Education records degrees and qualifications are the most stable part of a career profile and the most quietly inconsistent. The shape of an education entry barely changes around the world — an institution, a qualification, a field of study and a pair of years — yet almost every attempt to compare two such entries directly runs into the fact that the words inside them mean different things in different places.
This article covers what an entry holds, why a degree and a qualification level are not the same axis, how the timeline constrains the dates, and what to check before trusting an education field in a test record.
What does an education record usually hold?
Four things, and the fourth is the one that causes most of the trouble: the institution, the qualification or level reached, the field of study, and the year the study finished.
Most real records also carry a start year, which is useful for telling a four-year course from a part-time one, and most carry nothing else. Grade, ranking, thesis title and language of instruction appear often enough to need a home in the schema, and rarely enough that they cannot be treated as required.
A record should lead with the highest qualification reached rather than with the first one, because that is the fact a reader is looking for. The earlier entries exist to explain when study happened, and in a career record they also exist to constrain when work could have started. That is the property people forget: an education entry is not just a fact about the past, it is a bound on the entries above it.
Are degrees and qualification levels the same thing?
No. They are two different vocabularies answering two different questions, and systems which collapse them lose information that cannot be recovered.
A degree vocabulary describes an academic award. A qualification-level vocabulary describes a rung in a framework — a way of placing any learning, however it was obtained, on a common scale. The two overlap heavily in practice and are not interchangeable, because plenty of learning sits on the second scale without producing anything from the first. A completed apprenticeship, a vocational diploma, a professional certification route and a military technical course may all occupy the same level while sharing no award with each other.
The confusion is amplified by the fact that different regions lean on different axes. Some education systems describe everything in terms of the award, and readers there will look for a degree even where none exists. Others describe everything in terms of a level or a certificate, and readers there expect a level rather than a title. A form designed from one tradition will be unable to record a perfectly ordinary qualification from the other.
The practical consequence for anyone building hiring software is that the qualification field needs to be able to say “this is a level in a framework” as well as “this is an award”, and needs to record which framework the level came from. Otherwise the field is a string, and the string will be compared to another string, and the comparison will be wrong.
Why are field and institution names hard to compare?
Because they are proper nouns, and proper nouns are not stable identifiers. The same subject is named differently by different institutions, differently again in different regions, and differently across the decades as curricula are reorganised.
Three patterns account for most of the mismatch. Naming conventions split the same discipline along different lines, so a course that is one department in one country is two in another. Translation introduces near-synonyms that carry slightly different scopes, and the closest available word is adopted permanently even when it is a poor fit. And institutions rename themselves, restructure or merge, so a perfectly accurate entry from twenty years ago may name a faculty that no longer exists in that form.
The same is true of institution names, and there is a further trap: transliteration. Any institution written in a script other than the Latin one has at least two reasonable Latin spellings, and different systems will have adopted different ones. A search that treats the institution as a string will miss matches that a human finds instantly.
The resolution is not to force agreement but to keep three things: the name as the person wrote it, a normalised form for matching, and a reference to whatever mapping the normalised form came from. Any automatic comparison should run on the normalised value and any display should show the original.
| Field | Stable across regions | Needs mapping to compare |
|---|---|---|
| Institution name | Structure only | Spelling, transliteration, renames |
| Qualification | No | Award vocabulary versus level frameworks |
| Field of study | Structure only | Discipline naming and translation |
| Graduation year | Yes, as a year | Nothing, once precision is agreed |
| Start year | Yes, as a year | Interpretation of a gap within study |
What range of graduation years is reasonable?
A graduation year has to satisfy two bounds at once, and both of them are relationships rather than constants.
It cannot be later than today, since a qualification that has not been completed is a different fact and should be recorded as in progress rather than dated in the future. And it has to be consistent with the start of the first role: study normally finishes before full-time work begins, and a career record which reports a graduation year after the first job started is describing either a career change or a defect. A career change is perfectly ordinary and is exactly why the check should surface the pair for a human rather than reject it.
The awkward cases are the interesting ones. Further study in mid-career is common and produces a graduation year that sits far above the first role. Part-time study produces a span longer than a full-time course, so any rule about duration has to tolerate it. And a qualification listed as in progress has no graduation year at all, which is a state the schema must be able to represent rather than approximate with a guess.
For developers: education fields and their checks
Model the education entry so that every derived value has exactly one home. Keep the graduation year as a year, do not store the duration, do not store the age at graduation, and never store a computed level alongside the award it was computed from.
Four checks belong at the point of entry. The graduation year must not be in the future for a completed entry. The start year must not precede the graduation year. The qualification must either be an award from a known vocabulary or a level with a recorded framework, and never both at once. And an entry marked as in progress must carry no graduation year at all — the same rule that applies to an open-ended role, for the same reason.
Everything else belongs to review rather than rejection. A qualification at an unusually young age is rare and legal. A long course is unusual and ordinary. A graduation year after the first job is a career change far more often than it is a mistake. Encoding any of those as a validation failure will block real people out of real forms, and the work history side of the same record is where the second half of each relationship lives.
Sketch the institution names with obvious placeholders. The institutions and qualifications shown in test records here are stand-in values, invented so that form handling and parsing can be exercised, and they stand as proof of nobody’s education. What a parsed document should recover from an education section is a related problem, and it is usually where a mismatch between the stored year and the written one shows up first.
Next steps
Take one education entry out of your test data and write down, in plain words, what each field claims. If the qualification field cannot distinguish an academic award from a level in a framework, that is the first thing to fix, and the career profile generator will produce entries covering both for you to look at. Then check the pair of the graduation year and the first role’s start month across the whole set — it is the cheapest consistency check available and it catches defects nothing else does.