Property talk:P21

From Wikidata
Jump to navigation Jump to search


sex or gender
sex or gender identity of subject: male (Q6581097), female (Q6581072), intersex (Q1097630), transgender female (Q1052281), transgender male (Q2449503). Animal: male animal (Q44148), female animal (Q43445). Groups of same gender use subclass of (P279)
Representssex (Q290), gender (Q48277), sex or gender (Q18382802)
Data typeItem
Domainperson (Q215627), animal (Q729), narrative entity (Q21070598), abstract being (Q15619164), fictional character (Q95074), mythical entity (Q24334685), conjoined twins (Q216866), alter ego (Q201662), fossil (Q40614), fictional humanoid species (Q24286894), organism (Q7239), robot (Q11012), sex doll (Q10832) and synthetic voice (Q79600797)
Allowed valuesmale (Q6581097), female (Q6581072), intersex (Q1097630), transgender person (Q189125), transgender female (Q1052281), transgender male (Q2449503), non-binary (Q48270), fa'afafine (Q1399232), māhū (Q3277905), kathoey (Q746411), fakaleiti (Q350374), hijra (Q660882), male organism (Q44148), female organism (Q43445), unknown, not applicable, androgyny (Q207959), two-spirit (Q301702), transmasculine (Q27679766), transfeminine (Q27679684), muxe (Q3177577), intersex organism (Q28873047), agender (Q505371), genderqueer (Q12964198), genderfluid (Q18116794), neutrois (Q1289754), eunuch (Q179294), pangender (Q7130936), cogenitor (Q64017034), neutral sex (Q52261234), hermaphroditism (Q16674976), cisgender female (Q15145779) and cisgender male (Q15145778)
Usage notesMain: male (Q6581097), female (Q6581072), intersex (Q1097630), transgender female (Q1052281), transgender male (Q2449503). Animals: male animal (Q44148), female animal (Q43445). Groups of same gender use "subclass of" (P279)
ExampleCara Delevingne (Q1395624)female (Q6581072)
Barney (Q2988585)male organism (Q44148)
Tintin (Q52986)male (Q6581097)
Grumpy Cat (Q7714263)female organism (Q43445)
Kim Petras (Q60964)transgender female (Q1052281)
Chaz Bono (Q979387)transgender male (Q2449503)
Format and edit filter validationAbuse filter #13
Robot and gadget jobsDeltaBot does the following jobs:
Tracking: usageCategory:Pages using Wikidata property P21 (Q28039789)
Tracking: local yes, WD noCategory:Articles with template:Bio and P21 value missing (males) (Q23823801), Category:Articles with template:Bio and P21 value missing (females) (Q23823797), Category:Sex or gender not in Wikidata, but available on Wikipedia (Q26220006)
  • Top missing properties by number of sitelinks/P21
  • Check for items with P31:Q5 and P21 as qualifier
  • People with occupation but without gender
  • <search Commons for files with depicts-statement and this property as qualifier>
  • Items with no other statements
  • <most recently created items>
  • Items with novalue claims
  • Items with unknown value claims
  • Usage history
  • Database reports/Recent deaths/missing/P21
  • Database reports/Complex constraint violations/P21
  • Database reports/Humans with missing claims/P21
  • Database reports/Constraint violations/P21
  • <random list>
  • Proposal discussionProposal discussion
    Current uses6,426,304
    Search for values
    [create Create a translatable help page (preferably in English) for this property to be included here]
    Conflicts with “instance of (P31): sibling duo (Q14073567), married couple (Q3046146): this property must not be used with the listed properties and values. (Help)
    Exceptions are possible as rare values may exist.
    List of this constraint violations: Database reports/Constraint violations/P21#Conflicts with P31, SPARQL, SPARQL (new)
    Conflicts with “coordinate location (P625): this property must not be used with the listed properties and values. (Help)
    List of this constraint violations: Database reports/Constraint violations/P21#Conflicts with P625, hourly updated report, SPARQL, SPARQL (new)
    Conflicts with “country (P17): this property must not be used with the listed properties and values. (Help)
    Exceptions are possible as rare values may exist. Known exceptions: hitchBOT (Q17450611)
    List of this constraint violations: Database reports/Constraint violations/P21#Conflicts with P17, search, SPARQL, SPARQL (new)
    Conflicts with “located in the administrative territorial entity (P131): this property must not be used with the listed properties and values. (Help)
    List of this constraint violations: Database reports/Constraint violations/P21#Conflicts with P131, hourly updated report, search, SPARQL, SPARQL (new)
    Conflicts with “is a list of (P360): this property must not be used with the listed properties and values. (Help)
    List of this constraint violations: Database reports/Constraint violations/P21#Conflicts with P360, hourly updated report, search, SPARQL, SPARQL (new)
    Conflicts with “original language of film or TV show (P364): this property must not be used with the listed properties and values. (Help)
    List of this constraint violations: Database reports/Constraint violations/P21#Conflicts with P364, hourly updated report, search, SPARQL, SPARQL (new)
    Conflicts with “author (P50): this property must not be used with the listed properties and values. (Help)
    Exceptions are possible as rare values may exist.
    List of this constraint violations: Database reports/Constraint violations/P21#Conflicts with P50, search, SPARQL, SPARQL (new)
    Conflicts with “subclass of (P279): this property must not be used with the listed properties and values. (Help)
    Exceptions are possible as rare values may exist. Known exceptions: anonymous (Q4233718)
    List of this constraint violations: Database reports/Constraint violations/P21#Conflicts with P279, search, SPARQL, SPARQL (new)
    Conflicts with “has part (P527): this property must not be used with the listed properties and values. (Help)
    Exceptions are possible as rare values may exist. Known exceptions: Yuzu (Q75840084), Platon Kuzmich Koavalyov (Q55039999), Penthesilea (Q55052424), hitchBOT (Q17450611)
    List of this constraint violations: Database reports/Constraint violations/P21#Conflicts with P527, search, SPARQL, SPARQL (new)
    Qualifiers “start time (P580), end time (P582), instance of (P31), sourcing circumstances (P1480), nature of statement (P5102), reason for deprecation (P2241), reason for preferred rank (P7452), applies to part (P518): this property should be used only with the listed qualifiers. (Help)
    Exceptions are possible as rare values may exist.
    List of this constraint violations: Database reports/Constraint violations/P21#Allowed qualifiers, SPARQL, SPARQL (new)
    Single value: this property generally contains a single value. (Help)
    Exceptions are possible as rare values may exist. Known exceptions: Sammy Jo Carrington (Q7409772), Eivor (Q92600458)
    List of this constraint violations: Database reports/Constraint violations/P21#Single value, SPARQL, SPARQL (new)
    Pictogram voting comment.svg Unsourced notname gender
    People with instance of (P31) set to notname (Q1747829) with an unsourced sex or gender (P21) set to male (Q6581097) (Help)
    Violations query: SELECT ?item WHERE { ?item wdt:P31 wd:Q1747829 . ?item p:P21 ?genderstatement . ?genderstatement ps:P21 wd:Q6581097 . MINUS { ?genderstatement prov:wasDerivedFrom ?source }. } LIMIT 1000
    List of this constraint violations: Database reports/Complex constraint violations/P21#Unsourced notname gender
    Value masculine (Q499327) will be automatically replaced to value male (Q6581097).
    Testing: TODO list

    Value boy (Q3010) will be automatically replaced to value male (Q6581097).
    Testing: TODO list

    Value man (Q8441) will be automatically replaced to value male (Q6581097).
    Testing: TODO list

    Value Männchen (Q1957296) will be automatically replaced to value male (Q6581097).
    Testing: TODO list

    Value Male (Q360210) will be automatically replaced to value male (Q6581097).
    Testing: TODO list

    Value husband (Q212878) will be automatically replaced to value male (Q6581097).
    Testing: TODO list

    Value woman (Q467) will be automatically replaced to value female (Q6581072).
    Testing: TODO list

    Value Feminine (Q251810) will be automatically replaced to value female (Q6581072).
    Testing: TODO list

    Value Female (Q639354) will be automatically replaced to value female (Q6581072).
    Testing: TODO list

    Value girl (Q3031) will be automatically replaced to value female (Q6581072).
    Testing: TODO list

    Value Weibchen (Q1304841) will be automatically replaced to value female (Q6581072).
    Testing: TODO list

    Value Female (Q15600976) will be automatically replaced to value female organism (Q43445).
    Testing: TODO list

    Value stallion (Q757833) will be automatically replaced to value male organism (Q44148).
    Testing: TODO list

    This property is being used by:

    Please notify projects that use this property before big changes (renaming, deletion, merge with another property, etc.)


    Rename (en) label 'gender'->'sex'[edit]

    The label for this is wrong - either it's sex:male/female/intersex or it's gender:masculine/feminine/neuter. I'm presuming the former is intended, in which case, should we re-label? James F. (talk) 22:21, 4 February 2013 (UTC)

    Should this be male/female or man/woman or what? I personally don't like male/female. --Stryn (talk) 07:16, 5 February 2013 (UTC)

    Is that solved? Conny (talk) 12:39, 5 February 2013 (UTC).

    No, I think... I haven't seen any discussion elsewhere. :/ --Stryn (talk) 12:40, 5 February 2013 (UTC)
    Well, now I looked the English description: "MUST BE ONE OF: male (Q44148), female (Q43445), or intersex (Q1097630)". --Stryn (talk) 12:41, 5 February 2013 (UTC)
    It does not work in French (and, seemingly, in other languages like Chinese - see the Project chat). In French, using "femelle" is laughable, we use "femme". Through interwikis, "femme" is twinned to "woman" while "femelle" is linked to "female". I am trying to see how to circumvent this, but obviously the requirement to use either Q44148 or Q43445 or Q1097630, probably well adapted to English language, is a non-sense in some other languages. Touriste (talk) 13:12, 5 February 2013 (UTC)
    In german it works after renaming "Männliches Geschlecht" in "männlich" - this seems no problem, so it should be solved here. But maybe the meaning (in look of other languages) has a lag still. Conny (talk) 13:35, 5 February 2013 (UTC). It is very important to discuss this central! Conny (talk) 14:25, 5 February 2013 (UTC).

    What's the reason why we can't use man/woman? --Stryn (talk) 14:22, 5 February 2013 (UTC)

    It seems that we can - I mean there is no software constraint that forbids us to do that, it is possible that we should not - don't know. I tried to modify the item Marie Curie a few minutes ago (I replaced Q43445 by Q467 ) and it worked : [1] OK you don't see anything on this diff ? Neither do I, but the edit history tells us that I removed two bytes. I was reverted eight minutes later. Since then I asked the Village Pump in French and received the confirmation that any item can be inserted as a value of any property. Touriste (talk) 14:55, 5 February 2013 (UTC)

    ✓ Done OK, based on feedback here and on IRC, I've now altered this as proposed. James F. (talk) 19:44, 5 February 2013 (UTC)

    I know I'm a newcomer here, but I have a question. Will it be confusing to use man/woman if the subject is a child? --Auntof6 (talk) 06:27, 2 March 2013 (UTC)
    In Finnish, which Stryn and I speak, I don't actually see it causing that much confusion. As a definition of sex it's not inappropriate. At least I think so, and at least it would be better than the current koiras/naaras. We could also use the derived forms miespuolinen 'of the male sex' and naispuolinen 'of the female sex' if necessary. But yeah, I definitely would have English use male/female, along with other languages using what's appropriate for them. That's why I support dedicated items for use with this property, as discussed in the section below. Enoirditalk 14:59, 2 March 2013 (UTC)

    ਲਿੰਗ is the word for gender in Punjabi (Pa) language. So, you are hereby requested to update the section.--Dr. Manavpreet Kaur (talk) 07:18, 17 December 2015 (UTC)

    Values not making sense[edit]

    Problem with value in czech language: male/female should be translated as samec/samice which is used only for animals. For people should be man/woman (muž/žena).

    Maybe in this case not straight link to article but give options which can be translated different for different languages. JAn Dudík (talk) 13:12, 6 February 2013 (UTC)

    This is the third language (after Chinese and French, see above) in which items Q44148 and Q43445 are obviously unsuitable. I have tried a new approach : I have simply blanked the English description quoted above ("MUST BE ONE OF: male (Q44148), female (Q43445), or intersex (Q1097630)") which does not seem to be consensual. It was strange to do a non-trivial edit without an edit summary handy. Let's see if this is a good idea. Touriste (talk) 13:58, 6 February 2013 (UTC)
    Problem in Finnish language also. I don't think that your edits helps. We need some other solution. It's not good when some people has "man" and some people has "male". --Stryn (talk) 14:14, 6 February 2013 (UTC)
    I've reverted pending discussion; do we need items for "Male human", "Female human" and "Intersex human" so that this works for those languages too? If so, we'd need to label this to make sure this is sex for humans, not in general (which is a pity). James F. (talk) 18:27, 6 February 2013 (UTC)
    What if any item has a property "is a" -> "human person"/"horse" (e.g. Varenne)/"any other living being" AND the "sex" property -> "male"/"female"? So we need to know what is it and then which sex it has, no man/woman/etc.. --Viscontino (talk) 22:12, 6 February 2013 (UTC)
    Yes, that was what we originally intended to do, but apparently some/lots of languages don't have the concept of male-ness translated properly as a thing independent of human-ness, so we can't do that. :-( James F. (talk) 23:22, 6 February 2013 (UTC)
    Since this is going to be an issue very much complicated by language, might I suggest that either a) two custom items be created that return the desired values (titles) that are most appropriate for each language, with no wikilinks, or b) ultimately the data type for sex/gender not be "Item". I am not aware if properties will ever have constraints, but obviously this one needs a hard constraint in software or otherwise. Espeso (talk) 03:09, 7 February 2013 (UTC)
    @Stryn: I'm not a polyglot but why isn't right Koiras/Naaras in Finnish language? is it used only for animals? --Viscontino (talk) 14:53, 7 February 2013 (UTC)
    Mostly for animals, yes. We never use words koiras/naaras when speaking about people. --Stryn (talk) 15:38, 7 February 2013 (UTC)
    I think the problem here is that the labels for Q44148 (etc.) have been mis-translated. The concept is very clearly about both male humans and male non-humans, but the fi label is "Koiras" (and Uros?). Could you instead translate the labels as they are used - "Mies or Koiras"? i.e.:
    • male (Q44148) -> "Mies or Koiras"
    • man (Q8441) -> "Mies"
    • male beast (????) -> "Koiras" - we might need to create an Item for this concept!
    Of course it would be even better if is there a term about the concept of maleness in Finnish that's shared between both humans and non-humans? James F. (talk) 17:06, 7 February 2013 (UTC)
    Eh, I understand your confusion. It's just that while koiras/naaras are the 'shared' concepts of maleness/femaleness, they're skewed towards animal-related/biological use. It's inappropriate – impolite – to refer to humans as koiras/naaras. It's kind of like calling someone an animal: yes, it's technically correct, but not appropriate.
    Also, remember this is also a problem in French, Chinese, and Czech. As Espeso said, there should probably be special items for use only with this property. Enoirditalk 17:29, 7 February 2013 (UTC)
    Agree with James F. We should separate the man/woman property from the male/female (animals), or is there again some problems with other languages? --Viscontino (talk) 22:53, 7 February 2013 (UTC)
    How? Make a property for both of them? Where we then use them? --Stryn (talk) 07:18, 14 February 2013 (UTC)
    Clearly, we would us it here. This is the same as English being deficient in having two different words for 'free' (roughly translated as gratis and libre); for a Property about a piece of software's status, we would have to use (or create) a term for the two as a concept as well as for the individual concepts. James F. (talk) 19:24, 14 February 2013 (UTC)

    Estas malbone. Pole "płeć" devas 4 eblecoj:

    • "mężczyzna" - por homo
    • "kobieta" - por homo
    • "samiec" - por animalo
    • "samica" - por animalo

    Esperante ankaŭ:

    • viro - por homo Q8441
    • virino - por homo Q467
    • Vira sekso - por animalo Q44148
    • Ina sekso - por animalo Q43445

    Mi proponas Sekso (P21) skribi Q8441 kaj Q467. Marek Mazurkiewicz (talk) 01:34, 21 February 2013 (UTC)

    There is problem with this also in Swedish. hanne/hona can be used of Homo Sapiens, but not for specific human. It is ok if youre an Alien doing experiments on a "male of female human". It is also ok if youre God, creating human as male and female, but not after you have named them as Adam and Eve. -- Lavallen (block) 16:44, 9 March 2013 (UTC)


    See [2]. Somebody added sex=eunuch to castrates, but it was reverted. Technically they were males, but I think castartion is signifcant for them, see existing categories (q5110164 and q5110160) and there is no other good property for this. Or leave it blank? JAn Dudík (talk) 10:59, 22 February 2013 (UTC)

    Castration/eunuch would probably be better as a qualifier to sex(male). /Ch1902 (talk) 13:07, 22 February 2013 (UTC)

    Deutsch Intersexualität[edit]

    Müsste es nicht statt Intersexualität Intersexuel heißen? Mauerquadrant (talk) 05:04, 5 March 2013 (UTC)

    Das Problem ist das die Bezeichnung aus dem Label gewonnen wird. Das Item welches verlinkt ist sammelt jedoch die Artikel zu Intersexualität (dies ist auch das deutsche Lemma). --Sk!d (talk) 02:32, 20 March 2013 (UTC)


    Does this property only refer to "persons" (homo sapiens) or can it also apply to other creatures (Chi-Chi the panda; Jumbo the elephant). If so what about the weird hermaphroditic animals that exist? How would we cope with a specific and named amoeba? -- SGBailey (talk) 11:42, 11 March 2013 (UTC)

    It refers to all species that have a sex, though there are some difficulties with the wording, see below. --Zolo (talk) 09:16, 12 March 2013 (UTC)

    French translation (note to French-speaking users)[edit]

    J'ai changé le libellé français genre -> sexe. La version anglaise emploie "sex" alors même que "gender" est plus répandu que le français "genre" et nettement moins polysémique --Zolo (talk) 08:32, 12 March 2013 (UTC)

    la version anglaise est passée à « sex or gender », du coup j'ai également changé le libellé fr: vers « sexe ou genre ». -Ash Crow (talk) 19:16, 8 December 2014 (UTC)

    change recommended values[edit]

    Unless there is opposition to it, I will change this evening the set of recommended value for humans from Q44148 and Q43445 to Q6581072 and Q6581097. It has been repeteadly remarked by users that Q44148 and Q43445 sound weird in some languages and no one has opposed this proposal at Wikidata:Project chat#Statement p21 (sex) should not use q43445 (female) as value. --Zolo (talk) 09:17, 12 March 2013 (UTC)

    I changed the descrption (forgot to marks as done)--Zolo (talk) 13:49, 19 March 2013 (UTC)
    I could do this by my bot. --Sk!d (talk) 13:45, 19 March 2013 (UTC)
    Thanks, that would be nice. I think user:Legobot has done some too. --Zolo (talk) 13:49, 19 March 2013 (UTC)
    Just some statistics: I looked up the first 10500 items with this property. Theses are the sum of the linked properties: Q251810: 2, Q871: 1, Q6581097: 74, Q1621322: 1, Q6581072: 8, Q467: 2, Q44148: 8599, Q43445: 1833, Q8441: 5. --Sk!d (talk) 15:31, 19 March 2013 (UTC)
    So no one is confused, I have/will fix a few of these invalid entries. Espeso (talk) 15:49, 19 March 2013 (UTC)
    Some more Statistics: There are over 559.000 itmes using this property the first 50.000 have those values: {6581097: 144, 6581072: 12, 44148: 42980, 43445: 7426, 5442715: 1, 499327: 1} --Sk!d (talk) 18:27, 20 March 2013 (UTC)

