Proposal talk:ISCED 2011 Level of Education

Including orientation

I'm diving deep into this matters, as I have to build a data collection tool this data for my work. While reading the ISCED and doing practical mapping, the 'orientation' of the program turns out to be quite important. Therefore, I propose to have isced:2011:level=vocational_lower_secondary, isced:2011:level=general_lower_secondary, isced:2011:level=vocational_upper_secondary and isced:2011:level=general_upper_secondary

Similar, we can have a '(professional|academic)_(bachelor|master)' values too Pietervdvn (talk) 14:32, 8 June 2022 (UTC)

What about special education, adult education and evening schools

This being said, I was wondering how other types of schooling can be fit in, e.g. a special education school. I skimmed through the [Official Unesco definitions] and point 132 (page 31) states that isced-level 1 includes programmes suited to individuals with special needs if the programme provides systematic teaching and learning in the fundamental skills of reading, writing and mathematics, irrespective of the age of the participant . ~~This means that using tagging such as isced:2011:level=primary_education gives the impression that this is for kids from about 6 till 12; but isced:2011:level=1 might be applicable for a school to teach basic literacy to adults whom just never had the chance to learn how to read during their childhood.~~ This is not what the isced says, but the definitions in this proposal should make it very clear that this is a _skill level_ that is obtained and that, while there are typical ages to these schools, exceptions exists.

In a similar vein, in Belgium we have 'evening schools' where one can follow a part-time course - e.g. if one wants to retrain during the evenings while still working a normal job. These do sometimes offer official diploma's. I guess these could use isced:level:2011=post_secondary? (By know I'm sure this is the correct answer, but this should be phrased a bit better)

At last, what should be done with special education? E.g. how to tag a primary school that exclusively has special need-kids?

I do think that, for these special cases, a tag similar to social_facility:for=* should be used, e.g. school:for=adults,school:for=learning_disabilities, school:for=autism, ... If this tag is missing, we could however assume that the school is for normal students of the appropriate age. Pietervdvn (talk) 15:08, 7 June 2022 (UTC)

isced:2011:level=* vs. isced_2011_level=*

I choose isced:2011:level=* as name of the key, in order to keep it similar to isced:level=*, but would be willing to change it to isced_2011_level=* or similar. Are there strong opinions on that? --Skorbut (talk) 19:03, 16 May 2022 (UTC)

I like isced:2011:level=* way better. Underscore is mostly used as replacement for spaces, e.g. `highway=parking_aisle`, but as `2011` is a subcatagory of `isced`, ':' is more appropriate Pietervdvn (talk) 14:48, 7 June 2022 (UTC)

How is this to be verified on the ground?

Please add a section to the proposal explaining how these values are to be verified on the ground, by mappers. That has been a continuing problem with ISCED tagging, and it's not (yet) clear how this proposal would fix or improve that. JesseFW (talk) 13:51, 15 May 2022 (UTC)

@JesseFW: Thank you for taking the time to read the proposal draft. As stated in the description, the main goal of this proposal is to "remove the ambiguity associated to the tag isced:level=*", i.e. making sure that a given value from an ISCED level OSM tag can be associated to exactly one ISCED level and vice versa. (Quite a basic requirement, but unfortunately not given (anymore) for isced:level=*.) But still, mappers should obviously be able to associate a specific school they are encountering on the ground to the correct ISCED level. While most of the tags/values in OSM have a (usually) simple and globally usable definition, this works badly for ISCED. Why is that? In order to accommodate the varieties of about 193 different countries, the definitions for the ISCED levels contain a lot of ifs and buts. Let's look at ISCED level 0 (early childhood education) (usually called "kindergarten" or similar) on page 26 of the ISCED 2011 document : There are 4 main criteria that must be met

1. Educational properties of the programme
2. Institutional context
3. Typical target age of children for whom the programme is designed
4. Programme intensity/duration

and 3 subsidiary criteria that usually are met:

1. Staff qualifications
2. Existence of a regulatory framework
3. Typically not part of compulsory education

Each of these criteria is described in detail in its own paragraph, thereby taking up about one and half page when printed out.

I therefore propose to break with the usual OSM habit of providing definitions and go with the country-specific list of school programmes that can be found at http://uis.unesco.org/en/isced-mappings . In my experience they are conclusive and they take up significantly less space than the ISCED definitions. See the examples (USA and Germany) how that would look like. I estimate that should cover something like 99.9% of all (official) schools in an average country. Do you happen to know an example from your country where this list would not suffice in determining the ISCED level?

Sidenote: If you actually stand on the ground in front of a school, it may be kind of hard to determine what is taught there, but this is not a problem specific to ISCED but generally to mapping levels of education.

