Professional Notes

• Date: 01 September 1981
• Published date: 01 September 1981
• DOI: 10.1177/0734371X8100200108
• Author: Gail F. Jones
• Subject Matter: Articles

85

PROFESSIONAL

NOTES

USEFULNESS

OF DIFFERENT

STATISTICAL

TECHNIQUES

FOR

DETERMINING

ADVERSE

IMPACT

IN

SMALL

JURISDICTIONS

Gail

F.

Jones

Examination

Technician

Grand

Rapids,

Michigan

A

major

problem

faced

by

small

jurisdictions

is

how

to

determine

adverse

impact

in

regards

to

their

employee

selection

test.

The

Equal

Employment

Op-

portunity

Guidelines

(EEOC)

do

not

give

specific

enough

information

in

regard

to

such

tests

to

be

of

practical

help

in

determining

adverse

impact

when

dealing

with

small

numbers.

Small

jurisdictions

seldom

have

the

number

of

applicants

necessary

for

good

statistical

analysis

and

it

is

difficult

to

use

the

&dquo;4/5

rule&dquo;

(U.G.E.S.P.,

1978)

when

it

is

necessary

to

discuss

4/5

of

one

or

two

persons.

The

only

help

provid-

ed

by

the

EEOC

Guidelines

is

stated

in

4D:

Greater

differences

(than

4/5)

in

selection

rate

may

not

constitute

adverse

impact

where

the

differences

are

based

on

small

numbers

and

are

not

statistically

significant.

While

this

statement

may

be

helpful

in

a

courtroom,

it

is

of

no

help

in

deter-

mining

which

of

your

backlog

of

tests

are

in

greatest

need

of

validating.

What

is

needed

is

information

concerning

the

behavior

of

different

simple

statistical

techniques

when

dealing

with

small

amounts

of

data

from

the

real

world.

The

statistical

techniques

used

should

be

as

uncomplicated

as

possible

in

order

to

facilitate

the

use

by

small

staffs

which

may

have

members

of

limited

technical

expertise

and/or

no

hardware

assistance.

The

data

used

must

be

from

the

real

world

since

a

major

obstacle

in

the

field

is

the

indeterminability

of

the

actual

variable

or

variables

causing

the

adverse

impact.

While

many

hypotheses

may

be

advanced

as

to

the

cause,

e.g.

cultural

familial

retardation,

institutional

discrimination,

test

anxiety,

etc.,

there

is

no

way

that

the

total

reality

of

the

situation

can

be

ascertained.

This

indeterminable

variable

causes

an

inapplicability

of

theoretical

infor-

mation

regarding

the

different

statistical

techniques.

It

should

also

be

noted

that

since

we

have

insufficient

information

concerning

this

variable

there

is

the