    Legobot and my bot have changed all items. I think I will run this task in the future on a regularly basis to fix false entries. --Sk!d (talk) 12:21, 31 March 2013 (UTC)

    I think that it's not right to have recommended values, because we should try to describe the world, rather than to prescribe it. Recommending is mild form of prescription. I think a better wording might be common values, and make a point say that any Value that describes a Person's identity is appropriate. I think we're restricting ourselves to a small number of acceptable values is for the ease of performing queries. If that were true, we'd be sacrificing rich description to circumvent a technical problem; whereas I'd like to see accurate description and a attack the technical challenge that accompanies it. Maximilianklein (talk) 21:54, 6 May 2013 (UTC)


    I created the Item Q7745125 for unknown sex. The German wikipedia has a whole category with people where the sex is unknown see de:Kategorie:Geschlecht unbekannt. --Sk!d (talk) 02:28, 20 March 2013 (UTC)

    There is a "special value" unknown that works for all datatypes, I think it is better to use it. --Zolo (talk) 10:44, 23 March 2013 (UTC)

    copied from Talk:Q7745125

    Hm, the feature of a special item for this is that it would be much more easier to get every item linking to this item instead of looking up all items with the Property and then checking if the value is unknown. But I don't know if this would be possible with the query syntax in the future. --Sk!d (talk) 20:03, 23 March 2013 (UTC)
    I hope that phase 3 will give more possibilities for this kind of things, that seems like a basic feature if we want to be able to generate list. Actually, wdstats already does something rather similar, but it does not yet work for special values. Can I delete this item for now ? I know that P21 other special items were created for P21, but this one does not seem necessary from a linguistic point of view and too many porpoerty specific values may be rather hard to maintain. --Zolo (talk) 06:00, 25 March 2013 (UTC)
    Another problem with "unknown" is that it looks ugly and rather confusing, though that would rather be an issue for the development team. --Zolo (talk) 08:04, 27 March 2013 (UTC)

    only Persons[edit]

    We should make it clear that this Property should be only used for Persons (see Domain on the top of the page). This is why I removed the part in the descriptions which says: "for other species: male (Q44148) or female (Q43445)". There should be another Property if this is needed. --Sk!d (talk) 02:37, 20 March 2013 (UTC)

    Is there any reason to any reason why it should not be applied to other species. My understanding had always been that it could, just with a different value. The discussion at the top mentionned en:Category:Men and en:Category:Women because we have vastly more humans than ostriches with their own item, but I think it makes sense to use it for all species, as long as it refers to a biological property, and not say, to the grammatical gender of a word. --Zolo (talk) 08:02, 20 March 2013 (UTC)
    I think there will be easy to make mistakes for those who, in there own language, do not separate female humans from females of other species. People can be offended by being classified as an animal, and since we have special precautions for biographies... -- Lavallen (block) 08:49, 20 March 2013 (UTC)
    Hm if the GND value Person is correctly set it might be possible to control if the correct Item for Humans is set. The other way round is a little bit harder as there is no clear why of identifying if a item is not-human or if the GND typ is just missing. An wikidata-index with the possible value "animal" would be a great help for this. --Sk!d (talk) 10:19, 25 March 2013 (UTC)
    If the item has any of these properties, it is not human, unless those properties have some very specific values. - Soulkeeper (talk) 23:37, 24 April 2013 (UTC)
    I agree it doesn't make sense for us to be classifying humans by sex. We need to change it back to gender (and adjust the value suggestions accordingly). Kaldari (talk) 17:18, 24 May 2013 (UTC)

    in Fiction and Myths[edit]

    How to do with people i fiction, like James T. Kirk as an example? He is not categorised as a 'man' on svwp, since he is a tv-/film-character. (It looks to me like en:Category:Men and sv:Kategori:Män should have separate items, but that is another question.) On svwp the union of the sets of of "Män" and "Kvinnor" should contain all biographies. Fictional and Mythical persons are not included. (Some semi-mythical persons are of course always difficult to classify.) -- Lavallen (block) 09:14, 20 March 2013 (UTC)

    I think we should treat them the same way as real people. Otherwise, we would also need a special treatment for fictional birth places, fathers, that would be kind of endless. We certainly need a "is fictional" property once the boolean datatype is available. --Zolo (talk) 09:28, 20 March 2013 (UTC)
    I am the same opinion. Another Property which says they are fictional should be enough. --Sk!d (talk) 10:25, 20 March 2013 (UTC)
    Sounds good enough, even if I doubt it is always easy to treat boolean properties. (True/False/Doubts/Plausible). -- Lavallen (block) 11:06, 20 March 2013 (UTC)
    I think "is a" would be the right property just use always "is a fictional charakter" or something like this. --Sk!d (talk) 11:34, 20 March 2013 (UTC)
    Not sure of that. There are also fictional objects, animals, places and probably other things. We may not have items for all of that, and I am not sure it would be very useful to create them.
    In some cases, we will have to say that some sources state that someone existed, and some state that he did not exist, and yet another that does not know. As we can give several values to the same property, and there is a special value for unknown, we can use a "is fictional property" to say "source A says Jesus existed, B says no, C does not know, and hopefully also that one of the 3 values is ranked as "preferred" because it reflects the current scholarly consensus. That sounds more difficult to get with the "is a" property (is a fictional character + is a real person + is a possibly-non-existing person ?). --Zolo (talk) 12:04, 20 March 2013 (UTC)
    On svwp we have categories for fictional persons birthplaces. Kirk above is a "Fictional character from Iowa". But 'from' and 'birthplace' is not always the same thing, I'm afraid. We have endless talks about this on svwp. Is 'from' the location of your birth (for Kirk in outer space) or where your mother lived when you were born or the place you are mainly related to. Swedish statistics changed the "birthplace" from "location of birth" to the "location of the mothers home" some years ago. -- Lavallen (block) 12:40, 20 March 2013 (UTC)

    Also in the Swedish Wikipedia, the categories are named men/women (män/kvinnor) and no animals, fictional, or Biblical characters are allowed there. These two categories are only used for biographies. That's what makes them useful. It would be a pity of Wikidata couldn't be as useful. Please describe how I can extract a list of all men and women from Wikidata. --LA2 (talk) 16:35, 9 May 2013 (UTC)

    "Biblical characters" are de facto allowed as long as they are not considered to mythical, several Israelitian kings are historically confirmed. Mose is not in Män, but Jesus is. The existens of a Jewish Rabbi by the name Jesus is possible, and I think therefor he is among the Män. King David is, I guess, at the frontier where he can be considered to mythical. At present he is in that category. -- Lavallen (block) 16:46, 9 May 2013 (UTC)

    Commons categories[edit]

    Should I link commons:Category:Male humans and commons:Category:Female humans to Q6581097 and Q6581072 ? Or should those items be considered "special" and be left statement-free ? In many languages those items use adjectives as labels, so that it may not sound very logical to add this kind of statements. --Zolo (talk) 09:37, 6 April 2013 (UTC)

    I would consider them as special items. It is not clear that the item is about humans. I think there are better items to link from. --Sk!d (talk) 21:43, 10 April 2013 (UTC)

    ISO/IEC standard for human sexes[edit]

    See also w:ISO/IEC 5218. - Soulkeeper (talk) 23:48, 24 April 2013 (UTC)

    Argument for Multi-Valued Items[edit]

    Hello, I've been looking into how historically sex has been stored by catalogs in more sophisticated ways. One method, I really like is how the Library of Congress approaches recording sex. Basically each item can have 1 or more sex values. If there is more than 1 sex value, then those values are associated with start and end dates.

    I think the best way to use this approach in Wikidata is to allow items to have multiple claims for Property Sex, and if there are more than one claims for an item, then qualifiers should be used to describe the dates for which those properties are accurate. Maximilianklein (talk) 21:48, 6 May 2013 (UTC)

    This seems like a good idea; temporal coverage are what qualifiers were initially discussed as being meant for, anyway. :-) James F. (talk) 20:54, 9 May 2013 (UTC)
    Note that the w:Text Encoding Initiative have recently moved away from w:ISO/IEC 5218 (see ticket) and the w:W3C are in favour of more open classification. I think we need to be more open here. Stuartyeates (talk) 03:46, 16 May 2013 (UTC)
    Sure, but gender and sex are entirely seperable, and that is about trying to combine the two into one field rather than just have a distinct one. Merging the two concepts would be a poor start. :-( James F. (talk) 15:53, 17 May 2013 (UTC)
    Agree that separating 'sex' and 'gender' would probably make sense for English at least. I wonder, however, if there are languages that do not differentiate between sex and gender and how that could be handled. Kaldari (talk) 23:08, 21 May 2013 (UTC)

    unknown gender[edit]

    As I said before, we need a solution for people who´s gender is unknown. e.g. Jack the Ripper or authors under pseudonym have unknown gender.--Giftzwerg 88 (talk) 10:51, 19 May 2013 (UTC)

    The built-in solution is to use the special "unknown" value, and this is what is currently done in Jack the Ripper (Q43963). I agree that it does not look nice, but that is a developer's problem. I do not see why we need anything else than what has been especially designed for this use. --Zolo (talk) 11:04, 19 May 2013 (UTC)
    There seems to be some confusion about the special "unknown" value at Jack the Ripper (Q43963). Despite the discussion here, it showed up on User:SamoaBot/sex doubts and Delusion23 removed it.
    A solution could be to create an explicit value instead, e.g. Q7745125.
    This would also make it easier to do cross-checks on that (e.g. Special:WhatLinksHere/Q7745125). --  Docu  at 06:27, 31 August 2013 (UTC) (edited, adding existing item)
    was there a decision made on this? - many historic art catalogues only provide initials which makes it difficult to assess gender. This may be resolved with later research but it does complicate creating items as there are mother fields ie. date of birth which are linked to this field Eothan (talk) 02:22, 8 January 2020 (UTC)

    Gender data from be/ja/ru/zh Wikipedia[edit]

    Are there categories at these Wikipedias we could use with "array properties gadget" to add this P21 to items? Frequently items with articles just in these Wikipedias don't have this property. --  Docu  at 20:18, 19 May 2013 (UTC)

    What do you enter when sex and gender do not agree?[edit]

    If someone's gender does not match their sex, what do you enter? ('Gender' currently redirects to 'sex' as a property.) If the answer is "There isn't a solution for that yet", that's fine :) I realize it's a complicated problem. Kaldari (talk) 23:02, 21 May 2013 (UTC)

    When do we have that problem? I am maybe lost in translation now, but as far as I can see, that only affects animals?! -- Lavallen (block) 10:39, 22 May 2013 (UTC)
    In English, it only affects humans, not animals. Humans have sex and gender, while animals only have sex. Gender is a masculine or feminine gender identity, while sex is a biological trait - what kind of chromosomes you have (or genitalia). Hope that makes sense. 16:41, 24 May 2013 (UTC)
    Maybe we could add both with different references. Like transsexual person we could add a time qualifier to state that the gender (e.g. male) has started on a given time wile the sex was always the same (e.g. female). --Sk!d (talk) 02:19, 16 June 2013 (UTC)
    I agree Jane023 (talk) 15:26, 26 October 2013 (UTC)

    This seems particularly topical given the edit war over Bradley Manning. Sj (talk) 23:33, 22 August 2013 (UTC)

    I agree and would add we need more categories than just three: M, F, Trans->M, Trans-F, Cross->M, Cross-F, Herm, and unknown. For the cross-dressers, maybe we should just leave them ambiguous, along with the hermaphrodites. I do think we need to establish a time frame for transexuals, especially if they have children before the switch, otherwise the data will seem full of errors when you start linking up families. Jane023 (talk) 15:26, 26 October 2013 (UTC)
    People might read the relevant wiki articles before opining. Hermaphrodites are generally called Intersex. And fully sexual/gender identity males or females can nevertheless have the chromosomes typical of the other sex. 21:37, 27 October 2013 (UTC)

    Rename (en) label 'sex'->'gender'[edit]

    In modern English people are categorized by gender, not sex. Animals, however, are always categorized by sex. Having 'sex' be a property only for humans is backwards. This appears to also be an issue with Czech, Finnish, and possibly French and Chinese. If we are going to keep this as the property for humans, it should be 'gender' in English, and the values should be changed to 'male', 'female', or 'other'. We should also create a 'sex' property for animals, with values of 'male', 'female', and 'intersex' (although I'm not sure how needed such a property would actually be). Kaldari (talk) 17:14, 24 May 2013 (UTC)

    Pictogram voting comment.svg Comment. If by "gender" you mean the sex someone identifies herself with, that seems rather problematic: really how do we know ? Until recently it was not really common to publicly say "deep in my vagina, I feel I am a boy". If you just mean that gender should be used for biological sex, I guess that makes sense to follow usage, but I doubt it would apply to many other languages than English. --Zolo (talk) 18:23, 24 May 2013 (UTC)
    By 'gender', I mean the gender someone is publicly known as. I don't agree that it's easier to know someone's sex than their gender. If someone is a notable transgender person, it's generally known which gender they identify with. It is not always known, however, if they have actually had sex-reassignment surgery. And it is very rarely known if someone is actually intersex since intersex people normally pass as male or female. Either way, it is rather awkward and clinical to classify people by 'sex' instead of 'gender'. Kaldari (talk) 20:44, 24 May 2013 (UTC)
    I guess it depends on places and cultures, but in most places and time, you were assigned rights and duties based on your real sex, not on perceptions. Of course, you were not supposed to cross-dress or "cross-act", which makes the distinction a bit moot, but still it would be a bit odd to consider Pope Joan a male. Even nowadays, you cant take part in a women's sport competition based on your gender identity. Also, when someone gets sex-reassignment-surgery, that sounds sensible to change P21. --Zolo (talk) 05:55, 25 May 2013 (UTC)
    Yes, professional sports competitions are the one and only field where sex is actually determined and important, rather than gender. In virtually all other fields, people are classified by gender. No one has to get a blood test to be classified as a "female novelist". Kaldari (talk) 17:24, 25 October 2013 (UTC)
    What do you mean ? When are you supposed to be be classified as a female novelist at all ? --Zolo (talk) 23:02, 13 November 2013 (UTC)

    -- 19:44, 27 May 2013 (UTC) Agree - Recommended reading Representing the Reality Underlying Demographic Data

    A distinction can be made for "gender identity", which is internal, and "gender presentation", which is public. Someone can present one way and identify another. Gender identity is not something that we usually know about historical figures. Gender presentation we often would. Blue Rasberry (talk) 18:36, 25 October 2013 (UTC)
    Pictogram voting comment.svg Comment. There is currently a proposal to create a 'gender identity' property here. Gobōnobō + c 17:55, 26 October 2013 (UTC)
    Symbol support vote.svg Support Strong support, do we really know the chromosomes of those historical figures? Maximilianklein (talk) 17:42, 25 October 2013 (UTC)
    Symbol strong support vote.svg Strong support per Maximilianklein. + We have no way to source chromosomes, hormonal levels, various biological phenomena (en:Androgen insensitivity syndrome) or the presence and functionality of primary sex organs. We can however source how a person presents itself to the world. --Tobias1984 (talk) 16:36, 13 November 2013 (UTC)
    At least we do for popes :) That is only half a joke. Yes, we know the sex of historical figure. Basically, you were not allowed to be King of France, or to vote in the 19th century or to marry a female if you were not a biological male. Your gender has nothing to do with it.
    Symbol oppose vote oversat.svg strongly oppose making "gender" the default. "Gender" is a mixture of subjective feeling, externally imposed categorization, and biology. Even the meaning of the word is not entirely clear. Wikipedia defines it as a "person's private sense, and subjective experience, of their own gender" (and if so how do we know ?), but that does not appear really compatible with concepts like "hijra" that appear to socially imposed categories. Actually, according to Wikipedia hijra is a "challenge Western ideas of sex and gender". So it seems we have various closely related concepts with no clear boundaries, and given that the propertion of complex cases is small I think the best solution would be a single property, and use qualifiers to state that the person "is biologically male", "has been socially categorized as male", "claims to feel like a male" etc. Alternatively, I am not strongly opposed to having two properties, but just dont make "gender" the default. --Zolo (talk) 23:02, 13 November 2013 (UTC)
    I think your making a good point, and I am also against splitting the property. People need to understand that our male and female category is a mix of biology and culture. Neither sex nor gender can be adequately sourced or even inferred. By "having" (who knows where they are from?) children some people fool us into being X- or Y-chromosome donators and others fool us into that they have a working uterus (Did they also provide the other X-chromosome?). --Tobias1984 (talk) 15:22, 14 November 2013 (UTC)
    Symbol oppose vote.svg Oppose This proposal is similar to a a more recent proposal, and shares many of its problems:
    Given that, I think it would be best to keep this property labeled 'sex', and its values one of the biological sexes of 'male', 'female', or 'intersex' (or 'unknown'). Emw (talk) 13:14, 29 November 2013 (UTC)

    Separate fields for 'sex' and 'gender'[edit]

    These are different properties in many cultures, and should both exist; though in almost all cases the two will agree. The case of trans athletes is a good one - someone who may be sorted according to their gender in most walks of life may be sorted according to biological sex, and sorted that way, in sports databases and rankings. For more about distinguishing these two and multivaluing, see the Genderbread discussions. Sj (talk) 23:35, 22 August 2013 (UTC)

    Let's not do this until we figure out how it would work in Far Eastern languages. Two properties with the same name would be confusing. Shii (talk) 15:46, 2 September 2013 (UTC)

    This will disrupt i18n with Japanese where "sex" and "gender" are the same word (性). Shii (talk) 22:03, 22 August 2013 (UTC)

    Symbol support vote.svg Support We would be very modern to mark the two if they are separate. If we can get a japanese user to provide a translated distinction then we can move ahead it seems. Maximilianklein (talk) 17:42, 25 October 2013 (UTC)
    Symbol support vote.svg Support If sex is worth mentioning then gender is worth mentioning as well. I do not speak Japanese but I think the Japanese must have some way of describing the odd things they put in their cartoons, so I am not immediately prepared to accept disruption of Japanese projects as a counterargument. Blue Rasberry (talk) 18:32, 25 October 2013 (UTC)
    Pictogram voting comment.svg Comment Japanese native speaker here. When Japanese speakers import "new" ideas/distinctions for which the language has not yet found the right word, the Western pronunciation is imported and written down in phonetic syllabary. The field in question would be ジェンダー (jendā), and this is the socially accepted terminology [3]. So, I don't know about the other languages, but this is not a problem for Japanese. --朝彦/Asahiko (talk) 17:13, 26 October 2013 (UTC)
    Please participate in the discussion below as well Jared Zimmerman (talk) 08:19, 14 November 2013 (UTC)
    Symbol support vote.svg Support Separating concerns of sex and gender into separate properties would be a good data modeling choice. This is explored more here and here. Emw (talk) 13:14, 29 November 2013 (UTC)
    Symbol support vote.svg Support Sex and gender are two different aspects (please see Sex and gender distinction. Fred (talk)

    There seems to be consensus to split "sex or gender" into "sex" and "gender" as separate properties. This should address many of the problems brought up elsewhere on this talk page such as what to do when sex and gender don't match, i.e. Q297865, etc. The main concern raised about this change is that it would be confusing for Far Eastern languages, for example, Japanese. However, a native Japanese speaker said that it would not be a problem since there are separate words in Japanese for the two concepts: 性 and ジェンダー. It also seems that Chinese has separate words for the two concepts: 性别 and 社會性別. Even if a language doesn't have separate words, it shouldn't matter since they can be disambiguated by the description. Since this property is most commonly used for identifying gender, it would be simplest to change this property to just "gender" and create a new property for "sex". I'll follow up with such a request at Wikidata:Property proposal. Kaldari (talk) 03:46, 1 August 2019 (UTC)}}

    It is misleading to present this as a formally closed discussion after six years. It looks like the purpose of this action was to give this short discussion section more status than the much longer discussion below which resulted in changing the original 'sex' definition of this property to 'sex or gender'. Therefore I undid this closure. Bever (talk) 02:38, 16 April 2020 (UTC)

    The above discussion is preserved as an archive. Please do not modify it. Subsequent comments should be made in a new section.

