21.11.2022 Views

Theory of Knowledge - Course Companion for Students Marija Uzunova Dang Arvin Singh Uzunov Dang

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

IV. Ethics

old adag, widly itd and oftn misattriutd

An

Mark Twain, is “Thr ar thr kinds of lis:

to

damnd lis, and statistis”. This stion taks

lis,

losr look at th impliations of knowldg

a

drdging is on xampl of a statistial

Data

with murky this. It involvs th

prati

sanning of larg data sts for

spulativ

without having a prdfind

orrlations,

Sin a 95% onfidn intrval is

hypothsis.

usd, maning a 5% han of finding

usually

orrlation twn ompltly random

a

spurious yt statistially signifiant

varials,

ar almost rtainly found. On ovious

rsults

is that any hypothsis found using data

solution

must tstd against anothr data st

drdging

usd in th original drdging.

not

w may not s any rason to qustion

Oftn,

and statistial vidn. Most of th

orrlations

as non-xprts, w may not hav th skills

tim,

dig dpr, and must instad valuat th

to

of xprts to prform in th funtion

rdiility

puli intlltuals.

of

paradox is a phnomnon in

Simpson’s

and statistis in whih a trnd that

proaility

in svral diffrnt data groups rvrss or

xists

whn th groups ar omind, du

disappars

a “lurking” varial (also alld a onfounding

to

that is not aountd for. This paradox

varial)

how orrlations an distortd y

rvals

rrors, and th dangr in making

sampling

th xampl shown in Tal 11.1:

Considr

Univrsity of California, Brkly 1973

th

11.1 shows that mn appar mor likly

Tal

womn to admittd to th univrsity,

than

an admissions ias. Howvr, on

indiating

admissions rats y dpartmnts,

analysing

trnd rvrss, and it appars that mor

th

(A, B, D and E: s Tal 11.2)

dpartmnts

a ias in favour of admitting womn. Th

had

and orrtd data showd a “small ut

poold

signifiant ias in favor of womn”

statistially

et al 1975).

(Bikl

B 860 63% 25 68%

D 417 33% 375 35%

E 191 28% 393 24%

F 373 6% 341 7%

paradox ariss from th fat that mn

Th

to dpartmnts suh as nginring

applid

hmistry that had fwr appliants

and

highr admission rats, whras womn

and

to dpartmnts with a gratr numr

applid

appliants and a lowr rat of admission. Of

of

th laim of a statistially signifiant

ours,

in favour of womn dos not nssarily

ias

thr was a pratial ias in favour of

man

womn.

paradox itslf is xplaind in th TED-Ed

Th

linkd hr.

vido

trms: Liddll How

Sarh

an mislading

statistis

11

IV.2 The trouble with statistics

admissions statistis.

Women

Men

Applicants Admitted Applicants Admitted

Total 8442 44% 4321 35%

drivd from statistis and th authority it arris.

Table 11.1 Univrsity of California, Brkly: admissions 1973

(Bikl t al 1975)

Women

Men

Department

Applicants Admitted Applicants Admitted

A 825 66% 108 82%

C 325 37% 593 34%

Table 11.2 Dat for dirnt dpartmnts (Bikl et al 1975)

ausal intrprtations of orrlations.

350

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!