On 29-1-2010 20:46, Warp wrote:
> Neeum Zawan <m.n### [at] ieeeorg> wrote:
>>
http://www.nytimes.com/2010/01/14/nyregion/14fire.html?adxnnl=1&adxnnlx=1264791670-GxVD+sfKjvw04Ji/zop/fQ
> 
>>         It fails to explain how exactly the test discriminated. I looked at
>> some other articles, and none gives any detail whatsoever.
> 
>   It's puzzling indeed. I'm trying to think how you could create an exam
> which would be disadvantageous to black people, and I can't think of anything.
> How could one possibly make a test disadvantageous to people with a certain
> skin tone? I just can't fathom it.

It is not the skin tome, it is the cultural background. (almost) all IQ 
test favour white Europeans, because they refer in subtle ways to things 
that appear to be common knowledge by the people who design these tests. 
They also put an emphasis on things that we as Europeans (or is 
Caucasian the term mostly used in the US?) find important. Sometimes 
that is valid and sometimes it is not.

If a test uses a phrase that became well known because of being used in 
a 60's sitcom, people with a social background that included a TV in the 
60's will have an advantage. Use one from a 90's black sitcom and 
another group is a at an advantage.

One of the things that my sister mentions in her book on cultural 
background in the job market is that it appeared that a disproportionate 
number of Dutch citizens from a non-Dutch background failed the written 
test for nurse. The reason turned out to be that in the questions a lot 
of expression were used that were derived from our famous 17th century 
bible translation. Even if the tested persons were familiar with the 
bible, they too often did not know our particular translations. One 
could argue that that would indeed imply that they would not be able to 
communicate effectively with the rest of the staff and the patients. One 
could also argue that these expression seldom turn up in spoken 
communications, especially if the one you are talking to is easily 
identified as probably not a 5th generation pure bread Dutch. Also note 
that the reciprocal problem is never mentioned. The persons that made 
the test might not be able to communicate effectively with a subgroup of 
patients.

Creating a good set of questions is a skill. Most that are used today 
fail to meet the principle that they only test the relevant skills. The 
test that I gave today to my students is an example of such a biased 
test. I know it is, I can identify (some of) the questions that are 
contaminated by cultural bias. I know many people would not even notice, 
but I did and I didn't find a way to solve it.
So, I know that even a test on digital technology can have a cultural 
bias. Therefore it wouldn't surprise me if a test for fireman would 
also. You would have to see the actual test to see if that bias is 
justified or not.

Does this answer your question?

Post a reply to this message