    VIAFbot imported 44,000 Sex statements from, conflict report posted.[edit]

    I just finished running VIAFbot to import new sex data where possible. I wrote a blog about the details if you're interested. A list of statistcs, conflicts and edge cases are available at User:VIAFbot/Property_Sex_Report. Maximilianklein (talk) 16:58, 20 June 2013 (UTC)

    • Thank, good job. Interesting case is VIAF errors or something very similar to errors. For example Q9038. What we can do with it? — Ivan A. Krestinin (talk) 17:49, 20 June 2013 (UTC)

    Description for other language still retained other property[edit]

    Even if the constraints for the 3 values was already agreed upon.

    male (Q6581097), female (Q6581072), intersex (Q1097630)

    I notice that the description for other language is still saying more than 3 values. I tried to change the language that I know but how about for the others.

    --Napoleon.tan (talk) 11:51, 5 July 2013 (UTC)

    Hm, the discussion under Property talk:P21#only Persons had no clear conclusion. Time to reopen it? Bever (talk) 02:33, 27 July 2013 (UTC)
    Well - in Polish description there is info that if you wrote about animal you should use different properiety (what I assume is very ok) - so maybe first we need to have examples or somebody who understand other languages, to make a choice? PMG (talk) 18:25, 27 July 2013 (UTC)
    Parallel discussions on Wikidata:Project chat#How do we identify animal genders? suggest that this is the right property for animal gender. I would suggest that we restart allowing male organism (Q44148) and female organism (Q43445) for animal. Objections? -- LaddΩ chat ;) 00:20, 28 July 2013 (UTC)
    @PMG: Polish and other languages advise to use a different item as a value for this property. That is not the same as using another property. Bever (talk) 23:21, 12 August 2013 (UTC)
    Yes, there was no consensus on the naming, and frankly the current English naming makes no sense. Unless you examine someone gonads, you have no way of knowing if someone is male, female, or intersex. What we actually classify people by is gender, and this is also the most commonly used term in English language sources. Kaldari (talk) 17:16, 25 October 2013 (UTC)


    Please see Wikidata:Bot_requests#Simultaneous_edits_.28Property:P21_duplicates.29. --  Docu  at 10:50, 4 May 2013 (UTC)

    I think the constraint can be change from only using gnd type person to also allowing creative works for fictional character? --Napoleon.tan (talk) 11:30, 5 July 2013 (UTC)

    Wrong sex from VIAF[edit]

    A blog post explaining the problem of some VIAF entries containing the wrong sex: "Sex Ratios in Wikidata, Wikipedias, and VIAF part 2". --AVRS (talk) 13:34, 22 July 2013 (UTC)

    Non-binary animals?[edit]

    What about animals like the common earthworm, which are hermaphroditic; or effectively single-sex animals such as the all female whiptail lizard? Not to mention how many reptiles can change sex depending on their enviroment and social conditions. Shouldn't there be a catch-all for animals too?

    The traditional term 'hermaphrodite' is nowadays considered to be stigmatizing, misleading and technically incorrect when applied to intersexual people. I do not know if the opposite would be feasible: using intersexual for hermaphroditic animals. It seems that the word intersex(ual) can be used for non-humans but primarily for individual animals in species which usually do have a distinction between male and female.
    So in theory you may be right. But then, to which items would we apply this catch-all? I think this item was set up to be used for individuals, like Dolly, Herman, Laika, Lassie or Pal (the dog who played Lassie, interestingly Lassie was said to be female but Pal was male). I cannot imagine many earthworms get their own Wikidata item... Bever (talk) 21:36, 20 August 2013 (UTC)
    I would wait until such an items is created. Maybe a catch all would be to use "some value" or "no value". --Sk!d (talk) 23:31, 20 August 2013 (UTC)
    The term "hermaphrodite" to me preserves the dignity of animals. I cannot speak to whether it is still the preferred scientific term. Let me ask some others... Blue Rasberry (talk) 18:27, 25 October 2013 (UTC)
    In my opinion you can certainly use hermaphrodite with P21 for earthworms, especially if you include a source. Filceolaire (talk) 00:56, 13 November 2013 (UTC)
    The term "hermaphrodite" is problematic. For animals it typically means having gonads of both sexes (as a normal state of the animal), but for humans, it means being gynandromorphic (having some characteristics of both sexes, but not fully both). Confusingly, animals can also be gynandromorphic, but are not usually called "hermaphrodite" in such cases. Kaldari (talk) 21:15, 21 May 2014 (UTC)

    Wikidata:Requests for deletions#Delete unnecessary gender Items[edit]

    I nominated some values for deletion as I don't think we should split up the gender values into subgroups. Please discuss the deletion there. And add your opinion there. --Sk!d (talk) 00:51, 27 August 2013 (UTC)

    Would you ask them to be undeleted so people can discuss? --  Docu  at 02:38, 27 August 2013 (UTC)
    • The items deleted were basically "male (fictional person)", "female (fictional person)" and two others. (Please correct me if I'm assuming wrongly).
    • My personal thoughts are basically Sk!d's. I think p21 should simply be "male" or "female". Any characteristics about whether the person is a real person or not, can be done with "instance of". Legoktm (talk) 21:21, 28 August 2013 (UTC)

    Perspective from India[edit]

    There is very little scholarship existing in English, and I suspect almost none in native languages as well, on the Indian concept of a hijra. Whatever else this community may be, they often are called the third sex or gender. In this system, sex, gender, and sexuality are conflated. There are a lot of other tough classification issues. There is a history of activism of this group wanting non-male non-female legal recognition.

    The situation in India is particularly relevant because it models the perspective of a large percentage of the world's population. The views on this demographic in India often apply to beliefs in Nepal, Pakistan, and Bangladesh, so it would not be unreasonable to add the populations of those countries to the pool of people who need to be represented in this discussion. Finally, aside from India, other parts of the world have their own perspectives of this issue and their thoughts do not mesh well with Western thoughts.

    I have no idea how to find an Indian authority on this issue. I have looked for some time and was unable to find any authority on the subject. I would encourage non-committal advancement in developing this Wikidata property by making it somehow open to change in the future when other perspectives are found to integrate here. Blue Rasberry (talk) 18:46, 25 October 2013 (UTC)

    If you have a source showing that the person has identified as a hijra then you can certainly use that with this property, provided you include the source details. Filceolaire (talk) 00:54, 13 November 2013 (UTC)

    Transgender / Cisgender changes[edit]

    Jaredzimmerman has recently added (without consent) the following items to the description of this property: transgender male (Q15145783), transgender female (Q15145782), cisgender female (Q15145779), cisgender male (Q15145778) (Diff: The community should still discuss if these changes should be reverted or kept. --Tobias1984 (talk) 21:53, 11 November 2013 (UTC)

    Which includes this edit: --Tobias1984 (talk) 21:58, 11 November 2013 (UTC)
    This edit: was also not discussed. Should we really remove "gender" as an alias from this property? --Tobias1984 (talk) 10:48, 12 November 2013 (UTC)
    This property isn't about gender or gender identity, it is about the biological sex. No matter what someone views themselves as, they have one type of sex organ or another (or none), and that is what this property should reflect. This avoids any subjective view of self and sticks to the objective facts. Perhaps there could be a gender property as well, but it should be separate from this one. Ajraddatz (Talk) 19:50, 12 November 2013 (UTC)
    Seems rather confusing that in one sentence we're equating sex with gender when discussing this property and in the next we're not. according to and and pretty much anyone who doesn't fit into binary genders, these things are quite different, thats why I reccomend we remove the aliase and make two properties if neccesary.
    Ajraddatz: We know very little about the genitalia and the chromosomes of persons on wikidata. In most cases all we know is how people choose to present themselves and, in my opinion, this property in most cases represents that - not biological sex. Filceolaire (talk) 00:48, 13 November 2013 (UTC)
    We know quite a lot about those things by deduction. If a person has a biological child and a spouse and presents themselves as male, then it seems safe to classify them as biologically male. In lieu of that, there are secondary sex characteristics that can be used to infer biological sex.
    The comment above seems to be arguing more that we should relabel this property 'gender', rather than maintain the current conflation between 'sex' and 'gender'. Do we know roughly what percentage of transgender people do not disclose their transgender status? If that number is an exceedingly small proportion of all people, then such a relabeling might not be worthwhile. Emw (talk) 05:04, 13 November 2013 (UTC)
    If a person presents male then we have no idea if they fathered the child unless a DNA test was done. We do not know what percentage of transgender people conceal their status. Their isn't even a single agreed definition of what transgender is. Hence my proposal we should continue to conflate sex/gender under P21 and use qualifiers to explain the special circumstances of those few who fall outside the convention categories. Filceolaire (talk) 08:09, 13 November 2013 (UTC)
    Requiring such a stringent and pedantic standard of documentation to assign a subject's sex seems odd to me. Consider if we were to apply such a standard to assign gender. Let's say we establish that a subject's gender is whatever they self-identify as. Requiring a karyotype or sex organ verification to establish sex would be like requiring a statement directly from the subject that they self-identify as a male, woman, or some other gender (such statements would be awkwardly explicit, and unlikely to come by). Both tests are unreasonable.
    Inferring that a subject is biologically male if they are said to express a certain set of sociologically or biologically male properties (and do not self-identify as non-male) seems reasonable. In other words, if a subject is referred to with a gender-specific pronoun and is not known to be transgender or intersex, then it seems reasonable to infer that they have one of two kinds sex organ, express secondary sex characteristics, are more or less likely to be express color-blindness, hemophilia or other sex-linked conditions. Emw (talk) 03:52, 14 November 2013 (UTC)
    The sex and gender of most people are not controversial, but there are still many people for whom this can be very controversial. At least in those cases, being pedantic is exactly what we should be doing. Maing inferences carte blanche is not appropriate. --Avenue (talk) 00:27, 15 November 2013 (UTC)
    I agree that the sex and gender of most people is not controversial, but for some people it is (and if we have reason to believe someone falls into the latter group, then introducing pedantic criteria makes sense). My point was that the tests that have been suggested here for establishing a subject's sex are unduly stringent, and a similar level of stringency is not being applied for establishing a subject's gender.
    Inferring that a subject has certain base properties if it has certain strongly-correlated derived properties is not carte blanche inference. It is the basis of statistical classification. I think the benefit of enabling downstream inferences (e.g. Alice has sex female, so if she is a parent of Carol then she must be Carol's mother) is greater than the cost of misclassifications (mislabeling undisclosed transgender people, a population which is likely very small). I see no compelling arguments why a very small false positive rate should cause us to abandon this kind of classification. Emw (talk) 13:22, 15 November 2013 (UTC)
    One option that I can think of is that we have male/female/intersex be valid properties of sex and transgender male/transgender female/cisgender female/cisgender male/genderqueer (new) be the valid inputs of a new property "gender" for the issue above about some languages using the same word for sex and gender, we can have gender be "gender identity" and have the descriptions clarify the difference.

    Jared Zimmerman (talk) 21:36, 12 November 2013 (UTC)
    I agree sex and gender are quite different things, and we should have two separate properties for them. I'd expect the most commonly used one (for humans) would be gender, and so it would minimise work if this property was renamed "gender" and a new one was created for sex. Gender roles depend on culture, so I think we should allow a wider range of options than just the Western ones listed by Jared above (e.g. fa'afafine (Q1399232), māhū (Q3277905), kathoey (Q746411), fakaleiti (Q350374), hijra (Q660882)). --Avenue (talk) 23:08, 12 November 2013 (UTC)
    For me sex or gender (P21) should be used for both, with the appropriate qualifier 'instance of (P31):sexual identity (Q1165905)' or 'instance of (P31):gender identity (Q48264)'. If necessary a person can have sex or gender (P21) with two (or more) values, each with appropriate qualifiers - time qualifiers if things change. That way a query doesn't have to guess which property was used for a person; it's always sex or gender (P21). Filceolaire (talk) 00:48, 13 November 2013 (UTC)
    I like that approach. Separate properties would make people be precise about which they meant, which I initially thought was a good idea, but if sources aren't clear about what's meant, perhaps we should leave it unspecified as well. We'll need to allow quite a variety of values though. --Avenue (talk) 01:54, 13 November 2013 (UTC)
    That approach seems problematic to me, given that this property is labeled "sex". Sex and gender have different semantics -- they are indeed different properties of a subject. Using this property should not require users to understand the fine-grained differences between sex, gender, gender identity and sexual identity. Accommodating these overlapping but distinct properties in one property requires convoluted compound statements like those seen in Chelsea Manning. Separating them would enable simpler claims that still capture important information where necessary. Emw (talk) 04:28, 13 November 2013 (UTC)
    So let's change the label. I agree sex and gender are different, but they are usually conflated, and in most cases all we know is the person's gender. If we have separate properties, then the one for sex should usually be absent or "unknown". Editors using the property would need to understand at least that much.
    Convoluted situations will require convoluted statements, regardless of whether we have one sex/gender property or two. In typical situations, the combined property would be simple to apply - and simpler than deciding whether to use the sex property or the gender property. --Avenue (talk) 15:32, 13 November 2013 (UTC)
    Given that this property has an alias 'gender', Jared's changes seem valid. I think it would be best to remove that alias (which he did, then got summarily reverted). Then the current constraints would seem more reasonable.
    Even then, though, the 'one of' constraints currently available for biological sex strike me as needing improvement. Having one pair of sexes for humans (male, female) and another pair of sexes for non-human animals (male animal, female animal) seems like poor design. The claim Fido (Q3744562) sex or gender (P21) male animal is attempting to do too much: we know that Fido is not a human because he's an instance of (P31) dog (Q144) (and dog is not a subclass of human). It would be better design to simply say Fido (Q3744562) sex or gender (P21) male, and let P31 deal with the separate concern of whether Fido is human or not. In other words, Q44148 should have its label changed from 'male animal' to 'male', and Q6581097 ("male" / "male human") should be deleted.
    There is further room for improvement, but the steps above would be a good start. Emw (talk) 03:54, 13 November 2013 (UTC)
    The split between "male human" and "male animal" was just because in some languages it sounds offensive to use the same word for human and for other animals. It was deemed better to use different items than having users complaining everyday about the label (which was what was happening ). --Zolo (talk) 05:27, 13 November 2013 (UTC)
    For the curious, those complaints can be found in the 'Values not making sense' section of this talk page (permalink). Emw (talk) 13:21, 14 November 2013 (UTC)
    Emw: I think we should rename this property "sex/gender" since in most cases we only have one value 'human male' or 'human female' and we are not really sure if that is the persons sex or their gender - since we haven't examined their genitalia. For the small number of people for whom these two values are not enough, we can have other values or multiple values or use these values with qualifiers. We don't need another property. Filceolaire (talk) 08:09, 13 November 2013 (UTC)

    Jared Zimmerman (talk) 09:37, 13 November 2013 (UTC)
    Saying that this does not affect a large number of people is not an appropriate argument against making the change, as a foundation and movement we strive for inclusivity of all race, sex, gender, religion, and cultures. We actively seek to make projects open to all. And while small every change affects how users see the project. Not taking some action here to be inclusive can and will be seen as a move against diversity I think.

    Jared Zimmerman (talk) 15:33, 13 November 2013 (UTC)
    @Emw & Zolo, Also I agree completely that we should merge male animal and female animal in to male and female, It makes very little sense to have these separated out. For both frogs and greek gods it might make sense to have hermaphrite as an alternate, but certainly not for humans.
    Re gender being an alias, it shouldn't be. I thought that it was removed. The point raised that we usually only know gender, not sex, does seem valid. I suppose the property could be repurposed to identity not the biological organs present. Ajraddatz (Talk) 18:29, 13 November 2013 (UTC)


    Jared Zimmerman (talk) 19:01, 13 November 2013 (UTC)
    Let me see if I can summarize what I understand from the conversation:

    Rename sex or gender (P21) as "Gender" alias "sex," "gender identity," "sexual identity"
    Suggested values are male (Q6581097), female (Q6581072), but may also be intersex (Q1097630) rename to "intersex (human)"
    Alternate values are transgender male (Q15145783), transgender female (Q15145782), cisgender female (Q15145779), cisgender male (Q15145778), fa'afafine (Q1399232), māhū (Q3277905), kathoey (Q746411), fakaleiti (Q350374), hijra (Q660882), or other appropriate value.

    For non-humans use male organism (Q44148) rename to “male (non-human animal)”, female organism (Q43445) rename back to “female (non-human animal)”, hermaphrodite (Q303479) rename to "hermaphrodite (non-human animal)"

    Allow as many values as appropriate for this property with qualifiers 'start date', 'end date', 'instance of:gender identity', 'instance of"sexual identity' etc. as appropriate. Be open to other values, and allow for multiple values as well.

    Gender should have the disclaimer that they should respect the individual's choices, sex or gender (P21) is not a means of "outing" a transgender individual. I know this sounds like it muddies the waters a bit but I think it's important. And is in-line with the wording/warning in property of sexual orientation (P91)

    Please feel free to edit the proposal above directly if I've gotten something incorrect.


