Difference between "cases" (C) and "persons" (P)
Closed this issue · 9 comments
What is the difference between "cases" (C) and "persons" (P) in the datasets on asylum decisions and asylum applications? Our guess is that cases include several people (e.g. families). But, if yes, is there any way to tell how many people it includes?
Cases can consists of several people. Cases and persons are two different ways of counting.
@erictapen Gibt es über die API einen Weg, einfach abzufragen, wie viele Länder in Europa mit "Cases" arbeiten (statt Personen)? Diese Unterscheidung könnte nämlich womöglich noch ein Problem darstellen, wenn viele Länder damit arbeiten, könnte sich aber auch erübrigen, wenn sie in Europa kaum vorkommt. Über den UNHCR Data Finder kann man das aber nur schwer herausfinden, also dachten wir, dass es über die API vielleicht einen einfachen Weg gibt.
Oh das Thema hatte ich schon komplett vergessen! Ich hab mal schnell das Attribut dec_pc in die Webapp integriert, das müsste die Unterscheidung in der API sichtbar machen. dec_pc ist an sich undokumentiert, aber ich gehe mal davon aus, dass es die Unterscheidung vornimmt, zumindest sind die einzigen möglichen Werte P oder C.
D.h. ab jetzt sieht man auf dem Prototypen, was das für Zahlen sind.
Apparently this even can switch inbetween years for the same COO/COA combination...
See e.g. AFG -> FIN
So there really is the question how to interpret these values, even if we limit the selectable COA to european countries.
I think it is a good way (at least for now) to make the ways of counting transparent as you did it here @erictapen. There are some years and countries where the data is both based on persons and cases, how is the webapp dealing with this right now? summing them up?
There are some years and countries where the data is both based on persons and cases, how is the webapp dealing with this right now? summing them up?
Yes, currently they are displayed like this:
I also think it's good to be at least transparent with it. But imo the biggest problem here is that, if a Case can mean more than one person, that our units don't check out anymore. We're then comparing cases with the population of a country (which itself would be not too bad) but also in the worst case we would compare cases/population in one COA with persons/population in another COA.
Currently I'd lean into the direction of removing Cases and Mixed based data alltogether to not run into this pitfall.
What is the difference between "cases" (C) and "persons" (P) in the datasets on asylum decisions and asylum applications? Our guess is that cases include several people (e.g. families). But, if yes, is there any way to tell how many people it includes?
-->answer by Léo:
--Confirmed. All asylum claims in almost all countries are processed in terms of cases – e.g. a family applying for asylum together would likely be processed as one case. Because of legacy administrative systems, some countries are unable to publish the actual number of people included in these cases. Over time, this has reduced to now just two or three countries and only for specific types of applications (e.g. in the UK it is only those on appeal I think). The big recent improvement was from the US in the mid-2021 statistics where they were able for the first time to provide statistics for the number of people relating to all asylum applications. Previously this was only available from one of the two departments responsible for processing asylum cases.
But, if yes, is there any way to tell how many people it includes?
-->answer by Léo:
-- For the legacy US data, yes we publish the average in Global Trends each year. For the UK, Finland and Malta, this is not available, but the numbers are much smaller so we simply have to typically present the cases as the number of people. Bear in mind for the US, that this is the overall average so we would not recommend applying this to specific countries of origin as the average case size can vary substantially. We only use it to estimate the total number of asylum applications globally (for which the US data is a significant contribution).
I also think it's good to be at least transparent with it. But imo the biggest problem here is that, if a Case can mean more than one person, that our units don't check out anymore. We're then comparing cases with the population of a country (which itself would be not too bad) but also in the worst case we would compare cases/population in one COA with persons/population in another COA.
Currently I'd lean into the direction of removing Cases and Mixed based data alltogether to not run into this pitfall.
This will not solve the problem as generally one decision does not mean one person, even if we remove the cases data. As one person can receive decisions over several applications in the same year. Definitely this leads us to the problem if it is misleading to use the number of inhabitants, but I guess now it is too late :/ :D