dplace-data
dplace-data copied to clipboard
D-PLACE
wrong population size (EA202) for Koreans (Ed1)
https://d-place.org/society/Ed1
the value for this variable is 200, based on the population of a specific island in 1947. Surely this should be modified (it was more around 20 million around that time...)
Similar issue for Brazilians (Cf4), population size (EA202) equals 2700 presumably since this is the population of a village near Sao Paulo in the ethnographic source. Clearly should be larger (also, EA031 is 50000+).
There appear to be more with a similar problem - here are more ea_id's for which EA031 is 50000+ but population size is smaller:
Ca7
Ch1
Ch5
Ea13
Ed6
Ef11
Eg10
Thanks Dor, these look like something we should check in more detail. I'll add it to the to-do list
I looked in the EA to check these values.
According to the Murdock (1967). Ethnographic atlas: a summary, the population sizes are: Koreans (Ed1): 30,000,000 Amhara (Ca7): 2,000,000 Bulgarians (Ch5): 7,000,000
Other population data I found: Brazilians (Cf4): 2,700 in Cruz das Almas in 1940. SOURCE: Ethnology 1963, Vol. 2, No. 1 Uttar Pradesh (Ef11): 1,400 in 1941 (the village of Madhopur or Senapur). SOURCE: Ethnology 1971, Vol. 10, No. 1.
For Serbs (Ch1), Punjabi (Ea13), Min Chinese (Ed6) and Telugu (Eg10), I found no population data in EA.
@SimonGreenhill Do you know what the (other) sources were used for this variable?
I suspect the number recorded here is for this subcase i.e. Kanghwa Island. Perhaps we should think about renaming these cases to e.g. Korean (Kanghwa Island) to be more transparent.
I'm not sure that updating this value to the modern population is the right thing to do (because where do we stop - do we recode other variables that are different?).
I looked at the description of EA202 variable and this is probably not a bug, but a feature:
"Population of ethnic group as a whole, unless otherwise noted in Comments. Note that source differs by society; EA bibliography is source where possible, otherwise Ember (1992)." https://d-place.org/parameters/EA202#1/30/152
So I'd leave it as it is. As long as the culture in question has a disclaimer that the values relate to a specific location.
What could theoretically be done is to split this variable into two: "Population of ethnic group" and "Population of subcase". Each would have a bunch of NAs, but it wouldn't be as confusing.
I'd even go so far as saying that confusion is a feature and not a bug here. It basically signals that one needs to do a bit of research before using these numbers in any analysis. Prefering the smaller numbers also seems to be useful, because extrapolating D-PLACE data to - essentially - the level of big national languages is burdened with all sorts of theoretical problems and getting the speaker numbers for these is probably the smallest issue here :)
I'd recommend signaling these exceptions more clearly to researchers. Generally speaking, I don't think it's a good idea for one variable to refer to two things, and one can't expect all researchers to review each datapoint.
I agree that a variable being interpreted differently for different datapoints is generally bad. But since d-place only aggregates data from existing sources, the question becomes whether this is something the editors can/should change or whether we simply document this issue in the variable description.
dorshilton @.***> schrieb am Mi., 22. Nov. 2023, 10:34:
I'd recommend signaling these exceptions more clearly to researchers. Generally speaking, I don't think it's a good idea for one variable to refer to two things, and one can't expect all researchers to review each datapoint.
— Reply to this email directly, view it on GitHub https://github.com/D-PLACE/dplace-data/issues/321#issuecomment-1822410292, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAGUOKEBAKOO7BIZPFPO353YFXBLJAVCNFSM6AAAAAARUB44U6VHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQMRSGQYTAMRZGI . You are receiving this because you commented.Message ID: @.***>