    My understanding is completely different from yours. I have edited the summary above as my understanding of the conversation above. Filceolaire (talk) 20:08, 13 November 2013 (UTC)

    I don't think we were that far off actually, I've integrated your feedback and updated the proposal, why do you think we should remove cisgender(fe)male? I think that it allows for a different level of specificity than (fe)male especially in the case where a person explicitly idenitfies as such. Jared Zimmerman (talk) 08:15, 14 November 2013 (UTC)

    My thinking was that, in general, "Cisgender male" means pretty much the same as "male" but, thinking about it, that isn't a reason to ban it, especially where a person uses that description for himself. Same for 'cisgender female'.
    Symbol support vote.svg Support as this rewrite, except that I would not rename hermaphrodite (Q303479). There are fictional humans this applies to. Filceolaire (talk) 18:08, 14 November 2013 (UTC)
    Symbol support vote.svg Support, also apart from renaming hermaphrodite (Q303479). We have separate sex properties for animals due to the fact that several lnaguages use different words in that context, and conflating animal/human sex can be offensive there. I'm not aware of any similar problem with hermaphrodites. The most important point is that sex or gender (P21) should be renamed gender, since that is what we will most often be observing. --Avenue (talk) 00:06, 15 November 2013 (UTC)
    I agree with Bever below; "sex or gender" would be a better title than either term alone (at least in English). --Avenue (talk) 03:18, 18 November 2013 (UTC)
    Pictogram voting comment.svg Comment I'm a bit worried about adding cisgender because people trying to use the data won't always realize that they have to include cisgender in their search. So, for example, if I wanted a bot to give me a list of every female mayor in a given year, it will miss all of the mayors tagged as cisgender females unless we have a bot making sure that pages that link to cisgender always also link to one of the three main values. The same problem might also exist for the cultural third genders; a biologically male person who is a Fa'afafine or Fakaleiti should still be included with the males for many or most sorting purposes, the same way a metrosexual in the West would be. --Arctic.gnome (talk) 03:26, 15 November 2013 (UTC)
    Symbol support vote.svg Support, obviously. I'm ok with not renaming hermaphrodite (Q303479) but I do feel we should suggest it not be used for humans in its description (animals/mythology/fiction only when self described of "historians agreement"). Arctic.gnome, I see your point but like Avenue said, better data is better than searchability in this case, I bet we can solve the search problem later.
    Jared Zimmerman (talk) 13:37, 15 November 2013 (UTC)
    • Is it possible to create a constraint saying that if the item has cisgender or a cultural gender as its value, then it must also have one of the three suggested values? If not, then for the sake of sortability I'd rather see sexual identity given its own parameter with a constraint saying that the item must also have a P21 value, like sexual orientation (P91) does. --Arctic.gnome (talk) 23:47, 17 November 2013 (UTC)
    Support adding new possible values, oppose renaming to gender (for the reasons suggested above). Also, if the property is about biological sex, I see what a date qualifier means, but if it is primarily about the subjective sex identity, that become much more fuzzy.--Zolo (talk) 15:03, 15 November 2013 (UTC)
    Note that many languages do not have the difference between the words 'sex' and 'gender' as in English. In Dutch, when people want to be clear that they mean 'gender', they use the English loanword 'gender', because both 'geslacht' and 'sekse' are ambivalent. Bever (talk) 02:00, 18 November 2013 (UTC)
    I agree with Jared's proposal by the way (Symbol support vote.svg Support). Regarding the English label, I understand that 'gender' can be vague, like Zolo said above (under the heading "Rename (en) label 'sex'->'gender'"), but 'sex' is not much better. We could also combine both terms in the label: 'gender or sex', or 'sex or gender', to make the ambivalence immediately clear. That also stops people from thinking it is about grammatical gender... Bever (talk) 02:25, 18 November 2013 (UTC)
    Symbol support vote.svg Support, sensible enough, actually it's true that using "sex" when it should be "gender" is not much better than the other way round :). --Zolo (talk) 08:00, 18 November 2013 (UTC)
    Symbol oppose vote oversat.svg Strong oppose the alternative values. Cisgender is already covered by the items for male and female. Dividing transgender into female and male seems beside the point. If you want to commit to transgender, why commit to a sex again? Or is it gender in this case? And adding all these cultural terms again erodes the purpose of this property. Cultural anthropologists usually inflate these terms in order to highlight how exotic a culture is. It is not true that these terms are so unique to their culture that no other culture can understand them. And it is also not a western/european/(call it what you want)-bias to sort these people into male, female and transgender. In these cases I would even propose setting the property sex to "unknown". Creating a property e.g. "part of subculture" could then store e.g. = "Mahu". --Tobias1984 (talk) 08:58, 18 November 2013 (UTC)
    With many of them fitting as subclasses of male, female, genderqueer (or similar) this argument seems like its more against Cultural anthropologists than against this proposal. Throwing everything that isn't male or female into "unknown" is not only kind of offensive its also discarding plenty of data just because it doesn't fit into the model, rather than adjusting the model. Right now wikidata is growing, and solidifying things beyond the point of change seems rather against the point of the project.
    Jared Zimmerman (talk) 03:22, 24 November 2013 (UTC)
    Symbol oppose vote.svg Oppose. We need a better way to handle the sex-gender distinction and transgender people, but I think the proposed changes cause more problems than they solve. The proposal has several problems that have not been adequately explored or addressed. It's also based on premises that seem flawed.
    • 1. "In modern English people are categorized by gender, not sex."
    That's an overstatement. Yes, the gender-specific pronoun used to describe a person in most modern-English media is determined by the person's gender expression. But a gender-specific pronoun in itself is not much of a categorization -- more consequential categorizations in modern English are based on sex, not gender. For example, US government forms generally categorize people by sex, not gender. This includes the United States Census, upon which much demographic and economic research is based, and Social Security records. (Interestingly, that SSA document conflates gender with sex, but only uses language about sex and biological attributes to define the criteria to change gender in Social Security records.)
    I'm sorry I don't understand your point. citing government recordkeeping methods isn't a great argument I think? as its something rather mired in history that doesn't really acknowledge those outside the norms. It seems that if the issue is that you want there to be some 1:1 between government records and wikidata items having the primary gender identities as subclasses of male and female and the non-binary genders be subclasses of another 3rd property like non-binary (Q48270) will resolve this.
    Social security is a bad example as it is one of the largest databases and therefore slow to change, lets not emulate something so dated with the open possibilities we have before us with wikidata.
    Jared Zimmerman (talk) 03:12, 24 November 2013 (UTC)
    My point was to demonstrate that saying "In modern English people are categorized by gender, not sex" is an overstatement, and thus not a valid premise to base the proposed changes upon. None of your assertions above contest that.
    Whether the US Social Security Administration allows categorizations for those outside the norm doesn't seem relevant to the specific question of whether people are classified by gender -- i.e. an expression or identification with certain societal roles and norms -- or sex -- i.e. biological properties. The question is about which property people are categorized by, not which values within that property are permitted. For example, if some reliable source for categorization of people accommodated those outside the norms of sex and allowed an "unspecified" or intersex value in how they categorized people, then they would still be categorizing people by sex, not gender. Emw (talk) 16:29, 28 November 2013 (UTC)
    Traditionally sex is determined by asking the person what sex they are or by looking at the sex they present as. Even after a person married infertility was common enough that failure to reproduce would not call a persons sexual identity into question. Filceolaire (talk) 15:15, 16 December 2013 (UTC)
    Filceolaire, that doesn't seem to be contesting the fact that major governmental entities like the US Social Security Administration are concerned about the biological attributes of sex rather than non-biological attributes of gender expression. As you can see in the referenced documents, SSA requires more than self-identification. Also, whether the US Census asks a person what their sex or determines what their sex is by examining gender expression (for reference, they do the former) does not change the fact that they are asking specifically about sex and not gender identity or gender expression -- please see the document linked above. Also, a note about terminology: sexual identity != gender identity (!= gender expression, etc.).Emw (talk) 13:58, 23 December 2013 (UTC)
    Emw The US Census document you linked to says that there is a lot of confusion between sex and gender. I am guessing they are referring to earlier censuses. Given this confusion, it is better for us to have an ambiguous property with qualifiers in those few cases where we have better information. The US Census is trying to improve the quality of it's data by being more specific about how this question should be answered and when that Census data is released to the public (in a hundred years) then we can incorporate it into Wikidata. In the meantime ..... Filceolaire (talk) 00:10, 26 December 2013 (UTC)
    Filceolaire, the document goes to pains to emphasize that sex and gender are not the same, and that they are specifically interested in sex, not gender. Also, the document uses the present tense throughout, giving no indication that they are referring to only earlier censuses and not future ones as well. For others' convenience, here is what the US Census Bureau says about sex and gender:
    In general discussions, the concept of gender is often confused with the concept of sex, and the terms are used interchangeably. The meanings of these two concepts are not the same: sex is based on the biological attributes of men and women (chromosomes, anatomy, hormones), while gender is a social construction whereby a society or culture assigns certain tendencies or behaviors the labels of masculine or feminine. These assignments may differ across cultures and among people within a culture, and even across time. Gender may or may not correspond directly to sex--depending on the society or culture or period. That means, for example, that people may associate themselves with femininity (as defined by their culture) while being biologically male. At the Census Bureau, the sex question wording very specifically intends to capture a person's biological sex and not gender. Ambiguity of these two concepts interferes with accurately and consistently measuring what we intend to measure--the sex composition of the population.
    This Issue 1 with the proposed changes is independent of the question of whether sex and gender should be represented in one property or two properties. It is concerned specifically with whether people are categorized by gender and not sex. The documents from the US Social Security Administration and Census Bureau demonstrate the assertion that "in modern English people are categorized by gender, not sex" is an overstatement. Emw (talk) 18:14, 26 December 2013 (UTC)
    I accept that is an overstatement. I agree that sex and gender are not the same. I agree that having separate properties for sex and gender would be more accurate and specific than having separate properties for these two properties. The problem is that specific properties need specific information and in many cases we do not have specific information on biological sex and cultural gender. Given this lack of information we then need to choose between
    • leaving it blank except for the few cases where we do have specific information on (biological) sex and/or (cultural) gender.
    • guessing the (biological) sex based on the limited information we have.
    • using an ambiguous property like 'sex (or gender)' which reflects the ambiguous information we have, supplemented by the 'specifically' qualifier where we have more specific information.
    I favour the third option. Filceolaire (talk) 16:28, 27 December 2013 (UTC)
    Filceolaire, I think the information we have in virtually all cases unambiguously concerns at least gender expression. Can you think of an example where it would concern sex and not gender? For reasons I describe in my reply to you regarding correlation and inference with sex and gender, I think making statements about biological sex based on information about gender expression is not a guess, but a statistically sound inference. Research estimates that only 0.3% of the population is transgender. This means at most about 1 out of 333 people would have an incorrect 'sex' statement based on gender information. Thus, while gender is not a perfect signal of sex, it seems like a sufficiently accurate signal of sex. Emw (talk) 00:23, 28 December 2013 (UTC)
    Emw We could make an inference, knowing that it would be wrong in 3,000 cases in every million or we could find some other way of recording this which would be wrong even less frequently. Remember that there are a lot of people on Wikidata. Now that we are creating items for non-notable children, husbands, and wives of notable people and authors of reference texts, the number will shoot up even further and we will have even less info about these people. There is a strong feeling against classifying people as homosexual or transgender based on inferences; I would prefer not to make inferences about anyone if we can come up with a way of encoding what we know without making inferences.
    From what you say this could be achieved by redefining P21 as "Gender expression" and adding new properties for 'biological sex' and for 'gender identity'; properties which would only be used where we have more information (e.g. where the queen's labour pains were observed by a panel of witnesses). Remember that en:intersex lists 5 different factors which can be used to determine a persons biological sex. There will be a very few cases where we will have accurate information on all 5 of these and will want to be able to record all of these in those few cases.
    or we could allow P21 to be used for all the various meanings of male and female with qualifiers for the fewer than 1 in three hundred persons where we have additional information on exactly what is meant by 'male' or 'female' or 'intersex' in those cases. This is the solution I prefer. Filceolaire (talk) 21:40, 28 December 2013 (UTC)
    Filceolaire, to be clear, stating that a person is biologically male or female based on gender expression information is categorically different than stating that a person is transgender or homosexual based on gender expression information. I am espousing the former, not the latter. In this approach, a false positive would mean that a transgender person would be incorrectly said to be a cisgender person. There would be no cases where a cisgender person would be said to be transgender.

    As you note, Wikidata is creating items for people that do not meet notability criteria on Wikipedia. Information on these individuals is sparse. If we want to make P21 claims about them, then inference is inevitable -- even if we confine our claims to gender expression and not sex. To my understanding the "source" for virtually all P21 statements is a combination of gender-specific pronouns and gender-specific names, which, assuming the as-yet-undefined criteria for gender expression includes more than a person having a certain name and being called 'he' or 'she', suggests that the error rate even for strictly gender expression is significantly higher than 0.3%. This undefined nature of gender expression means we are almost certainly making inferences for cultural gender like we are for biological sex. I'd go further and assert that insofar as gender expression has more meaning than a gender-specific pronoun, using our current criteria for assigning P21 likely has a much higher error rate for gender expression than it does for biological sex. Emw (talk) 13:54, 30 December 2013 (UTC)
    @Emw:: I agree. And that is why I want to use an ambiguous non-specific property as that reflects the ambiguous non-specific information we have. Then use a qualifier like 'instance of' or 'specifically' to create more specific imitation subproperties for those cases where we have more specific information. Filceolaire (talk) 00:54, 31 December 2013 (UTC)
    Filceolaire, both sex and gender have meaning. Labeling this property "sex (or gender)" implies that unqualified P21 claims are making an assertion about sex or gender or both. But we don't have direct evidence of either: a pronoun does not entail that a person has certain chromosomes or anatomy nor certain behavior or styling. A pronoun is not strict evidence of gender. Strictly speaking, if we want to avoid inference, then we must say the information we have in virtually all cases is evidence of only gender attribution. Gender attribution is "calling oneself a woman, being called a woman" -- a label with no meaning; only what a person is called, and nothing more whatsoever.
    If we are not making assertions about sex or gender, then labeling this property "sex (or gender)" is misleading. If we want to accurately reflect the ambiguous, non-specific information (i.e. pronouns) that we have, then we should change the label of this property to 'gender attribution' or 'grammatical gender'. It would be precisely what you're arguing for.
    Such usage of P21 would be error-free, but meaningless. Meaningful and occasionally erroneous claims are much better. Since noone supporting the updated proposal has provided a non-circular definition of gender nor contested the notion that making claims about gender would yield much higher error rates than claims about sex, using unqualified P21 statements as assertions of sex seems to be best option for this property. Emw (talk) 20:16, 1 January 2014 (UTC)
    • 2. "Unless you examine someone gonads, you have no way of knowing if someone is male, female, or intersex."
    This is an unreasonably high threshold of evidence to make a statement about an individual's sex on Wikidata. Wikidata editors should not have to perform a sex organ examination nor conduct a karyotype on someone to state that they are sexed male or female. If someone is said to be the biological father of another person, then Wikidata should be able to state that someone is sexed male. Suggesting that such a claim is unfounded without a paternity test (as has been done in discussions above) is bizarrely stringent.
    Why do we need to introduce extreme standards of evidence to make assertions about someone's sex? To my understanding, that would only be necessary if we had some reason to believe that there was a statistically significant amount of people who were transgender and did not disclose their transgender status. Is there any evidence to suggest that is the case? I'm not aware of any such evidence.
    Agreed, I don't see how this is an argument against this change, also it seems to invalidate your own argument you make in point 7.
    Jared Zimmerman (talk) 03:12, 24 November 2013 (UTC)
    (Note: Jared's reply above was originally inserted after the paragraph ending "bizarrely stringent". To avoid confusion arising from complex indentation, I've moved Jared's comment below my complete comment, as is convention in Wikimedia talk page discussions.)
    The assertions I make in point 2 immediately above are complementary, not contradictory, to those I make in point 7 below. The gist of point 2 is that requiring a sex examination to evidence sex statements is unreasonable, and the gist of point 7 is that, in the absence of signals to the contrary, evidencing sex statements based on secondary sex characteristics is reasonable. The sentence in bold is what my point is rebutting, not supporting. Emw (talk) 16:29, 28 November 2013 (UTC)
    Emw I don't know many transgender persons (5 that I know of) but none of those I do know advertise their status if they can avoid it. They are in fact very careful to conform to conventional images of sexual identity as far as they possibly can. So that is anecdotal evidence that a significant proportion of transgender persons conceal their status - a greater proportion than those who reveal it. Filceolaire (talk) 21:10, 15 December 2013 (UTC)
    Filceolaire, about how many people do you know? A 2011 research article from UCLA estimates 0.3% of people in the United States are transgender. I'd like to base this conversation on research instead of anecdote -- I'm sure there are plenty of people who know 0 transgender people, but that doesn't seem like a valid basis for this discussion. Do you know of any academic research on what the rate of transgender non-disclosure in anonymized surveys that would bring the 0.3% transgender figure into question? Emw (talk) 13:58, 23 December 2013 (UTC)
    If you have a 'sex' property that is specified as strictly reflecting the persons 'biological sex' then that specification is what sets the 'unreasonably high threshold of evidence'. If you have a 'sex (or gender)' property that is specified as reflecting the sex (or gender) of a person as generally accepted' then there is no need for a high threshold of evidence'. The same applies if you have a 'gender' property that is strictly limited. You need to have evidence that the person complies with those strict limits before you can use it. We could end up having two properties but not able to use either one because we don't know which applies to a particular person. Filceolaire (talk) 00:10, 26 December 2013 (UTC)

