the same rule as everywhere else on this site and produces the same result: no figure is published unless the institution’s own page supplies every required field and the amount can be tied to the category, level and year it belongs to. Programmes taught in different languages are frequently priced separately in these systems, and a place funded by the state and a fee-paying place are charged differently, so a single institutional figure often does not exist to publish.

No United States profile on this site publishes a tuition figure, and the reason is consistent across every institution examined. What is published is usually a cost of attendance, which adds accommodation, food, materials and everyday spending to instruction and exists for budgeting; a net price, which is an estimate after aid particular to one household; or an interactive estimator that asks about a reader before showing anything. Each of those is a legitimate and useful document. None of them is a tuition rate.

Where a rate does exist it is rarely one number. It can differ by residency, by level, by the school within the university, by the programme, and by the academic year, and a figure detached from those is not a simplified fact but an unattributable one. A reader who is not in the category it belonged to would plan against a number that was never theirs.

So the profiles record tuition as not reviewed and name what to look for on the institution’s own student financial pages. That is a statement about what U4S has verified rather than a comment on any institution: the figures exist and are published, they are simply not published in a form that satisfies the field set.

Institutions here commonly publish a schedule that lists a number of components together — an admission or one-time charge, a recurring academic charge, examination and registration charges, and hostel or mess charges where residence is involved. Only one part of such a schedule is tuition, and reading the total as a tuition rate would overstate it substantially while attaching charges to students who may not incur them.

The same requirements apply as everywhere: the component that is tuition has to be identifiable as tuition, the amount has to stay attached to the programme, the student category and the period it covers, and the academic year has to be stated. Where a schedule lists components without saying which is which, or without a year, nothing is published and the profile says which document was opened and what was missing from it.