--Skorbut (talk) 19:25, 15 May 2022 (UTC)

The problem is that interpreting a list of definitions (or even a list of county-specific category names) regarding their applicability to a specific institution is wildly out of scope of the sort of thing generally asked of OSM editors. This sort of thing seems much more appropriate for Wikidata (or even better, a separate, authoritative source compiled by relevant educational accredation/licensing organizations). Expecting OSM editors to make these judgements seems basically wrongheaded, and I'm sorry to hear that this proposal doesn't (so far) seem to be making an attempt to fix that. JesseFW (talk) 03:24, 16 May 2022 (UTC)

@JesseFW: You wrote: The problem is that interpreting a list of definitions (or even a list of county-specific category names) regarding their applicability to a specific institution is wildly out of scope of the sort of thing generally asked of OSM editors. Well, I disagree and would generally trust mappers to be able to do such an assignment. Having 9 different items with most of them usually not necessary to be considered is really not that a difficult situation... But anyways, since the aim of this proposal is to undo the damage that time did to isced:level=* and you seem to not like ISCED in general, I have a hard time finding out how I actually could improve the proposal. Any pointers?

--Skorbut (talk) 19:03, 16 May 2022 (UTC)

My main suggestion for improving this is to decouple it from ISCED and read up on world-wide education instead of just US and DE. For instance in Nigeria https://en.wikipedia.org/wiki/Education_in_Nigeria they write: The education system is divided into Kindergarten, Primary education, Secondary education, and Tertiary education. Which would not fit your list. The list would also not fit the Dutch education system which I'm familiar with. You can't expect mappers to shoehorn what they see on the ground into ISCED Kaartjesman (talk) 21:09, 16 May 2022 (UTC)

According to the Wikipedia article you linked "The education system is divided into Kindergarten, Primary education, Secondary education, and Tertiary education.". That sounds like textbook ISCED levels _0_, _1_, _2+3_ and _6_. More detailed mappings are described in on the UNESCO website So, pretty verifiable, especially if a specialized editor would show these options. Pietervdvn (talk) 01:54, 10 June 2022 (UTC)

@Skorbut: OK, so the classification in the U.S. happens to be based on grade numbers, making it verifiable by proxy. But the classifications are lossy, completely unrelated to what's on the ground and, in the interest of international harmonization, incompatible with the classifications that people actually use inside the country. Several years ago, this high school's ninth grade moved to a separate freshman campus. Even though it remained a high school, it would've transformed from a lower_secondary;upper_secondary to an upper_secondary, a distinction no renderer or geocoder should care about. This analysis of isced:level=* applies just as equally to this proposal; just replace "Level 3" with "upper secondary" etc. I don't think we can call a classification scheme a good fit if it can only neatly classify 10% of schools.

I really appreciate that you're trying to fix a technical problem with the de facto tagging. The ambiguity has been a bugbear of mine as well. Unfortunately, as much as other commenters have pushed you to come up with keywords, I'm afraid it's just window-dressing for the key's more fundamental problems. I think it's time to formally endorse the use of school=* or country-specific subkeys like school:FR=* as the preferred scheme for school classification. Then if people still want to tag ISCED levels for some reason, all it will mean is "Here's what the school offers", not "Here's what the school is".

 – Minh Nguyễn ^💬 04:33, 17 May 2022 (UTC)

@Minh Nguyen:I interpreted that the 'fundamental problem you see is that ISCED-levels don't map nicely on the various school systems and that we should rather create a `school:<country-name>`-tagging system?

Personally, I do disagree. That would also imply that every country would have their own `highway=*` tags, their own `shop=*` tags, ... For example: our `residential streets` in Europe have different properties and expectations then the ones on say Africa or the US, but still we use the same tagging. Regional differences are accounted for by wiki pages stating what those expectations are - yet at the same time, the definition of such a road is the same worldwide: a road where people live next to. It doesn't matter if they are paved or not, if they take certain kinds of traffic, ...

At the same time, the ISCED manual did a great job looking at all the regional differences and crystallizing the big lines. I read the ISCED as `a primary school is a school where one learns skills such as reading and simple maths. These schools are normally attended by kids of ages around 5 till ages around 12; and take around 7 years (but those ages are but an indication)'

In other words, ISCED elegantly solves the problem of e.g. schools for adults on learning how to read and to do simple maths. (e.g. for adults who never had the chance to read during their childhood, have medical issues, ...). Such a school would get `isced:2011:level=primary`, and an additional `school:for=adults`.

The differences between countries can be coded by editors, which map the general names of the schools onto their respective ISCED-levels.

Furthermore, ISCED:2011 is pretty stable as tags, whereas 'grades' might be not. For example, when my mum started primary education, the first year of it was called '6th grade', the second year was called '5th grade'. By the time she was in her 4th year of primary education, the government decided that counting grades backward was retarded and they reversed the numbering - so my mum was in 4th, 5th and 6th grade again!