    I disagree on two fronts.
    First, defining a property as specifically concerning a certain set of attributes does not mean that we must have direct evidence of those particular attributes to make a claim about the property. In this case, defining sex as concerning attributes of chromosomes, anatomy and hormones does not mean we need clinical data to make statements about sex. We can reasonably use one set of attributes to make statements about another set of attributes if the two sets very highly correlated.
    For example, in a hypothetical world in which there were 1 transgender person and 7 billion cisgender people, then requiring examination of sex organs and chromosomes to make statements about sex on Wikidata would be absurd. Gender would still not entail sex, but the false positive rate would be so low that would could reasonably infer sex from gender. This is why I return to the fact that research estimates 0.3% of people in the United States are transgender. This means biological sex can be correctly inferred from gender expression in at least 99.7% of cases (about 332 out of 333), at least recently for people in the US. Now, one might say that accuracy is too low, but that begs the question: what is a reasonable level of accuracy in inferring people's sex? 99 out of 100, 999 out of 1000, 7 billion minus 1 out of 7 billion?
    Second, your comment about a separate gender property seems like a non sequitur. Having a property that concerns gender and not sex does not suggest that we need a higher threshold of evidence to make statements about gender. You have consistently argued that the source material for P21 covers gender expression, not sex. If there is evidence for property A but not property B, then having a separate property for only A does not suddenly mean we need more evidence for A. Emw (talk) 18:14, 26 December 2013 (UTC)
    Emw: As I said above, we should avoid making inferences. As Wikidata will be the easiest database to access and one of the most comprehensive therefore any information we put in it will likely spread far and wide and pollute all other databases making it very difficult to correct mistaken inferences. 1 error in 300 will mean we have thousands of items with the wrong information. Let us look at the alternative proposal which should have a much lower error rate and which will not give our users any false impressions as to the accuracy of our information. Filceolaire (talk) 21:52, 28 December 2013 (UTC)
    The discussion on inference of sex or gender is ongoing in Issue 1 above, specifically here and comments proceeding that. Emw (talk) 13:56, 2 January 2014 (UTC)
    • 3. A similarly stringent threshold for establishing gender is not being applied.
    If we need to "examine someone's gonads" to determine their sex, then a similar level of stringency should be required to establish someone's gender. For example, a comparable test for gender would be requiring a statement directly from the subject that they self-identify as a male, woman, or some other gender. Of course, such statements would be awkwardly explicit and unlikely to come by -- but since the suggested threshold for sex evidence is so high, it seems reasonable.
    I don't believe anyone in favor of this change is making this argument, so again, not sure what your point is.
    Jared Zimmerman (talk) 03:12, 24 November 2013 (UTC)
    Right. My point was to put into perspective the unduly stringent thresholds for establishing sex that have been proposed, by proposing a similarly stringent threshold for establishing gender. Arguing for an extremely high threshold of evidence for one property (sex) and not for another (gender) as folks have done on this talk page is inconsistent. Emw (talk) 16:29, 28 November 2013 (UTC)
    • 4. Specific criteria for establishing gender have not been proposed.
    With all the pedantry surrounding criteria for establishing sex, one would expect some specific criteria for establishing gender. What are they? Getting explicit declarations of gender self-identification directly from a subject isn't feasible, so how would claims about gender be supported? Circular and vague criteria like "masculinity", "femininity", "presents/expresses as female/male" seem prima facie inadequate.
    I think claims would only need verification if there was some sort of general disagreement. I don't think anyone is arguing for this in general, only when there is some complexity involved. In general as you say in point 7 its is "usually" obvious which gender identity an individual is expressing, when it is not obvious statements from the individual will need to be used as a source, much the same as with sexual preference
    Jared Zimmerman (talk) 03:12, 24 November 2013 (UTC)
    Point 7 is not about establishing statements about gender identity, it is specifically about doing so for statements about sex. My point here is that those who have been suggesting ironclad criteria for establishing statements about sex have -- rather inconsistently -- not proposed any specific criteria for establishing gender. Like circular criteria mentioned at the end of my post above, vague criteria like "usually obvious" do not seem adequate for statements about gender. Criteria for establishing gender need to be more specific. Emw (talk) 16:29, 28 November 2013 (UTC)
    Emw The strict criterion for gender is "What sex did (or does) this person present themself as?" Filceolaire (talk) 21:33, 15 December 2013 (UTC)
    The question here is about specificity. Please list specific criteria; "presents/expresses as female/male" is not specific. Emw (talk) 13:58, 23 December 2013 (UTC)
    • 5. Combining sex and gender into one property is a poor design choice.
    Sex and gender are different properties, and should be represented as such on Wikidata. Accommodating these overlapping but distinct properties in one property requires convoluted statements like those seen in Chelsea Manning, and would play out similarly for any transgender person. The sex and gender statements used in Chelsea Manning are poster children of the type of kludge we should be avoiding in the design of statement conventions. Representing a transgender person should be as simple as saying gender female, sex male.
    I don't think any transgender person would be particularly happy or accepting of this "simple" classification you proposed. Jared Zimmerman (talk) 15:03, 18 November 2013 (UTC)
    Can you elaborate? By "simple", I don't mean to suggest that male and female be the only possible values for gender. Instead I mean that distinct properties should be represented as distinct properties. Others have proposed this; see Separate fields for 'sex' and 'gender'. Emw (talk) 04:50, 20 November 2013 (UTC)
    That was my original proposal but that brings you to the issue where transgender people might have male (Q6581097) and transgender female (Q15145782) which might not only be confusing to many people, it is also pretty insensitive. So based on the arguments proposed by others I came around to the idea that a single gender property that expressed the desired gender presentation of a person would be the best single property to have. In actual practice what cases is the biological gender of a person more important than the way them present themselves?
    Jared Zimmerman (talk) 10:15, 20 November 2013 (UTC)
    I think your original proposal was on the right track by distinguishing sex from gender. I also agree with your removal of gender as an alias of this property (diff). However, I think the expanded values in your original proposal could be improved. You suggested we make transgender male, transgender female, cisgender female, cisgender male and genderqueer valid values for gender. The problem I see here is having a value 'transgender female' for gender violates a separation of concerns. The term 'transgender female' concerns both sex and gender; according to transgender female (Q15145782), the term denotes a "biological male who self-identifies as female". If we assert that gender is defined specifically by gender identity, i.e. "one's private sense of being a man or a woman, consisting primarily of the acceptance of membership into a category of people: male or female", then transgender female is not a gender; male, female, etc. are. In that case, compound classifications of sex and gender could be made by reasoners -- gender male, sex female: transgender female; gender female, sex male: transgender male; gender male, sex male: cisgender male; gender female, sex female: cisgender female; gender other: genderqueer, etc.
    If you'd like to discuss practical cases in which the biological sex of a person is more important than the gender-specific way they present themselves, then I'd be happy to. But let's move onto that after we come to some resolution to the topic above. Emw (talk) 13:36, 22 November 2013 (UTC)
    Combining these values seems to solve more issues that is causes, I think there will be few few instance like that of Chelsea Manning where public figures change gender at a point late in life in such a public way which would warrant the qualifiers used on that item.
    The one important point you failed to address was in what cases is representing somes biological gender more important than an individual's gender identity? I think without a valid use case for this this argument doesn't really hold up.
    Jared Zimmerman (talk) 03:12, 24 November 2013 (UTC)
    I've given a detailed explanation above about how combining values like "male" and "transgender male" into one property is a clear violation of the principle of separation of concerns. The frequency of cases like Chelsea Manning doesn't seem relevant to that. Whether such cases are common or rare, it would still be a better idea to have the separate concerns of gender and sex represented in separate properties.
    The question at hand is whether it is sound from a data modeling perspective -- i.e., whether it is good or bad design -- to represent sex and gender in one property rather than two properties. Whether there exist cases where sex is more important than gender is irrelevant to that question. That said, examples of such cases include family trees, pedigree charts, analyzing X-linked genetic disorders and biological conditions more broadly, professional sports and demographic records. Again, to emphasize: biological sex and gender identity are different properties of a person, and thus should not be represented as one property; whether sex is more important than gender is irrelevant to that assertion. Emw (talk) 16:29, 28 November 2013 (UTC)
    Combining sex and gender in one property reflects the confusion over whether stating that someone is 'sex:female' is a statement about their biological sex or about their gender. We know that in most cases the biological and gender sexual identity match. We know that in some cases they don't. We have no way of knowing which case applies for most of the persons on wikidata and making two statements would misrepresent what we know. Filceolaire (talk) 15:15, 16 December 2013 (UTC)
    That's an argument for labeling this property "gender expression" (note: not gender, but that could also mean gender identity, which is much more difficult to establish), not "gender or sex". Emw (talk) 13:58, 23 December 2013 (UTC)
    • 6. Conflating gender with sex would invalidate useful constraints.
    The constraints in properties listed at Wikidata:List_of_properties/Person#Relationship assume P21 has biological meaning. Conflating both gender and sex into this property invalidates that useful assumption.
    To me this sounds like "Technical limitations allow me to be an insensitive ass" (not calling you an ass, just the logic)
    Jared Zimmerman (talk) 10:15, 20 November 2013 (UTC)
    Elaboration would help here. Refraining from "no offense"-like comments would help too. Emw (talk) 13:36, 22 November 2013 (UTC)
    Jared Zimmerman (talk) 18:07, 22 November 2013 (UTC)
    Emw, seriously was not meant to be an attack on you, I just mean using a technical excuse for doing something insensitive is not an excuse for acting insensitively.
    You've simply rephrased your previous post without elaborating. So I'll take a guess at what you mean.
    My guess is you think that properties like mother and father should not be constrained by a person's sex. To that, I would reply that the notion that mother and father can be based on gender stereotypes independent of biological sex seems to be unsupported by reliable sources. Saying that a transgender man (i.e., a biological female that self-identifies as male) can be a father is at the very least a violation of the principle of least astonishment. More likely, though, such an assertion is simply incorrect by definition. Even in exotic cases like that of Thomas Beatie -- in which a trans man retained female sex organs and XX karyotype -- the argument for labeling the subject father is predicated on the idea that they are sexed (not simply gendered) male.
    A final note here: construing property constraints as "technical excuses" suggests to me that you might be unfamiliar with some important features of the Semantic Web. Constraints are basic features of W3C recommendations like OWL (see, and they underlie compelling applications based on semantic reasoning. Constraints are not bugs as indicated by labeling them "technical excuses". On the contrary, constraints are essential features of the Semantic Web, and thus Wikidata. Emw (talk) 05:03, 23 November 2013 (UTC)
    Again this seems solved by the change to use 3 (or more) classes with subclasses as holders for binary and non-binary genders, which will preserve these constraints. But to answer your other question do I think that properties like mother should not constrained by gender? perhaps, but lets leave that to be a separate issue.
    03:12, 24 November 2013 (UTC)
    Whether properties like mother and father should be constrained by sex is not a separate issue -- it is the founding premise of my point. Your new idea would discreetly but fundamentally alter the definition of male (Q6581097), without a robust rationale or basis in Wikidata convention. That item is about sex (not gender identity, as your new idea would have it) as shown by the fact that the corresponding article, male, explicitly states it is about the male sex, and confines itself to describing biological maleness. Subclassing transgender as your new idea proposes doesn't solve the issue. Not only is it questionable in itself as described in this post, it also doesn't address any of the problems described in detail in my previous post. Emw (talk) 16:29, 28 November 2013 (UTC)
    Emw Property P21 may have biological meaning but in most practical cases all we know is what sexual identity the person presented to the world i.e. their gender identity. Filceolaire (talk) 21:33, 15 December 2013 (UTC)
    Filceolaire, it seems that important distinctions have gone overlooked, so I will emphasize them again. First, sexual identity is not gender identity. Please read at least the beginning of the linked sexual identity article. Second, gender identity is not gender expression. Gender expression is what is "presented to the world" -- not gender identity, as you mistake above. Gender identity matches gender expression in cisgender people and transgender people who live openly, but it does not match in a notable group of people: those who internally feel like one gender but behave as another.
    I think what you mean to say is that "in most practical cases all we know is what gender a person expresses and not what sex they are". You are essentially asserting that biological inferences cannot be drawn when a subject is said to be a mother or father. As explained above in my reply to Jared, I think that is unreasonable. Emw (talk) 18:14, 26 December 2013 (UTC)
    • 7. It is reasonable to make statements about people's sex without examining their gonads.
    Secondary sex characteristics and other sexually dimorphic traits are very reliable (though not infallible) signals of a person's biological sex. Non-biological properties of a person -- gender-specific pronouns used to describe them, whether they express stereotypical gender behaviors -- can often be used to corroborate that signal. Unless there is evidence to suggest that there is a statistically significant amount of undisclosed transgender people among the general population, then using these traits as a substitute for a sex organ examination is entirely reasonable. Overriding this evidence for known transgender people would be straightforward.
    Yes, again, this seems like a valid statement that is not against this proposal, and I don't know that people are proposing otherwise when supporting this proposal.
    Jared Zimmerman (talk) 03:12, 24 November 2013 (UTC)
    Several of those supporting this proposal and similar notions have based their position on the premise that we cannot make statements about people's sex without examining their gonads or chromosomes. As I describe above, that's false: we can quite reasonably make such statements using derived traits, and refine them when there is contradictory evidence. Note that this point is complementary, not contradictory, to point 2. Emw (talk) 16:29, 28 November 2013 (UTC)
    Emw In my limited experience, as I wrote elsewhere on this page, transgender persons take great pains to make sure that their secondary sexual characteristics conform to their gender identity - exaggerating or concealing breasts, practising speaking in a lower or a higher register, growing a beard or getting electrolysis to remove unwanted hair permanently. They also, in my limited experience, take pains to conform to cultural stereotypes - transgender women being more likely to wear skirts than the average woman for example, practising male or female mannerisms as appropriate. This means that these sexual characteristics are, in practice, a better indicator of gender identity than they are of biological sex. Filceolaire (talk) 21:10, 15 December 2013 (UTC)
    Please see my reply to you in Issue 2, where I address this. Emw (talk) 18:14, 26 December 2013 (UTC)
    Feel free to comment above and indent responses below the relevant point. Emw (talk) 13:25, 18 November 2013 (UTC)

    Updated proposal[edit]

    I feel like we're pretty close to something since we only have two in disagreement, and much of the opposition seems to be related to data structure and searchability (addressed below)
    Based on the notes of those in support and opposition I'll summarize a new proposal:

    Rename sex or gender (P21) as "Sex or Gender" alias "sex," "gender identity," "sexual identity," "Sex," "Gender"
    Suggested values are male (Q6581097), female (Q6581072), or intersex (Q1097630) rename to "intersex (human)"
    Alternate values are transgender male (Q15145783), transgender female (Q15145782), cisgender female (Q15145779), cisgender male (Q15145778), fa'afafine (Q1399232), māhū (Q3277905), kathoey (Q746411), fakaleiti (Q350374), hijra (Q660882), or other appropriate value.

    Suggest the use of one value but allow as many values as necessary for this property with qualifiers 'start date', 'end date', 'instance of:gender identity', 'instance of"sexual identity' etc. as appropriate.

    "Have the disclaimer that the property should respect the individual's choices, "sex or gender (P21) is not a means of "outing" a transgender individual" or similar wording that is in-line with the wording/warning in property of sexual orientation (P91)

    Make transgender male (Q15145783) and cisgender male (Q15145778) subclasses of male (Q6581097) and transgender female (Q15145782) and cisgender female (Q15145779) subclasses of female (Q6581072). Make other sex/gender options above a subclass of either male (Q6581097) or female (Q6581072) where appropriate and a new property (non-binary (Q48270), third sex, gender non-binary, etc.) where needed for searchability. (new equally long discussion for this umbrella class will need to be had, I'm sure)

    For non-humans use male organism (Q44148) rename to “male (non-human animal)”, female organism (Q43445) rename to “female (non-human animal)”, hermaphrodite (Q303479).

    Jared Zimmerman (talk) 06:03, 23 November 2013 (UTC)

    Symbol oppose vote.svg Oppose This new proposal does not resolve any of the multitude of problems of the previous proposal. For convenience, here's the list of problems, with links to detailed explanations of each:
    1. A premise of this proposal, that "In modern English people are categorized by gender, not sex", is an overstatement.
    2. A direct sex organ examination is an unreasonably high threshold of evidence to make a statement about an individual's sex on Wikidata
    3. A similarly stringent threshold for establishing gender is not being applied.
    4. Specific criteria for establishing gender have not been proposed.
    5. Combining sex and gender into one property is a poor design choice.
    6. Conflating gender with sex would invalidate useful constraints.
    7. It is reasonable to make statements about people's sex without examining their gonads.
    To keep discussions about those points manageable, please comment in the linked sections. Emw (talk) 14:14, 23 November 2013 (UTC)
    BA candidate.svg Weak oppose I think it's problematic to include some aspects of gender identity but not the entire spectrum. Maybe allowing transgendered as an option makes sense. However, if we're including cisgender and the cultural gender identities, what is the rational for cutting out the minutiae of personal identities that relate to gender in western culture, like tomboy (Q845799) or metrosexual (Q192084) or househusband (Q7604987) or lipstick lesbian (Q899836). I don't see how these kind of identities are less worthy of inclusion than the cultural gender identities. I say we either make it biological-only or we include everything related to gender. Including everything would water-down the usefulness of the property, so I'd rather keep this biological only (though still allowing start and end dates as qualifiers). --Arctic.gnome (talk) 03:56, 29 November 2013 (UTC)
    I support keeping this property biological-only, as you suggest. I think it'd be best to have a separate, new property labeled 'gender' for gender identity claims. (Note: this has already been proposed by Sj in Separate fields for 'sex' and 'gender'.) Emw (talk) 13:39, 29 November 2013 (UTC)
    The dichotomy between 'sex' and 'gender' might be a too simple presentation of reality. Besides biological sex, the is the gender someone feels belonging to, the gender role someone uses in public life, the gender identity others attribute to the person, and legal gender/sex. In legal documents and registrations, both the words 'gender' and 'sex' seem to be used, and the relationship between legal, biological and social reality can vary from country to country. So when this property has to be split, there need to be at least three different properties.
    This would give a lot of redundancy, since for items, the three properties have the same content. Also many users will add only one of the properties, so the knowledge about persons will be divided between three properties, we'll be ending up with a lot op people having only a biological sex, others having only a gender and yet other people only having a legal gender/sex.
    Therefore, I prefer strongly the use of qualifications within one property, as proposed before. For simple cases, it is enough to add this property without qualifications. In complex cases, using qualifications persons can get a good description of there sex/gender situation and history. Bever (talk) 20:08, 14 December 2013 (UTC)
    Bever, the issues you cite don't go away by conflating the two separate properties of cultural gender and biological gender into one property. The redundancy you assert would exist by splitting sex and gender information into separate properties would also exist with qualifiers -- we would still have two claims, one for sex, one for gender, whether it existed in separate properties or separate qualifiers in a single property. Using qualifiers would simply be a more complex way to specify that information -- it is a poor separation of concerns.
    I think a reasonable way to represent gender identity would be to specify it only when it doesn't match the statistically normal value. In other words, for the overwhelming majority of cases where a person has a cisgender identity, cultural gender information would be assumed and omitted from the item. This would help reduce uninformative statements whether they were made via two properties or two qualifiers on one property. Emw (talk) 21:55, 14 December 2013 (UTC)
    My previous remark was based on the assumption that it would be possible to omit the qualifier in cases where the sex and gender are the same. Then we do not have to bother which gender/sex definition applies, and the English label from Jared's latest proposal, 'sex or gender', could be read as 'sex and gender'. For transgender people, two claims would be needed indeed. Bever (talk) 22:41, 14 December 2013 (UTC)

