[Corpora-List] Ability and Personality Traits from Text: 1st Call for Participation

Nihal Yağmur Aydın nyagmuraydin at gmail.com
Sat Dec 21 08:18:27 CET 2019


Hello everyone,

As I studied psychology in addition to my studies in computer engineering (bachelor's and master's level), I would like to comment on this issue, as I noticed a probable incorrect implementation of psychology in computer science domain or in any selection process published as a shared task within this community.

1- The quote mentioned by Ms. Rehbein previously about the psychometric test, which is the association between "who is the main person in the image " and " how the person is feeling" seems to be not a good measure, as it brings an ambiguity, with the involvement of emotions in the topic.

It could be OK to ask someone about an image and who is the main person there, in order to measure someone's intellectual capacity (to measure if the person is aware of the past public figure in Germany for instance), however, if the image is not very clear and recognizible by many others in a similar manner, and if someone is asked about it, it brings the issue of " selective perception" or one's "subconscious mind". Free association tests give some clues about someone's subconcious mind, but if the image is that much free, I doubt that it is not a measure at all! With free association or selective attention/perception you see how many people differ in perception etc. but you cannot generalize any result based on this. Secondly, use of emotions is something I never observed in any kind of test styles so far, and it brings a highly personal perspective, which is hard to measure. Moreover, personalization level should not be that much high in psychological tests used for analysis purposes. Based on these realisations, the test does not look very promising.

2- As mentioned by the publishers at the public link, it is true that socio-economic backgrounds of participants might affect selection processes in Germany. However, there is also another issue, which I learnt by reading some articles that some people are put into disabled categories in Germany (who are mostly immigrants) and sent to disabled's schools, whereas when these people go to their home country later on, psychologists or health professionals don't agree with the previous findings. Because of that, some foreign psychological health and well being centers are being opened in Germany, in order to help people from other countries tackle with such incorrect measures. Therefore, I guess that psychological measures done at schools for selection processes cannot be generalized, as someone's psychology is highly shaped by his/her cultural background and it is mostly not understood by national psychologists which cause them to arrive at wrong generalizations. Freud's psychological theories for instance, are highly criticized as it is done with Austrians mostly and not generalizible to every nation. All in all, considering that Germany hosts a large number of immigrants, any psychological test done for selection of students seems to be quite biased and not generalizible. *** Also, general and standardized tests, like SAT, GRE etc. are so much different than the test proposed in that study. As I entered GRE for my master's study in Turkey, (due to the fact that this exam is recognized also in some universities of Turkey and have equivalence to national exam), we all know that it is a world-wide recognized test and there is not any personal measure. Therefore, the comparisons which were made about standard tests and psychometric tests proposed don't reflect a good comparison method. Cover letters are common everywhere too, and it signifies one's own personal achievements so it is not the same in structure when compared to the test proposed.

3- Referring to the past discussions, I would also state that some problems in education system exist, as many people don't become open to criticisms even if they become scientists later on. You might be more open to discussions and open talk, instead of giving aggressive reactions. One's own social status might also be studied based on how they react to "criticisms", therefore, being scientist somewhere is also not a good indicator of "status", I believe.

Best Regards, Nihal.

6 Aralık 2019 Cuma tarihinde Arne Köhn <koehn at informatik.uni-hamburg.de> yazdı:
> Ines Rehbein writes:
>
>> For students who want to apply to NORDAKADEMIE, there is no option to
>> not take the test (as it's part of the application procedure) or, to my
>> best knowledge, to refuse the use of their personal test data for
>> research and other use cases.
>
> One of the organizers addressed this on twitter:
>
> ,----[ https://twitter.com/realTimoBaumann/status/1202844419029970944 ]
> | ppl are asked for consent on whether the data can be used for research |
> | and ppl can opt out. they cannot, however, opt out from the testing as |
> | it's part of the application process. try applying at a US university |
> | without SAT and TOEFL/IELTS scores, or without a cover letter. |
> `----
>
> There is also this public statement; the link was posted as a reply on
> twitter but not here as far as I can tell:
>
>
https://www.inf.uni-hamburg.de/en/inst/ab/lt/resources/data/germeval-2020-psychopred/germvval2020-task1-public-statement.txt
>
> Best
> Arne
>
> P.S. for the sake of transparency: I am sending this from my old Uni
> Hamburg mail address because I am still subscribed to corpora with
> this address; I left for Saarland University a year ago. I know
> some of the organizers of the ST but am not involved in it.
>
> _______________________________________________
> UNSUBSCRIBE from this page: http://mailman.uib.no/options/corpora
> Corpora mailing list
> Corpora at uib.no
> https://mailman.uib.no/listinfo/corpora
>

--

Best Regards, Nihal Yağmur Aydın -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: text/html Size: 6919 bytes Desc: not available URL: <https://mailman.uib.no/public/corpora/attachments/20191221/5b69ecdd/attachment.txt>



More information about the Corpora mailing list