And I do agree that it does not neatly map to american schools in many cases, but it does neatly map onto e.g. Belgian schools. Furthermore, a decent definition of schools via ISCED is more viable then the mess we currently have. The school=* is currently a mess, mixing education levels, education subjects and even where the entrance of the school is. grades=* is just as messy. At first glance, it reads as a more finegrained attempt to define "what is educated at what age" (precisely as ISCED does!). This mapping is even more a disaster for mappers who are _not_ familiar with the wiki page. (In belgium, we label the six years of primary education as "grades 1 till 6" and label the six years of secondary education as "grades 1 till 6". Some secondary schools only offer grades 1 and 2 (so roughly from ages 12 till 14). Someone seeing the 'grades'-field will thus fill out "1;2".). In other words - every problem that ISCED has will be amplified by grades=*.

In conclusion: ISCED might _not_ be the perfect solution, but it seems to me as a very workable solution for 90% of the world in 90% of the cases; it has clear definitions, clear mappings which cannot be said about other current tagging. Pietervdvn (talk) 01:54, 10 June 2022 (UTC)

@Pietervdvn: How do you know ISCED is a workable solution for 90% of the world? Which countries besides Belgium have you considered? The ISCED mappings cannot be taken as evidence of effectiveness for OSM's purpose. In fact, UNESCO's ISCED FAQ cautions against this kind of usage. [1] Anyone promoting this approach needs to explain how ISCED tagging can be consistent with the verifiability principle. The mapping to OSM tags is purely unofficial and derivative of anything official in the real world.

You may view school=* as a mess because it isn't fully preplanned like these other schemes. However, with this less structured approach, each local community has the flexibility to determine values that make sense for their local situation. If the Belgium community prefers values that correspond to ISCED 2011's Belgium mapping document, great – take a Belgium-specific set of values to a vote. Maybe that would give people enough confidence to break out of this years-long struggle over school classification.

I don't think we can draw an analogy to country subkeys of shop=* or highway=*. For one thing, designation=* is very commonly used to indicate the local classification of a road apart from the global definition. Moreover, shop=* is not a classification system, so it's irrelevant.

I think you misunderstand grades=*. It isn't a classification system at all. If grade numbers repeat within a school classification, that's no problem as long as grades=* is paired with a school classification, whether tagged as isced:level=* or school=*. It's rather like criticizing min_age=* because it doesn't unambiguously indicate a school classification, something it was never meant to do.

Your anecdote about grade numbers being revised is interesting but rather proves the advantage of grades=*. The public must've been much more aware of that change than the ISCED 2011 revision, so if OSM had been around back then, whatever transition process the educational system undertook would've been mirrored in OSM. There would've been no obligation to continue tagging a school with the grade levels that used to be educated there. It isn't as though your prospective employer uses OSM to calculate your personal educational qualifications. As a counterpoint, look what happened after every school in Puerto Rico was tagged with isced:level=*, and then the government changed the local classification system to one that no longer lines up with ISCED. We probably have to discard all that data, because there's no way to convert it to something accurate.

 – Minh Nguyễn ^💬 02:41, 10 June 2022 (UTC)

@Minh Nguyen: Thanks for taking the time to edit Key:isced:level. Currently I am tending to switch my focus on working on a country-specific school:CH=* or the like. Maybe it's just too hard finding a globally usable scheme that would fit well into the expectation of mappers...

--Skorbut (talk) 19:29, 18 May 2022 (UTC)

I agree with Minh's comments. Let's switch to school=* and similar. JesseFW (talk) 15:00, 17 May 2022 (UTC)

@JesseFW: school=* is an _utter_ mess: the wiki page describes how tagging has happened in the past and already describes three distinct concepts (subjects, age groups and general skill levels + entrances). ISCED-tagging tries to capture one of them pretty cleanly Pietervdvn (talk) 01:54, 10 June 2022 (UTC)

@Pietervdvn: Those orthogonal aspects can be broken out into separate keys. It's not much different than how crossing classification started out, but no one is proposing to avoid all of crossing=* in favor of a repurposed statistical standard because of some deprecatable crossing=* values. – Minh Nguyễn ^💬 02:41, 10 June 2022 (UTC)

@Pietervdvn: The problem I'm bringing up is the lack of verifiability, not whether the existing documentation for school=* is good enough yet. I agree about the similarity to highway classification tagging, and I think the very long and vicious (and continuing) fights over whether or not a particular road is "primary" or not demonstrate precisely the problem with encouraging this sort of difficult-to-verify tagging. But at least the highway=* tags aren't explicitly tied to a particular external (but unverifiable) source, so individual regional communities can come up with their own guidelines for what highway=primary means and where it should be used. That's harder to do with a key that purports to represent an official value (but doesn't actually). JesseFW (talk) 02:57, 10 June 2022 (UTC)

Opinion; ISCED doesn't fit OSM

The previous topic (verification) is not the only reason why ISCED feels like a less-then-ideal match to OSM. When people think about a school in America and when one thinks about a school in Uganda you will have a very different image connected to them. Nobody expects the same level of education (compatible diploma) from them. That is comparing apples and oranges. Great for some specialized education website, not so much for OSM.