    Bever, Filceolaire, yes, two claims would be needed for transgender people and one would be needed for cisgender people. My point in my previous post is that the same would be true if we split sex and gender into separate properties. However, one effect of the poor separation of concerns entailed by the two-properties-in-one approach is that it would make statements for transgender people more complicated than necessary. Compare:
    Option A: Two properties in one property (Filceolaire's proposal):
    sex and gender identity (P21) male (Q6581097) specifically sex (Q290)
    sex and gender identity (P21) female (Q15304678) specifically gender identity (Q48264)
    sex and gender identity (P21) male (Q6581097)
    Option B: Two properties in two properties (Sj's proposal), omitting gender identity claims for cisgender people:
    sex (P21) male (Q6581097)
    gender identity (Px) female (Q15304678)
    sex (P21) male (Q6581097)
    Option B is better. It properly organizes the separate properties of biological sex and gender identity into two separate properties. Following from that, while the number of claims in either option is the same (2 for transgender, 1 for cisgender), the claims in Option B are less complicated than those in Option A. Emw (talk) 15:26, 15 December 2013 (UTC)
    Emw Option B has the big disadvantage that it insists that Chelsea Manning and every other transgender female be labelled with sex:male even if they also have other statements as well. Option B also assumes we have a level of knowledge which we usually do not have. I know of a transgender male with an item on Wikidata where there is some documentation on the internet about this person when he was a woman and there is some more on him as a man but there is no public documentation I could find which links the two together or discusses his transition. The wikidata item uses his male name. I know this is a transgender male but there is no reliable source for this so I guess his item should says sex or gender (P21):male since that is what is publicly known and that is what can be guessed from his secondary sexual characteristics - he has grown a beard.
    If we have separate properties for sex and gender then how do you deal with these cases where we don't know if a persons maleness is due to their biological sex or to their gender identity? Having a combined property deals with these ambiguous case. While it is true that the number of transgender persons is small it is also true that such persons often conceal their condition. Even cisgender persons don't usually discuss their sexual identity much in public. My guess that is for every 100 males there are
    • 98 conventional males who have never discussed their sexual identity in any serious way in public,
    • 1 man who has some gender identity issue but chooses to conceal it and presents himself as a conventional male as far as he can, and
    • 1 man who we actually know enough about to make separate statements about.
    In other words there are 99 out of 100 that we don't know enough about to make separate statements about them. I think we would be wrong to insist on making such statements, based on guesses from their secondary sexual characteristics even if we would be right in 98 cases out of 99. Filceolaire (talk) 20:48, 15 December 2013 (UTC)
    • Filceolaire has a point, if we have a parameter dedicated to biological sex, then we're making an assumption that can rarely be sourced. Keeping this as one property might make sense: P31-Male and P31-Female would be used for the majority of people and would both mean "apparent sex", and editors would have all the freedom they need to explain special cases. Really, my only concern is people who want to sort by sex for social science studies, so as long as male/female, P31-male/Female, cisgender male/female, and trans male/female were all sorted under an appropriate subclass so they could be easily searched, I'll support it. --Arctic.gnome (talk) 22:11, 15 December 2013 (UTC)
    • Filceolaire, a 2011 UCLA research article estimates that 0.3% of adults in the United States are transgender. Gender expression matches sex in about 99.7% of the population, assuming US estimates are a decent proxy for global rates. So, even if all transgender people concealed their status, we're talking about shifting the meaning of this fundamental demographic property to accommodate a very, very small set of edge cases.
    Even then, 1 out of 300 probably far exceeds the number of undisclosed transgender people on Wikidata. The difficulty of being a closeted transgender person is likely much more difficult for people with Wikidata items than it is for the general population. Such taboo information is valuable for tabloids, and correct tabloid speculation often percolates into coverage by reliable sources. At the same time, the successes of the LGBT movement have made it less stigmatic to be a transgender person -- decreasing the pressure to keep one's transgender status a secret. So I think the rate of undisclosed transgender people on Wikidata is likely significantly lower than the total population of transgender people.
    In other words, secondary sex characteristics and gender-specific pronouns would be an accurate signal of biological sex in about 99.7% of cases, and much of the remaining 0.3% could likely be correctly annotated (or at least marked for manual review) by detecting words like "trans", "transgender", "gender" etc. from articles. Realistically, I think we'd be looking at a false positive rate due to undisclosed transgender people of at most 0.03% - 0.15%; i.e. 1 in 600 to 1 in 3000 people on Wikidata incorrectly described as male or female sexed. Thus, it seems unwarranted to shift sex (P21) to only make claims about gender expression by default.
    Finally, a note about terminology. Terms seem to be getting conflated a bit in this discussion: sexual identity is not gender identity, and gender identity is distinct from gender expression (see the HRC glossary). These things are each separate properties. To be precise, you are arguing that we should use gender expression, and not gender identity nor sex, as the default property of this property. (Now we're talking about having three properties in this one property. Yikes! If we actually need these additional properties (I think we need to add at least gender expression), then let's create separate properties instead of making P21 into one ill-defined, unwieldy amalgam of three properties.) Emw (talk) 04:07, 16 December 2013 (UTC)
    Emw: P21 is already being used for gender expression. You argue that secondary sexual characteristics can be used to assess biological sex yet in my experience these are a better indicator of gender expression. Cultural sexual characteristics (hair styles, clothing, which box people tick on forms, etc.) are almost the definition of gender expression. In practice I believe that P21 is more likely to be based on these rather than on secondary sexual characteristics.
    You doubt that there are many transgender people on wikidata who manage to conceal their status and yet, of the few people on wikidata who I know personally, one of them is in exactly this situation. Is my anecdotal evidence better than your guess?
    Using P21 without qualifier accurately reflects our lack of information about most of the people on Wikidata - it probably reflects their biological sex and their gender expression and their gender identity but it might be only two of those. In the few cases where we have more information this can be accurately represented using qualifiers. If we had 3 separate properties then we would have endless arguments as to which is the appropriate one to use with each person. Filceolaire (talk) 15:15, 16 December 2013 (UTC)
    Somebody I know personally (though only superficially) recently changed her first name from a male one to female one. It appears that she has breasts, so it is possible that she had an operation or uses hormones, but it is also possible that they are plastic. In other words, I know her gender but not her actual sex. So the 'sex' property could not be filled in, were she notable enough for Wikidata (which is not entirely impossible, she is an author).
    It seems to me that option A above is better for searches. I have no experiences with searches on properties (that's not (yet) possible with the web search form, is it?) but in option B some (or many) people would not have both properties; perhaps most of them would have at least the sex property, but others only the gender property. This means that you would have to do a boolean OR search on both properties, while in option A you could just do a search on P21 without qualifiers, if you are not particularly interested in the gender versus sex issue.
    One problem which maybe also has to be addressed is what is actually meant with biologically sex, especially when somebody has had a sex change (and we know about it). In Filceolaire's example, since the person has a beard, it seems that he is at least taking hormones, or perhaps even had an operation, so I would say that he is a man biologically as well, but others might say that his biological sex is still female, because of his chromosomes. That is also something I did not understand about the definitions for the transgender/cisgender items: 'transgender female' was defined as 'biological male who self-identifies as female', but if a man has operations to become a woman, this definition would not apply anymore. I see that the description of the item has changed later, so this problem has been solved.
    One last remark, missing is an option for people who feel 'in between'. The words transgender or transgenderism also apply to people who do not feel entirely man or woman. While the number of people feel belonging to 'the other sex' is perhaps 0.5%, those who feel uncomfortable with the male/female dichotomy may number up to 5%. Bever (talk) 23:50, 16 December 2013 (UTC)
    A source about the 5%: mens-en-gezondheid*infonu*nl/diversen/68465-transgender-of-transseksueel-wat-is-het-verschil.html says it in short, the original source of this web page was p. 189 of this 1 megabyte PDF (2006). A more recent report has somewhat lower figures, 4.6 of the men and 3.2% of the women having an 'ambivalent gender identity' (p. 4 of the PDF). (It's in Dutch, but perhaps machine translation gives an impression if you cannot read it.) Bever (talk) 00:41, 17 December 2013 (UTC)
    Bever we also have the valid properties of intersex intersex (Q1097630) and genderqueer non-binary (Q48270) to cover these cases you can also see on in the "One of" section there are other valid options, and the suggestion that other values can be added if nessesary. I, as one of the people who proposed and argued for this change am very happy with the consensus that was reached even though it was not exactly in line with my original suggestion. If you feel something is still missing from the options your suggestion is welcome. Jared Zimmerman (talk) 03:59, 8 February 2014 (UTC)


    I'm going to go ahead and rename this property tomorrow. OK? Filceolaire (talk) 20:48, 22 December 2013 (UTC)
    I'd like to get a wider community review of an alternate proposal before we make this significant change. Please see the property proposal for gender. Emw (talk) 05:58, 23 December 2013 (UTC)
    Emw has now withdrawn the proposal for a separate 'gender' property so I have gone ahead and renamed this property from 'sex' to 'sex (or gender)' and have added aliases based on anything people might be searching on related to this. Filceolaire (talk) 00:37, 26 December 2013 (UTC)
    I'm very happy to have come to a group consensus on this, I think it is actually a big win for being respectful with the way we represent people in the data, without compromising on integrity and principles of the project.
    Jared Zimmerman (talk) 02:31, 29 December 2013 (UTC)

    Still to do[edit]

    I'm a bit disappointed to see the outcome of this discussion being the renaming of this property. I'm sympathetic to Emw's arguments above that there should be separate properties for biological sex and for gender identity. It seems to me that conflating the two concepts, and attempting to use qualifiers to disentangle them, will make for much messier and harder-to-use data, for most use-cases. But, I don't think either solution is perfect, and it looks like the weight of the opinions is on the other side, so that's that.

    In reading through all the discussions above, and the "updated proposal", I see that there are a lot of loose ends. But my main concern has to do with how the outcomes are documentated, so that editors will have some guidance on how to use it. With some good documentation, maybe some of the messiness in the data can be obviated. I would like to work on the documentation, but I don't know the right way to implement it. Suggestions welcome.

    There are others -- the above isn't an exhaustive list. Klortho (talk) 06:11, 2 January 2014 (UTC)

    Klortho, I was always on the fence about the single property, I don't have a strong preference there. I do think we should make all of the non-binary genders instances of something like non-binary (Q48270) with all of the others as instances of male (Q6581097) and female (Q6581072) ✓ Done to your last point about cisgender male and cisgender female, while "cisgender female" always equals "female", is not always the same as "cisgender female" I think these should have a parent child relationship i.e. cisgender female (Q15145779) and transgender female (Q15145782) and should both be subclasses of female (Q6581072). I think this may be partly done already, but not finished. ✓ Done
    Jared Zimmerman (talk) 08:52, 5 January 2014 (UTC)
    Hi, I'm not sure you understood my concern about transgender female (Q15145782), but in looking at it again, I don't think it is a problem. I was concerned that, because it is a subclass of female (Q6581072), which is about biological sex, then a machine could infer that because Chelsea Manning (Q298423)sex or gender (P21)transgender female (Q15145782), therefore Chelsea Manning is a biological female. But, that inference could only be made if Chelsea Manning (Q298423)instance of (P31)transgender female (Q15145782). Klortho (talk) 03:23, 6 January 2014 (UTC)
    Well, I realize I am not the only one confused about this, and, in fact, if the statement transgender female (Q15145782)subclass of (P279)female (Q6581072) is not to be used to infer that a transgender female is a (biological) female, then I can't imagine what it is supposed to be for, so I will delete it. Klortho (talk) 02:21, 8 January 2014 (UTC)
    Klortho, the reason that was does what that during the proposal discussion the point was brought up that you should be able to query for "females" and get "all females" not that you should have to run multiple queries to get this set of items. So the decisions was made then to to make transgender female (Q15145782) → subclass of (P279) → female (Q6581072) so that query would resolve correctly, I don't know that anyone said that female (Q6581072) had to do specifically with biological females.
    Jared Zimmerman (talk) 04:30, 8 January 2014 (UTC)
    Jared Zimmerman, I don't see in the above discussion a description of that rationale, and I have read and/or skimmed it a few times now. I'd still really like to see a list of example cases somewhere, which could serve as documentation on how this property should be used. Following the pattern of templates, could we do this on Property:P21/doc? Even though it wouldn't be automatically transcluded, nevertheless we could reference it on the discussion page, and maybe (if what John Lewis writes is correct) it would be transcluded soon. Klortho (talk) 13:41, 10 January 2014 (UTC)
    Klortho, the discussion took place across a few pages, and might have been archived, I'll see if i can find it over the weekend.
    Jared Zimmerman (talk) 05:24, 11 January 2014 (UTC)

    Spelling mistake[edit]

    Just a spelling mistake: the British English gloss for this property mis-spells the word "species" as "specees". -- 19:39, 17 February 2014 (UTC)

    Anonymous person[edit]

    What I should set if there was some anonymous writer? We don`t know who he was, so I should use "unknown" value, but I don`t see such option in values.~So - what should I use? PMG (talk) 11:56, 1 September 2014 (UTC)

    • Use "somevalue" for this case. But is the anonymous author single person? Or group maybe? This property must not be used for human groups. — Ivan A. Krestinin (talk) 12:17, 1 September 2014 (UTC)
    "Unknown value" and "Some value" are available via the strange three little rectangular boxes that appear next to the value when you click the edit button. (Not the three squares - that is something else). When thre rectangles appear then click on them and select the appropriate option. Filceolaire (talk) 22:42, 16 March 2015 (UTC)


    Recently, in many languages the description was updated, but the last sentence of the English description was omitted in most cases: 'Add qualifiers as appropriate'. I wonder why, is this difficult to translate? It's a pity because I like the qualifier idea. Bever (talk) 22:34, 6 October 2014 (UTC)

    When adding the qualifier sentence to the Dutch description, I encountered a technical problem: the description must not be longer than 250 characters. Luckily I could shorten the first sentence a bit. But would this be the explanation for the omission in the other languages? Bever (talk) 22:50, 6 October 2014 (UTC)

    Hermaphrodite and Transgender[edit]

    I've added two values (which already existed as items) to fit use in Getty ULAN:

    • hermaphrodite (Q303479): described as "used for organisms with both 'male' and 'female' reproductive organs" and coreferenced to GND 4191351-6
      • Note: intersex (Q1097630) is described as "variation in sex characteristics that do not allow an individual to be distinctly identified as male or female". It's coreferenced to GND 4027484-6 "occurrence of male and female or intermediate characteristics in an individual of a single sex type", which has synonym Pseudohermaphroditism so it's not the same as Hermaphrodite
    • transgender person (Q189125): While we have transgender male (Q2449503) and transgender female (Q1052281), ULAN doesn't have such distinction, so we need a "super-value" Transgender in addition to these two

    --Vladimir Alexiev (talk) 13:38, 12 February 2015 (UTC)

    Hi Vladimir, hermaphrodite (Q303479) should not be used for humans, this is considered offensive. It is meant for organisms in which such combinations of male and female organs are common, like snails. I wonder if the description in GND is outdated. Bever (talk) 21:05, 9 March 2015 (UTC)
    Bever hermaphrodite (Q303479) should generally not be used for humans, as it is generally considered offensive. It is meant for organisms in which such combinations of male and female organs are common, like snails or for fictional characters which are described as this or for humans who claim this description for themselves (though I can't think of any examples of this). Filceolaire (talk) 22:33, 16 March 2015 (UTC)
    Hi Filceolaire, sorry for forgetting the word generally and the fictional/mythical characters. :-) Bever (talk) 02:12, 17 March 2015 (UTC)
    There is now an example of a person using the word hermaphrodite for herself, instead of intersex: the Dutch singer Elle Bandita, she dislikes the word intersex/interseksueel. (Hermaphrodite applies to her birth sex, she was raised as a girl, present gender is ambiguous.) (Source: a recent interview in the newspaper de Volkskrant.) Bever (talk) 00:36, 2 March 2017 (UTC)

    Correct typo[edit]

    Hello, I cannot edit the item page - there is a typo in the Esperanto label. "eleku" should be "elektu". Kind regards --Z. (talk) 20:07, 10 March 2015 (UTC)

    Done. Or farite, I should say. Bever (talk) 02:17, 17 March 2015 (UTC)

    Pie chart[edit]

    DONE as of June 13, 2015: 2,641,424 (93.8%)TO DO as of June 13, 2015: 174,789 (6.2%)Circle frame white.svg
    •   DONE as of June 13, 2015: 2,641,424 (93.8%)
    •   TO DO as of June 13, 2015: 174,789 (6.2%)

    Enjoy .. --- Jura 12:44, 13 June 2015 (UTC)

    Under such section title I'd expected to see a distribution along genders/sexes... And "done" requires some comments. --Infovarius (talk) 11:41, 15 June 2015 (UTC)
    There are a few more at User:Jura1/People_charts with slightly more text. --- Jura 11:52, 15 June 2015 (UTC)

    Incorrect Aliases[edit]

    Man, Woman, Male, Female, and Intersex are listed as aliases, but they are not aliases, they are values. "Hey John, what's your male?" doesn't make any sense. This is causing us some issues in our knowledge engine where we utilize wikidata properties, so I wanted to come here and fix the original data (instead of just fixing it in our own database). I verified my email, but Wikidata still won't let me submit an edit to this property - what's up with that? Dylanwenzlau (talk) 07:11, 22 August 2015 (UTC)

    An alias is also a way to find the correct property. Generally, it wont be displayed. To add "male", one needs to use this property. --- Jura 07:07, 22 August 2015 (UTC)
    I agree that finding properties by searching the values is important. Isn't there a more semantically correct way of doing that though? According to Merriam-Webster, alias is defined as ": also called : otherwise known as". male, female, man, woman, and intersex do not match that description, and therefore make it hard for anyone using wikidata properties to be able to depend on them at all. One solution could be to allow users to search the actual allowed values of a property, instead of just a property. Another solution would be to create a "keywords" field that allows keywords to be specified to help with search. Other than this, so far, Wikidata aliases have been incredibly helpful to our project, and we've been able to use them directly. I hope we can continue to rely on the quality and accuracy of Wikidata aliases. Dylanwenzlau (talk) 07:18, 22 August 2015 (UTC)
    It is known that this isn't the best solution (alias is used for both search terms and real aliasses) and the developers are looking at a better way to store search terms, however until then, we have to deal with the system we have even though it isn't perfect. Mbch331 (talk) 07:22, 22 August 2015 (UTC)
    Just wondering, what is your project?
    For anything that needs to be displayed, I think a monolingual string property is preferred. --- Jura 07:25, 22 August 2015 (UTC)
    Our project is, and we use aliases mainly to improve entity & property recognition in search. Dylanwenzlau (talk) 08:10, 22 August 2015 (UTC)

    People without sex or gender (P21) by nationality[edit]

    Nationality Items List
    Japan (Q17) 12485 list
    Turkey (Q43) 3393 list
    Thailand (Q869) 3048 list
    South Korea (Q884) 1958 list
    Indonesia (Q252) 1512 list
    Spain (Q29) 641 list
    United Kingdom (Q145) 470 list
    Finland (Q33) 455 list
    China (Q29520) 282 list
    Haiti (Q790) 239 list
    South Africa (Q258) 212 list
    India (Q668) 182 list
    Romania (Q218) 178 list
    Lithuania (Q37) 128 list
    Iran (Q794) 125 list

    Above a short list from this query.

