[ 3 / biz / cgl / ck / diy / fa / ic / jp / lit / sci / vr / vt ] [ index / top / reports ] [ become a patron ] [ status ]
2023-11: Warosu is now out of extended maintenance.

/sci/ - Science & Math


View post   

File: 156 KB, 790x910, statistics.png [View same] [iqdb] [saucenao] [google]
11157587 No.11157587 [Reply] [Original]

Pic related is the results from a UK Fire Service whole-time recruitment campaign from 2017.
The service covers two counties, population is divided up 96.4% white british/irish, 3.66% black/minority ethnic (BME)
Assuming advertisements for the campaign reached all members of the population equally and the selection process was carried out "blind" i.e. stripped of all gender/ethnicity/age data - do these results make sense? If not, why not?
Thanks in advance :)

>> No.11157590

>>11157587
*96.34%

>> No.11157604

You're asking if there is any racial bias?
Seems unlikely, the deviation is small.
If you wanna insist on being specific, just run a qualitative chi-square.

>> No.11157617

>>11157604
Any kind of bias at all really
BME 3.66% of population but 6.54% of applications
Nearly twice as likely to apply as white?
Intuitively, should be skewed the other way as historically BME are LESS likely to apply and also there would have been disproportionate applications from part-time firefighters compared to general public (service is 98% white)
?
looking up qualitative chi-square now
tks

>> No.11157632

>>11157587
Women were about 10.4% of the applicants yet were 14.8% of hires.
Was everybody applying for the same position or were women more likely to apply for some telephone/desk job?

>> No.11157644

>>11157632
not a desk job - role is whole-time firefighter

>> No.11157654

>>11157587
Total applicants by gender graph is fucked; baseline starts at 800 instead of 0 like all the rest.

>> No.11157656

>>11157604
>the deviation is small
how did you get this? Sorry - not a stats fag

>> No.11157662

>>11157654
like every other public body in UK they're pissing themselves in case they're accused of being racist/sexist/transphobic etc etc etc

>> No.11157725

>>11157662
That's what I don't get, because that error in data presentation is making them actively look sexist. It'd be far more egalitarian if it were presented properly.

>> No.11157729

>>11157587
Fuck off, go back to /pol/

>> No.11157765

>>11157729
dumbass nibba

>> No.11157778

>>11157617
>nearly twice as likely to apply
Yeah, rates of unemployment are usually higher among non-whites in white countries.
How's the UK's economy nowadays again? I recall there being some Brexit problems still in 2017.
>disproportionate applications from part-time firefighters compared to general public
That's entirely expected.
>looking up
Did you find it?
>>11157656
When a statistician says "statistically insignificant" he means "I've run some statistical tests for 95% and it failed a bunch".
Small means "I've glanced at the data and it's probably nothing."

>> No.11157829

>>11157778
>rates of unemployment are usually higher among non-whites
accepted, but rates of unemployment are high all over this area
as for the applications from part-time firefighters, the area is mostly rural and so most of this service is part-time (we call it "retained" in the UK)
we all knew the service was recruiting whole-time (for the first time in 10 years) and lots of us applied, as the service is currently 98% white this should have skewed the BME applications lower than the population percentage of 3.66%?
>qualitative chi-square
yeah watched a video but there's only two categories so the residual of one is the negative of the other and zero squared is zero
don't really see how it applies to this situation?

>> No.11157842

>>11157829
>the residual of one is the negative of the other and zero squared is zero
Look at the formula again.

>> No.11157871

>>11157829
I guess it's the Jews then good buddy. You thought of everything and figured it out via process of elimination.

>> No.11157889

>>11157842
ok - my bad
two categories - white british/irish and BME (ignoring prefer not to say)
expected is 96.34% white / 3.66% BME as per population and assuming equal exposure to advertising and "blind" selection of applications
observed is 93.46% white and 6.54% BME
residual for white is -2.88 and 2.88 for BME
residual squared /expected = 2.23 for BME and 0.12 for white
sum of these is 2.45? what do I do with this?

>> No.11157899

>>11157871
really helpful, thanks
this is why I'm asking:
https://metro.co.uk/2019/01/28/fire-service-changes-entrance-test-deliberately-harder-white-men-8402291/
that's not my service btw but suspect it was a thing in this process also
no-one wants it
no-one benefits
lots of us very fucked off with whole thing

>> No.11157909

>>11157889
Doesn't it explain on the page you got the formula?
Pick a percentage of significance and look up the number in a table.

>> No.11157951

>>11157909
Df = 1
chi-square = 2.45
p-value = 0.1 approx
not sure what you mean by a % of significance?

>> No.11157959

jesus those plots are HIDEOUS

>> No.11157962

>>11157909
10% probability that null hypothesis is correct?
so statistically significant?

>> No.11157999

bump and help!

>> No.11158021 [DELETED] 

>>11157871
Gas yourself, kike

>> No.11158065

anybody?

>> No.11158215

>>11158065
They are giving more job offers to minorities and women to try to promote diversity. You're not arguing science, you're arguing policy. Please take it to the appropriate board.

>> No.11158240

>>11158215
>You're not arguing science, you're arguing policy
I'm not arguing at all - I'm asking for help to interpret results in pic in first post

>> No.11158265

>>11158240
The interpretation is that minorities and women have a higher acceptance rate due to policies. I think you knew that to begin with.