What matters on a map, however, is basics. What age of people are (typically) going to attend this. What topic is educated here. This is immediately useful for a map. A real-estate broker will list in its listing the nearest primary school. It won't care about the details listed in ISCED. This looks too complex and I'd much rather have research go into the direction of 'school=*' and 'grades=*' as they are simpler. And simple is better. Kaartjesman (talk) 20:54, 16 May 2022 (UTC)

I agree with with this, as well as the comments from Minh and Jesse. ISCED is fundamentally not a good fit for OSM tagging. I would be opposed to any move that formalizes ISCED as the primary mechanism of tagging the type of school. The table aligning grades to labels is a further papering over of the fundamental issue of creating invented categories rather than just tagging facts on the ground as they are. --ZeLonewolf (talk) 17:38, 17 May 2022 (UTC)

@Kaartjesman: @ZeLonewolf: I do acknowledge that ISCED level has some weaknesses, among them:

1. Quite different school get assigned the same level (could partially be solved by tagging the difference between academic and vocational schools)
2. Splits between ISCED levels might not be where pupils usually change schools (seems to be the case in the USA)
3. No quick definitions

However, it also has distinctive advantages, among them:

1. Internationally known (even outside of OSM)
2. Tradition of usage in OSM (>200k usages)
3. Quite complete lists ready to use for all countries on this planet

It saddens me to see that not only is this seldom given credit for, but also there are no visible efforts by anyone to change this for better, i.e. coming up with a globally usable, cool, easy-to-understand, data-consumer-friendly tagging scheme. (Or did I just miss them?) I believe it's next to impossible to come up with a globally workable scheme and I know ISCED levels are not perfect, however I was aiming for the possible, i.e. un-breaking the little that we have (isced:level=*). How about starting coming up with some ideas for requirements that such a new tagging system should have?

--Skorbut (talk) 19:29, 18 May 2022 (UTC)

@Skorbut: I can understand the desire for global harmonization. After all, many data consumers operate globally and initially balk at the idea of having to hard-code region-specific tags, even predictable, well-documented ones like network=*. Unfortunately, there's simply no global shared vocabulary for school classification that laypeople would understand, unlike with POI classification in general. If we tag enough schools according to each national classification system in school:*=*, then the effort to research and devise a global classification system can happen at the data consumer level, to ensure that school:*=* results in intuitive icons on a map. ISCED's country mappings would be a useful starting point for that effort. – Minh Nguyễn ^💬 00:11, 19 May 2022 (UTC)

is this realy a goegrafic feature

I do not think that this information is geografical data. You may find "hardware" differences in the lower sections - but if you climb upp in the scale it becames more and more confusing.

It is usual today that a univerity is present att many different locations. "Linneuniversity" for exampel has his own facilities in Kalmar and Växjö and what kind of education they offer on witch location is chaning from year to year.

But to make it complete confusing we have thinks like "Campus Västervik", "Campus Hultsfred" that are more or less renting out rooms for a broad variaty of schools, universitys and outer education. One year you can study masterprograms from Lineuniversity their, next year not nessarcarly.

So i think tagging the ISCED level is quite unnessesary, i do not see any possiblity to use the data and it is impossible to maintain.

And this is just the beginning of the development. It becames more and more commen that you can join one university course from different locations or att home - and use a broad varity of services in "e-learning center" and libaries for thinks like exams and supervision.

—Preceding unsigned comment added by Skinfaxi (talk • contribs) 16:21, 2 June 2022‎ (UTC)

What authoritative educational references have we documented so far?

Rather than trying to encourage mappers to guesstimate an ISCED level, it seems much better to me to work on documenting as many authoritative sources for categorizations of schools as we can. E.g. many regions (countries, states, provinces, etc.) license or run various schools, and likely provide reference identifiers for them (and categorizations). Does anyone know if we have a central wiki page documenting these, and which ones we've documented? Once we've done that for quite a few regions, it will (hopefully) be much easier to figure out what global categories might make sense. And if it happens that they map well enough onto ISCED levels, great, we can document that on the wiki, and (automatically?) add relevant tags where the underlying (regional) categorizations systems have already been added. Also, I'm pretty sure a LOT of schools already have Wikidata entries -- what's the progress on making those linkages? JesseFW (talk) 03:04, 10 June 2022 (UTC)