    Not that this is limited to items with country of citizenship (P27) defined. Only 26,168 of 181,815 have that. --- Jura 18:55, 12 September 2015 (UTC)

    How to handle collective pseudonyms that also have another P31 that conflicts[edit]

    How to handle cases like Iny Lorentz (Q107024) & Claude Izner (Q743688)? The violate a mandatory constraint, because they have a P31 that is in violation. These are pennames/collective pseudonyms that also have another P31 and that P31 is violating the constraint (in the first case it's married couple (Q3046146) and in the second sibling duo (Q14073567). Turn the constraint in not mandatory and add exceptions or remove the violating statements? 07:05, 22 October 2015 (UTC)

    At Property talk:P735, I had to split the constraints (one with exceptions, one without). If it's a group, maybe you should switch P21 to P279. One could argue that gender is novalue. --- Jura 10:19, 22 October 2015 (UTC)
    You mean "switch from instance of (P31) to subclass of (P279) ? Then no, it's incorrect to do that. It must stay "instance of" : partnership or something, not "subclass of" : partnership because it's not a class of partnership like "married couple" for example, it's a specific partnership. If I understood well of course. author  TomT0m / talk page 11:06, 22 October 2015 (UTC)
    I'd say here that the solution would be to link the couple with its name item (or not), maybe switch the item, or just use a monolingual string naming property. Never now which one should be used. author  TomT0m / talk page 11:09, 22 October 2015 (UTC)
    Jura: I would say that a collective pseudonym doesn't have a sex or gender. instance of (P31) should only be used for individuals. That is my opinion. There are a few cases where a human has represented themselves as representing a house name to the point of trying to create confusion between the group of writers and the person who goes to the book signings but these are unusual cases and by no means typical. At least that is my opinion. Joe Filceolaire (talk) 03:54, 25 October 2015 (UTC)

    Fictional animals[edit]

    Are fictional animals to be tagged with male organism (Q44148)/female organism (Q43445)? I see that some items like Lassie (Q941640) use male/female, and I'd like to know if this needs correction, and I'd like to know so I can expand items like Rocko (Q4305579). --AmaryllisGardener talk 02:56, 25 October 2015 (UTC) :Amaryllis: Fictional animals (in my opinion) can certainly have a instance of (P31) claim. Joe Filceolaire (talk) 03:56, 25 October 2015 (UTC) Sorry. Don't know what I was thinking. That answer made no sense. Joe Filceolaire (talk) 04:16, 31 October 2015 (UTC)

    What about anthropomorphic animals (like Mickey Mouse (Q11934), Goofy (Q111135) and Donald Duck (Q6550)) behaving like humans but having the appearance of an animal? The three mentioned use male (Q6581097) and as it seemed reasonable to me I used these human attributes for other anthropomorphic animals, too - while using the non-human attributes (male organism (Q44148)) for non-anthropomorphic animals (as seen in Pluto (Q108732)). Now I came across some anthropomorphic animals using male organism (Q44148)/female organism (Q43445) and I'm wondering if I should "correct" it using male (Q6581097)/female (Q6581072) or if I should "correct" Mickey Mouse (Q11934), Goofy (Q111135) and Donald Duck (Q6550). (I myself have a strong preference for using male (Q6581097)/female (Q6581072) for anthropomorphic animals as they are treated also in every other respect like persons/humans - they have a job etc.)Valentina.Anitnelav (talk) 18:19, 15 December 2016 (UTC)

    @Valentina.Anitnelav: I would not care. My solution would be to make male organism (Q44148) View with Reasonator View with SQID a superclass of male (Q6581097) View with Reasonator View with SQID : and to always query for the superclass.
    For example the SPARQL code ?male_or_masc wdt:P279* wd:Q44148 . ?item wdt:P21 ?male_or_masc . would find all such items. Then if you want to know if ?item it's a "fictional animal" or a "fictional person" the question becomes if it's an instance of fictional character (Q95074) View with Reasonator View with SQID or of fictional animal character (Q3542731) View with Reasonator View with SQID. If it's a mess, we should focus on the classification of fictional figures and create some classes like "fictional person that behaves like a human but that is represented as an animal" or classes like that. author  TomT0m / talk page 09:58, 16 December 2016 (UTC)

    humanoid (Q502931), extraterrestrials in fiction (Q1307329), vampire (Q46721), Koschei (Q1140396), Slavic dragon (Q620939) (volkodlak (Q22815043), werewolf in legend or folklore (Q9410), mythical shapeshifter (Q1133563) - fictional canine (Q19594644)?), Cheburashka (Q161731) - fictional animal character (Q3542731)? Also male organism (Q44148)/female organism (Q43445)? ->fictional systematics -> Original research -> Pseudoscience? --Fractaler (talk) 14:17, 16 December 2016 (UTC)

    sitelinks and P21[edit]

    Also mentioned at Wikidata:Project chat#How many items about people (P31=Q5) have .28P21.29 defined.3F. --- Jura 13:23, 30 October 2015 (UTC)

    Check for the use of person values on animals[edit]

    We should probably try to add some check to ensure that male organism (Q44148)/female organism (Q43445) is used on animals (not male/female).
    --- Jura 09:02, 7 January 2016 (UTC)

    please remove animals/floras, please add grammatical gender and number[edit]

    We should create a separate property for animals, or flora, or other living species. For them, there's no question of personal identity (e.g. "queer" or "transgender"), but only of biological sex (not just male and female, also hermaphrodite and asexual, and some other special natural evolutions between asexual and sexuated types)


    Also, please add separate properties for the grammatical gender (which is also different but tied to languages: masculine, feminine, neutral, or other), and the grammatical number (singular, dual, paucal, plural, or other) for languages that do have such distinctions. Note also that the grammatical gender and grammatical number of a name (or adjective, or participle, or pronoun) do not always match the sex or gender of persons or living species having this name.

    Application: the "Gentilé" property (applicable only to inhabitants or natives humans of a location, but having only a grammatical gender: "queer" or "transgender" is irrelevant here), which is also distinguished from the adjective.

    E.g. for Swizerland, in French:

    • the "gentilé" property (always a substantive noun) is "Suisse" (with the capital) for masculine singular, "Suisses" for masculine plural, "Suissesse" for feminine singular, "Suissesses" for feminine plural; but
    • the "adjective" property (missing for now) is "suisse" (without the capital) for other singular, "suisses" for other plural ("other" here means either "masculine" or "feminine" or a combination of both in plural cases)

    Note that French does not have a distinctive grammatical "neutral" gender, unapplicable genders would be mapped to "other", frequently but not always the same as "masculine"; some rare nouns in French have their grammatical gender changing depending on the grammatical number (such as "amour": masculine singular -> "amours": feminine plural). Note also that the French term for "person"/"persons" is "personne"/"personnes" which is grammatically feminine, independantly of the gender or sex of the person itself.

    Mapping a person's sex or gender to a grammatical gender is language-dependant (don't assume that a named "male" person will map to a masculine grammatical gender, or even to the "singular" grammatical case in all circonstances). The grammatical rules do not follow exactly the semantic of the referenced persons:

    • English itself also has personal gender/sex for humans, but map them in most cases to a single grammatical neutral gender (except for pronouns: "he", "she", "it"), but frequently map them to a neutral plural ("they") for politely refering to a single person with an undetermined gender.
    • French may replace "je" (masculine or feminine singular 1st person) by "nous" (masculine or feminine plural 1st person, including for conjugating verbs, but not for dependant adjectives or participles) ; same remark for "tu" vs. "nous" (2nd person).
    • French has some job titles only existing in (grammatical) feminine forms, even occupied by men (e.g. "sage-femme"); the reverse being also true ("professeur", "docteur", "médecin"... even if there are unsuccessful attempts to feminize these terms). Generally, the concernded persons choose themselves the form of the job title they prefer ("docteur" or "doctoresse" or "docteure" for a woman?), and sometimes even the applicable grammatical gender, independantly of their actual claimed or official sex or gender.
    • The grammatical gender is not restricted to humans, or even to animals, it exists also for objects and abstract concepts.

    For this reason we clearly need separate properties for the grammatical gender and number, even if you apply them only to humans! Verdy p (talk) 08:32, 10 February 2016 (UTC)

    I don't quite understand, where do you propose to use "grammatical gender" property? In future "lexical" (part of Wiktionary) items? --Infovarius (talk) 21:27, 17 April 2016 (UTC)

    Sex and gender data are wrong[edit]

    I see this has been raised many times, but the data on intersex people are wrong. en:Intersex people generally also belong in intersex and female categories by default, or intersex and male categories by default, unless they change assignment to a different category such as intersex and non-binary. Some people such as en:Mauro Cabral Grinspan are male, transgender and intersex, i.e. assigned female at birth, with an intersex condition, and later transitioned to male. It is wrong to describe him as male and not also intersex and transgender, or intersex and not male and also transgender. Some people such as en:Sarah Gronert and en:Maria José Martínez-Patiño are female, assigned female at birth, and have an intersex condition - and have never been anything else, even though their intersex conditions were (or were likely to be) discovered after birth. Discovered during sporting competition for en:Maria José Martínez-Patiño. It is wrong to describe them as female and not intersex, or intersex and not female. Compare en:List of intersex people with the output of the intersex Wikidata feed and these discrepancies are evident. Ideally, in my view, sex should be renamed gender, including female, male, and other, and intersex should be a separate classification. Probably transgender should also be a separate classification also. Trankuility (talk) 22:59, 2 April 2017 (UTC)

    Why used for values only constraint (Q21528958)?[edit]

    @Yair rand: Why did you add used for values only constraint (Q21528958)? I don't see why we should forbid usage as a qualifier in cases where gender matters (like images). ChristianKl❫ 21:45, 4 January 2018 (UTC)

    @ChristianKl: Hm, I can see how that could be useful for images. I don't suppose there's a way to add a constraint only allowing use as a qualifier for certain properties? There are a lot of broken uses of P21 on all sorts of properties where it doesn't belong, and I'd like to be able to mark those as problems without adding qualifier restrictions to all those properties individually. --Yair rand (talk) 21:59, 4 January 2018 (UTC)
    What problems do you see with it's usage? ChristianKl❫ 23:06, 4 January 2018 (UTC)

    Conflicts with "author"?[edit]

    This property has a constraint conflicts-with constraint (Q21502838) author (P50). Surely a fictional character can have a gender and an author at the same time? Deryck Chan (talk) 14:49, 7 March 2018 (UTC)

    • I use creator on these, not author.
      --- Jura 15:04, 7 March 2018 (UTC)

    P21 vs. P518[edit]

    If there are two numeric identifiers (male / female) what kind of qualifier should be used:

    1. geographer (Q901402) uses --> Geschlecht (P21)
    2. occupational physician (Q831831) uses --> betroffener Teil (P518)

    @Jneubert, Nikki: this concerns your edits.

    More examples: Wikidata:Database_reports/Constraint_violations/P227. --Kolja21 (talk) 14:02, 29 May 2018 (UTC)

    Thanks for spotting that and bringing it up here. Since GND uses different identifiers for all male and female occupations, it makes a lot of sense to give a recommendation for one particular qualifier. If P518 is the more common way to handle such situations, I'm fine with that. 14:36, 29 May 2018 (UTC)
    Imho P21 is fine. Then we see at once how many identifiers male / female exist in Wikidata. --Kolja21 (talk) 01:09, 31 May 2018 (UTC)
    ✓ Done Please see Help:P227#Allowed qualifiers. --Kolja21 (talk) 20:50, 27 June 2018 (UTC)

    For humans Gender Identity should be used[edit]

    I believe it would make much more sense to use sex for animals and gender identity for humans. The "sex or gender" is used now for humans is not correct imho. Of course sex is not correct for humans as then only male female and intersex would be viable options. And the gender of a person is exactly their gender identity, but currently WikiData treats it like something in between. E.g.: A Transwoman who identifies as a woman, shouldn't have the property transfemale, but female, because that's what she identifies as. Some transwomen identify as transfemale, but not all do. If I had to guess I'd say most don't. I've seen some Transpersons on WikiData have two labels set for sex or gender: one with their birth-sex and a statement that it ended at one point and another with transfemale or transmale starting at that point. This only makes sense if the person in question clearly stated that they identified with one gender until that point in time and then identified with the other one, but in reality for most transpersons only one gender is correct for their whole life. Coming out or starting your transition is not an event that changes your gender, it's an event that lets the world know which gender you identify as. Also there shouldn't be restrictions for which gender identities can be used for humans. Angel Haze (Q2377185) has 3 references for her "sex or gender", all of them clearly state that she is agender, still it is set to non-binary and can't be changed. Another example: Eliot Sumner (Q262772) doesn't believe in the concept of genders, but still she is set to "non-binary". Non-binary would be correct if a person doesn't believe in a binary concept of genders, but if she doesn't believe in genders at all her gender property should be empty or agender.

     – The preceding unsigned comment was added by PKFP (talk • contribs) at 09:01, 6. Aug. 2018 (UTC).

    • At the very least, make "transgender fe/male" into an allowable qualifier. Right now, you can qualify sex/gender with "start date" and "end date", which is NOT how most trans people understand their gender. But using gender identity would be preferable. LadiesMakingComics (talk) 18:39, 3 October 2018 (UTC)
    • LadiesMakingComics & PKFP: I'm very new to Wikidata, but yeah, when I saw how this site handles gender my response was "oh dear." Whether someone identifies as "female" or "male", and whether they identify as "transgender" or "cisgender" are two totally different things!
    I did some Googling and found this set of recommendations for cataloging "Gender in Name Authority Records". If I'm understanding the recommendations correctly, it describes a schema where a gender category could have multiple individual gender-related statements. For example, the actor Jeffrey Tambor would be categorized under both Males and Cisgender people. A transgender woman from Thailand might be categorized under Females, Transgender people, and the Thai-specific term Katoeys. The two-spirit author Ma-Nee Chacaby would be under Two-spirit people.
    Here are some more recommendations this document makes: 1. "Record information about gender as the person self-identifies and explicitly discloses, taking information from readily and publicly available sources" 2. "Do not assume gender identity based on pictures or names" 3. "Do not dig for given names or genders assigned at birth" 4. "Record gender terms based on information in the not assume [someone] is cisgender or transgender" 5. "Take into account...Is there potential for this information to harm the [person] through outing or violating the right to privacy?" WanderingWanda (talk) 03:27, 8 March 2019 (UTC)


    Any reason we can't use n/a (Q929804) in the gender field? Some instances of humans (namely specific groups like artist collectives) still seem to want P21 as it always comes up as the first suggestion when adding a new property, but for mixed gender groups especially it seems weird to put anything else. And if you are trying to query (for example) artists by gender, there should be some indication that these (currently blank) entries have not merely had the gender property overlooked. Lexid523 (talk) 16:22, 23 January 2019 (UTC)

    New proposal[edit]

    I am proposing the following changes to this property:

    • Allow multiple values so that people's gender identities can be better represented.
      • Rationale: It's possible for someone to identify as intersex AND transgender AND male all at once.
      • Surprisingly, it looks like this change has been implemented already. However, on its own it may cause more harm than good. Several other steps need to be taken (see below.)
    • Add the following instruction to the description: Give precedence to a person's latest self-identification.
      • Rationale: It is not up to us to dictate or guess at someone's gender, and self-identification is how, for example, the RDA library cataloging standard says to catalog gender.
    • Remove the values "transgender female" and "transgender male". Replace with just "transgender". Add "cisgender".
      • Rationale: whether someone is transgender or cisgender is a separate question from whether they are male or female. Making a distinction between "males" and "transgender males", etc., ghettoizes transgender people. It would be like splitting men into the groups "males" and "gay males".
    • Prohibit the values "male" and "female" from being used at once
      • Rationale: If editors are given the ability to do so, unfortunately trans people will often wind up being listed as both "male" and "female" (See: how Chelsea Manning was listed on here a few years ago.) Listing a transgender woman as "male" and "female" is disrespectful, confusing, and does not confer any additional information vs. simply listing her as "transgender" and "female". (For specific cases where using both values at once could be justified, exceptions could be made.)
      • Unfortunately I don't know how technically feasible this is. Could any of Wikidata's constraints be used for this? If not could a custom constraint be created that does?
    • Prohibit "start time" and "end time" from being used.
      • Rationale: We should avoid presenting someone's gender in the form of date ranges. Gender transition is a lifelong, complex, multi-step process and someone coming out as trans is best thought of as a gender reveal rather than an overnight change. Note that if someone comes out as gay we would never indicate that they were straight for a certain date range and then gay for a certain date range.
      • The date someone came out as transgender could be recorded using a property like "significant event" or, perhaps, a newly designed property with a name like "announcement time".

    Conclusion: If we implement all of these changes we can record people's gender-and-sex-related information in a more accurate, nuanced, and respectful way, without losing any information that is currently being recorded (these changes will only add information, not take any away.)

    Mock up:

     – The preceding unsigned comment was added by WanderingWanda (talk • contribs).

    Link to previous discussion on this proposal: Wikidata:Project_chat#Let's_talk_about_gender. @WanderingWanda: Please don't make drastic changes to properties and modify the format of thousands of uses without first getting consensus. I think it would be appropriate to revert the collection of edits until there is consensus in favor of your proposal. --Yair rand (talk) 20:51, 12 March 2019 (UTC)
    Regarding the actual proposal:
    • Ranks, in the normal use, seem quite adequate for handling precedence. Preferred means current data, normal means data that is no longer current, deprecated means data that is now considered to have never been correct.
    • A complex constraint can check for items that have both male and female set: SELECT ?item WHERE { ?item wdt:P21 wd:Q6581097, wd:Q6581072 . }. Many of the 135 items on that list are fictional characters, mythological characters, humans like Tony Briffa (Q15615388) who, according to Wikipedia "chooses to live as both female and male", and many are simply mistakes.
    • Regarding start/end times: I don't know much about this topic, but are there not individuals who say that they were a particular gender until a certain point in time?
    • I don't think it makes sense to separate statements using transgender male (Q2449503)/transgender female (Q1052281) into two statements. Without splitting, the property allows easy access to both assigned-at-birth status and current self-identification. Splitting would make queries more difficult.
    --Yair rand (talk) 21:34, 12 March 2019 (UTC)
    • Please don't make drastic changes to properties and modify the format of thousands of uses without first getting consensus
      Fair enough. Sorry for overstepping. It looks like my changes have already been reverted.
    • are there not individuals who say that they were a particular gender until a certain point in time?
      Maybe in some cases but I think that would be unusual. The Radical Copyeditor's transgender style guide is a good primer.
    • Splitting would make queries more difficult.
      Well, that depends on what you're trying to query, right? What if you want a list of women that includes any woman, without regard for whether they're transgender or cisgender? Or a list of transgender people that includes any trans person, male or female? What about a list of any trans person that also identifies as intersex? Lana and Lilly Wachowski are two of the best known female film directors, and their film The Matrix was recently featured in this article The 100 All-Time Greatest Films Directed by Women. And yet I just did a query for "sex or gender: female" and "occupation: film director". Neither of them showed up because they're not listed as "female" but "transgender female". WanderingWanda (talk) 00:07, 13 March 2019 (UTC)
    Well, that depends on what you're trying to query, right? This point is really just about the making of the right query:
    I know it's possible to query any of the things I mentioned, but it would be easier and less convoluted to query them if transgender/cisgender was separated from male/female. As it is now, like your examples show, if you're doing a search for women, transgender women will be excluded unless you perform an additional step. WanderingWanda (talk) 09:44, 13 March 2019 (UTC)
    • I'm mostly agreed with WanderingWanda, but rather than having multiple values allowed, I think we should be using qualifiers. For instance, "transgender male" is not a gender; it's a gender ("male") with a qualifier ("transgender"). Perhaps it would also be worth splitting the property into sex (for all animals, humans included) and gender (for humans). Then the options for gender are quite simply: "male", "female", "non-binary". And there's qualifiers "cisgender", "transgender" etc. I believe sex would have the options "male", "female", "intersex", "hermaphrodite" (the last for non-humans only, and generally no need to use the sex category for humans). (Please mention my username when replying to this so I get a cross-wiki notification.) Bilorv (talk) 00:23, 18 March 2019 (UTC)
    • I agree that "transgender female"/"transgender male" as a value for property "gender" is shocking and should not be allowed. Rama (talk) 09:31, 7 May 2019 (UTC)
      The property's label is not, in fact, simply "gender". --Yair rand (talk) 18:40, 7 May 2019 (UTC)
    @WanderingWanda: I support all of these proposals except using "transgender" and "cisgender" as gender values. I like Bilorv's idea better of using them as qualifiers, although this would require creating a new property (maybe "transgender status"). Kaldari (talk) 03:10, 11 March 2020 (UTC)

    more transgender combinations or more inclusive descriptions?[edit]

    Valentina.Anitnelav Thierry Caro Shisma (talk) Arlo Barnes (talk) Tsaorin (talk) 16:37, 12 November 2019 (UTC)

    Pictogram voting comment.svg Notified participants of WikiProject Narration Amir (talk) 09:13, 17 June 2014 (UTC) ★ → Airon 90 10:29, 17 June 2014 (UTC) --Another Believer (talk) 15:00, 17 June 2014 (UTC) I am not terribly familiar with Wikidata, but offering my support! Thepwnco (talk) 16:05, 17 June 2014 (UTC) Zellfaze (talk) 16:11, 17 June 2014 (UTC) anjali (talk) 17:31, 17 June 2014 (UTC) Gobōnobō + c 00:25, 18 June 2014 (UTC) OR drohowa (talk) 15:37, 18 June 2014 (UTC) Blue Rasberry (talk) 13:43, 20 June 2014 (UTC) SarahStierch (talk) 16:47, 30 June 2014 (UTC) (Been adding LGBT stuff on Wikidata for months, had no clue this existed!) MRG90 (talk) 10:15, 19 August 2014 (UTC) Ecritures (talk) 16:37, 17 July 2016 (UTC) Erica (ohmyerica) (talk) 19:19, 2 August 2016 (UTC) Shikeishu (talk) 22:28, 17 September 2016 (UTC) OwenBlacker (talk) 20:34, 11 March 2017 (UTC) Ash Crow (talk) John Samuel 17:33, 2 July 2018 (UTC) SilanocSilanoc (talk) 12:18, 17 November 2018 (UTC) Daniel Mietchen (talk) 22:45, 2 February 2019 (UTC) Tdombos (talk) 22:14, 27 May 2019 (UTC) Mardetanha (talk) 17:01, 13 June 2019 (UTC) Theredproject (talk) 23:41, 20 June 2019 (UTC) Davidpar (talk) 20:30, 7 July 2019 (UTC) Gerarus (talk) 13:10, 1 August 2019 (UTC) Sweet kate (talk) 16:33, 4 August 2019 (UTC) Nattes à chat (talk) 22:45, 3 November 2019 (UTC) Lucas Werkmeister (talk) Hiplibrarianship (talk) 23:05, 8 March 2020 (UTC) Jamie7687 (talk) 22:27, 6 May 2020 (UTC) ...

    Pictogram voting comment.svg Notified participants of WikiProject LGBT

    I have a fictional character here that I would say has a gender of transgender female (Q1052281) but the descriptions don't match:

    transgender female (Q1052281): person assigned male gender at birth who self-identifies as female

    Soren (Q64017403) requires a description along the lines of

    person assigned no gender at birth who self-identifies as female

    I think the term transgender female is already correct, but the description should be changed to

    transgender female (Q1052281): person who self-identifies as female but was assigned a different gender at birth

    --Shisma (talk) 09:07, 25 May 2019 (UTC)

    Just to document a similar but slightly different case here (and put it up for discussion): I came along Asgard (Q528820) characters from the Stargate Universe which seem to be asexual as to their reproduction but have a certain gender ("personality") (I don't know this universe, I just followed their descriptions in the according wikia, e.g. [5]]). In this case I added two statements, each qualified with applies to part (P518):
    Heimdall (Q3129443) sex or gender (P21) no value / applies to part (P518) sex (Q290)
    Heimdall (Q3129443) sex or gender (P21) female (Q6581072) / applies to part (P518) gender (Q48277)
    - Valentina.Anitnelav (talk) 10:53, 25 May 2019 (UTC)

    creating new "intersex man" and "intersex woman" items to replace "intersex"?[edit]

    hi, what would be the best thing to do for someone born intersex, notable for being born intersex, who was assigned female at birth and identifies as a woman? i ask for Caster Semenya (Q5834), where i can't just use intersex (Q1097630) and female (Q6581072) together due to the single-value constraint. would creating new items "intersex man" and "intersex woman" be appropriate to include in the one-of constraint (Q21510859)? or is there a more respectful way to do this? --Habst (talk) 02:35, 6 July 2019 (UTC)

    It is by far preferable to permit multiple values, and not create new single values that add other terms together. Trankuility (talk) 00:52, 7 July 2019 (UTC)
    I'd propose to add "begin/end" qualifiers to these values instead. --Infovarius (talk) 15:16, 8 July 2019 (UTC)
    in this example, both qualifiers are simultaneous and they don't have a begin or end date. i am hesitant to just remove the single-value constraint because i understand that it's caused issues with other items in the past. --Habst (talk) 03:08, 9 July 2019 (UTC)

    Split of sex or gender (P21) into "sex" and "gender"[edit]

    Please note the proposal at Wikidata:Property proposal/sex. --- Jura 15:37, 9 December 2019 (UTC)

    Conflicts with country (P17)[edit]

    Why? country (P17) is not to be used on humans. However, this does not really solve the problem why it triggers on Hachikō (Q186486), which is obviously not a human, but also not a thing, so properties like country of citizenship (P27) or country of origin (P495) don't really apply either :/ --Rohieb (talk) 17:37, 16 December 2019 (UTC)

    Wait, isn't that the other way `round? Nations grant citizenship to legal persons (which for most countries is restricted to humans, the exception being called non-human person (Q18170708)), but Hachikō was born, lived, died, and was interred entirely within the borders of Japan. Arlo Barnes (talk) 22:02, 30 December 2019 (UTC)

    Gender is central for forming grammatical sentences and as such needed[edit]

    Wildly boy
    ZI Jony
    Ivan A. Krestinin
    Jonathan Groß
    Pablo Busatto
    Thierry Caro
    Tinker Bell

    Pictogram voting comment.svg Notified participants of WikiProject Properties

    I just removed the property that may violate privacy (Q44601380) statement from this property as I believe it to be harmful and write this post according to our policy. Nonenglish languages such as German need a statement about a persons gender to make statement about the person. Given the necessity for assuming that a person has a gender to make grammatical sentences about them people frequently have to use heuristics based on the name of the person. Any Wikidata statement that allows for a better inference then the name is useful for data users and will allow them to misgender less people. ChristianKl❫ 15:09, 22 February 2020 (UTC)

    • Is it really worth it? It seems to be me that there is a debate about 1/1000 or less of the entries and skipping them or accommodating them differently would largely avoid drama. --- Jura 15:13, 22 February 2020 (UTC)
      • Although few people will consider their gender as private information, the current situation is only 88% of people in Wikidata have gender information. Many other people (e.g. authors of papers) are only known from few sources that does not explicitly describe their gender (in some cases, there're no gender information available from their names).
        Using singular they may be an idea, but in Wikipedia there's no consensus for or against it, and in some languages there's no equivalent.--GZWDer (talk) 15:26, 22 February 2020 (UTC)
        • English has a singular they that can be used but plenty of languages such as German don't have a singular third person pronoun. In many languages gender also has a more far-reaching effect then just affecting pronouns.
    @Jura1: I don't think you need property that may violate privacy (Q44601380) to make expections for a handful of items that are subject to controversy. Using it that way also weakens the general understanding about what the protection means. Giving sex or gender (P21) the same protection status that phone number (P1329) has, is a way to make people understand that they can use phone number (P1329) much more widely then was intended. ChristianKl❫ 20:31, 23 February 2020 (UTC)
    • Not sure which problem should be solved here. Gender deductions from given names etc. are pretty common and widely accepted currently—unless someone complains about such a claim, which is rather rare. In case of complaints, we fall back to the requirement of references to serious sources, which often not exist. Do you really want to enforce gender statements based on deductions in these cases? It is pretty likely that these are either plain wrong, or someone is severely pissed about this statement. —MisterSynergy (talk) 16:00, 22 February 2020 (UTC)
    • In my opinion, both sex or gender (P21) and sexual orientation (P91) should have property that may violate privacy (Q44601380) and should not be assumed (without any statement clearly made by the persons themselves). Some persons may consider both these values as private information. Irrespective of the percentage (current higher percentage value may be related to assumptions), I still consider it private. and I feel that Like sexual orientation (P91), sex or gender (P21) should also have an additional constraint citation needed constraint (Q54554025). In my edits related to persons, I normally use preferred pronoun (P6553) where the reference article describes the person with the specified pronoun and where sex or gender (P21) information is unavailable. John Samuel (talk) 16:18, 22 February 2020 (UTC)
      • Your position is essentially that nobody should speak in languages that require the speaker to assume gender about certain people. That seems to me a very extreme position. ChristianKl❫ 20:31, 23 February 2020 (UTC)
    • The presence or absence of this property does not seem to me to be an obstacle to correct phrasing. Having said that, who would be offended with a wrong gender attribution? It is often a misinterpretation or vandalism. If someone asks us for a change, the problem is no longer private. I understand its presence in phone number (P1329). As well described in the Statements of this property, it is to replace it with property likely to be challenged (Q44597997). In summary, Symbol support vote.svg Support deletion because there is minimal interest in its presence in this property and not for the reason mentioned by ChristianKl. If a genre is incorrect, the repercussions will remain false although Q44601380 is absent. An explanation on this subject which escapes me… —Eihel (talk) 17:41, 22 February 2020 (UTC)
      I think you might have missed the discussions further up on this page and the new property proposal --- Jura 18:19, 22 February 2020 (UTC)
      I did not miss the proposal for the split. Whether it is for sex or for gender, what I write is identical. But I admit that I await the comment from ChristianKl. —Eihel (talk) 23:34, 22 February 2020 (UTC)
    • Does Wikidata potentially violate privacy by using sex or gender (P21) for an individual? To do so it would need to be making a claim that "Individual A has Gender B". However, Wikidata does not do this. Wikidata says that "Reference C claims that Individual A has Gender B". This is hearsay. Claiming that "Reference C claims that Individual A has Gender B" is very different from claiming "Individual A has Gender B". Even if a statement is not sourced, it is still only claiming that "Someone claims that Individual A has Gender B", not that Wikidata is stating that it is true. As we cannot conduct or include original research within the scope of this project it is impossible for us to directly claim that "Individual A has Gender B", and thus since we are not making a claim as to their actual gender, we cannot therefore be exposing their private details as to their gender. We only make a statement that a source has claimed a gender for the individual. That said, misuse of this property, such as using it to attach original research, may expose private information, just as with several types of properties, but those misuses should be deleted as original research. Legitimate sourced statements are not a violation of privacy, as they are merely commentary about a released statement in another source. Josh Baumgartner (talk) 00:59, 24 February 2020 (UTC)
    • The idea of the concept of gender is that it's about the role of how a person interacts with society. If a person publically interacts with society it usually can't do this in a way that's detached from gender. While it might be possible in English to leave any assumptions about someone's gender
    In French you need gender to decide on the proper form of a lot of adjectives. If Siri wants to make grammatical correct sentences in French having access to our gender information about a person is important. You could argue that Siri should simply assume people to be male if it doesn't know or that the French shouldn't have access to a lot of information via Siri that English speaking people can access via Siri. The policy of assuming male gender seems to go against ideals of the gender discourse. The policy of witholding the information from the French people because you don't like inherent features of their languages seems to me very discrimitory and against Wikidata's goal of inclusivity whereby inclusivity is not just about being inclusive towards all strains of the Anglo discourse. ChristianKl❫ 19:39, 26 February 2020 (UTC)
    Genuine question @ChristianKl:: is it really the case that downstream, commercial uses of wikidata (siri etc) dictate design decisions impacting how data on living people are modelled in wikidata itself? Furthermore, that inclusivity should be judged by the customer base of these downstream (commercial) projects? That is certainly not how the wikidata project is presenting itself to the scientific / academic community in encouraging them to support initiatives such as the inclusion of ORCID sourced profile data, and moving management of staff lists to wikidata - expanding the number of items representing living people (but without corresponding wikipedia entries) and therefore making privacy issues more prominent. 11:38, 27 February 2020 (UTC)
    I don't think anything about the argument I made is special to commercial downstream projects, it's just as applicable to non-commerical downstream projects.
    As far as the position towards the academic community goes, the academic community goes VIAF that gather the data from all the libraries together also lists gender and doesn't treat gender as being a property that deserves the kind of privacy protection as the address of a person which they don't include. Academically, use-cases such as being able to make statement about the gender distrubtion of the authors in a given journal is also valuable. The position that counting gender distrubtion of a journal is immoral because you need to assume gender to do so, is shared by very few people. The importance of knowing about gender gaps is often mentioned in WMF material. ChristianKl❫ 18:46, 27 February 2020 (UTC)
    @Joshbaumgartner: Wikidata may be violating privacy for some people. Address is used as an example BLP - this may be found in a WHOIS entry (ie sourced) but user may flag privacy violation if added to wikidata. For some people P21 is private information and should be treated similarly. Just because something can be added does not mean that it should. Centralisation of previously scattered data can result in a violation of privacy. 11:38, 27 February 2020 (UTC)
    I agree with the idea that not everything that can be added should be added, and would say that just because something is sourced doesn't mean it is necessarily appropriate to include here. However, if a piece of information is published, the published information is by definition not private, and thus stating the published information does not reveal private information. If the Guardian publishes that Joe is a spy, that may be revealing private information, but a Wikipedia statement (Joe > occupation > spy > stated in > Guardian) would only be revealing the published statement made by the Guardian. Whether or not the statement should be added to WD is one thing, but it is not a privacy violation. As for centralization, Wiki projects are not the center of the universe, and adding sourced statements is replication of published information. Characterizing it as 'centralization of previously scattered data' is one way to put it, but that kind of view is subjective. Josh Baumgartner (talk) 19:45, 27 February 2020 (UTC)

    Constraints on citation[edit]

    I have now added a constraint on citation. We need to add sources to the gender values. Reliance on, e.g., given name may cause problems sometimes. There may be privacy issues that we need to adhere to. It is unclear to me what a source needs to be. For instance, would an article where the person is coreferenced with a pronoun be a sufficient source? — Finn Årup Nielsen (fnielsen) (talk) 01:45, 23 February 2020 (UTC)

    @Fnielsen: sounds like lot of work for very little gain in the overwhelming majority of cases. If there are privacy issues, Wikidata:Living people says how to deal with them appropriately. [[User:Finnusertop|phone number (P1329)]] (talk) 20:35, 23 February 2020 (UTC)
    I don't see how gender is in special need of sourcing for all people in a way that isn't important for other property. ChristianKl❫ 20:44, 23 February 2020 (UTC)
    While it may be the case for the majority, there might be issues for 1%. We have over 6 million humans in the Wikidata. If there are issues for 1% then it is an issue for 60,000 people. — Finn Årup Nielsen (fnielsen) (talk) 23:19, 25 February 2020 (UTC)
    I would expect the error rate to be under 1% but I do agree that there will be wrong values. However if I wrongly put down "female" on the basis of someone's name being "Lisa" I'm not violating privacy given that I didn't add the statement based on private information.
    Such an error leads to wrong gendering if a data-user needs the gender to form a gramatically correct sentence or they want to make gender gap statistics but it's easier to fix that error when gender is noted explicitely on Wikidata then if the data-user implictely guesses the gender. ChristianKl❫ 09:01, 28 February 2020 (UTC)

    Should male organism (Q44148) be used on items of anthropomorphic character (Q27921916)?[edit]

    What do you others think? --Trade (talk) 19:00, 20 March 2020 (UTC)

    I am came here about the same question. I see previous discussion here. I try to make some statistics of which values used for non-human characters (I've filtered out Ancient greek characters as they are mostly non-marked humans, and different deities):
    SELECT ?sex ?sexLabel ?cnt ?itemSample ?itemSampleLabel WHERE { 
      SELECT ?sex (COUNT(?item) AS ?cnt) (SAMPLE(?item) AS ?itemSample) WHERE {
        ?item wdt:P31/wdt:P279* wd:Q95074.
        MINUS {?item wdt:P31/wdt:P279* wd:Q15632617.}
        MINUS {?item wdt:P31 wd:Q22988604.}
        MINUS {?item wdt:P31/wdt:P279* wd:Q178885.}
        ?item wdt:P21 ?sex.
      GROUP BY ?sex ?sexLabel
    SERVICE wikibase:label { bd:serviceParam wikibase:language "[AUTO_LANGUAGE],en". }
    ORDER BY DESC(?cnt)

    Try it!

    I agree with Valentina that we should use "human" sex for antropomorphic characters and "animal" sex for gendered objects. But the boundary is vague. What about Disney ducks? puppets? robots (hey, Bender)? --Infovarius (talk) 12:46, 23 March 2020 (UTC)

    On a related note: Wikidata_talk:WikiProject_LGBT#Q1097630_used_on_various_robots_(or_non_human)_of_fiction. --Misc (talk) 16:18, 23 March 2020 (UTC)
    I think of male (Q6581097)/female (Q6581072) as representing human gender/sex including ideas of gender identity and roles (not only reproduction) and female organism (Q43445)/male organism (Q44148) as representing biological sex as to reproduction (without connotations of gender identity or role). Due to this perception I tend to use for all anthropomorphic characters (also robots who don't reproduce sexually but have some kind of gender identity) male (Q6581097)/female (Q6581072) and for all non-anthropomorphic characters with sexual reproduction female organism (Q43445)/male organism (Q44148). So Goofy (Q111135) would be male (Q6581097) and Pluto (Q108732) male organism (Q44148). - Valentina.Anitnelav (talk) 09:05, 24 March 2020 (UTC)
    • I disagree with this. It is bear, not a man. Maybe in English you don't see the difference, but in russian "мужчина" (male man) and "самец" (male beast) are completely different. Monkeys are also anthropomorphic. Should we also write them as "men"? No. ·Ɔ (talk) 11:36, 30 May 2020 (UTC)
      • Hm. But you don't see any problem with male organism (Q44148)? Winnie the Pooh (Q188574) is a toy, meaning strictly speaking he does not have any sex (Q290) at all, he is just referred to as male. - Valentina.Anitnelav (talk) 16:20, 30 May 2020 (UTC)
        • You can't know this from the perspective of a fictional world. In a sense, his gender is just a role, but if the author would write that he had found a female-bear and they would have kids, it would be clear that they had mated. Considering that at all werewolves it seems to be customary to indicate the sex of the animal - why should a fictional bear have to do otherwise? Of course, it would be nice to separate the invented gender from the real one, but I suspect such genders will be used to troll transgender people. I did not notice that the lexemes have indicated gender, although in German, for example, it is very important what kind of "gender" a word is in order to correctly use it.·Ɔ (talk) 11:49, 31 May 2020 (UTC)
          • The English label reads "male organism". I'm a bit doubtful if it would be appropriate to call a teddy bear (a stuffed toy) an organism. Even if they would have kids it would not be clear that they had mated. Who knows how teddy bears would reproduce in this fictional universe unless it is made explicit. Maybe they sew a new teddy bear together with a bit of their filling. The only thing we know is his gender (as presented in the book series). - Valentina.Anitnelav (talk) 15:51, 31 May 2020 (UTC)
      • I wonder if it would be best to create own items referring only to gender, absolutely independent of any reproductional aspect for these cases of anthropomorphic objects and maybe animals, too, if there may be problems with male (Q6581097)/female (Q6581072) in some languages as they strictly refer to humans. - Valentina.Anitnelav (talk) 16:30, 30 May 2020 (UTC)