Chapter 4 Psychological Testing 99

think you'll be finished taking tests

when you graduate from
college, think
in. You
be e
will expected
to take tests for
many more years. How well you per-
rm on them can delermine the path of your career, and even your life. Do iou
plan to continu. your education in graduate school to get a master's degree in
paloov? Perhaps a law degree or an MBA? Graduate and
cchools all have admissions and certification tests. professíonal
Do vou expect to find a full-time job right aftet graduation? Few
ans will be interested in hiring you until
organiza-
they see how you score on their tests
far iob applicants. Once you're hired, you'll be asked to test for
training pro-
orams and possible promotions. Even NFL football teams administer
cal tests to prospective players. One uses personality tests to detect psychologi-
whether a
nlayer might tend to be overly aggressive and likely tò cause trouble on or off the
field.
Psychological testing is big business in today's workplaces. 1-0 psychologists
are responsible lor
designing, standatdizing. and scoring the tests that will play
an important part in your future.

For general information abaut tests, log on to the American

Psychological Association's FAQ site at www.apa.org/science/
test.html. Just for fun, try www.davideck.com for.online
tests
of personality, intelligence, health, and career
interests. This
site has some surprises, too.

PRINCIPLES OF PSYCHOLOGICAL TESTING

Carefully developed and researched
psychological tests have several characteris-
tics that them apart from the tests
set
published in the Sunday newspaper or on a
self-help Web site, the ones that ask "Are You a
Good Spouse?" or "What Is Your
Sex Quotient?" A good test involyes more than a list of questions thát
be relevant to the variable appear to
being measured. A proper psychological test is stan-
dardized, objective, based on sound norms, reliable, and valid.
Standardization
Standardization refers
to the
coñsistenicy
proceduresfor administering a
or uniformity
of the conditions and
sStandardization The consis
performance of several job
psychological test. If we
expect to compare the tency or uniformity of the
conditions and procedures for
that test under applicants on the same test, then
they must all take administering a psychological
identical circumstances. This means that every student or job ap- test.

plicant takjng the test reads or listens to the same set of instructions. is allowed
ne same amount of time in which to respond, and is situated in a similar
cal environment. physi-
Any change in testing procedure
produce system
change in
in theindividual
may a
test
perlormance. For example, if the air conditioning testing room
Dreaks down
as well as
during a summer day, the people taking. the test
may not pertorm
people who took the test under more comfortable
conditions. lf an
 Human Resources
100 Part Two The Development of
instructions
fails to read the complete
to.
or careless tester the test under thu
inexperienced are not taking
then those applicants
ofjob applicants,
conditions as other applicants
same
by its devel.
into a test
procedures can be designed
Appropriate testing is the responsibility
of the perS
standardized conditions d
but maintaining administrators in prona
Therefore, the training of test
ministering the rendered useless by an untrain Po
test.

vital. An excellent test c a n be ined

cedures is
inattentive tester.
is helping to maintain
standardized conditions ofbf .

Computer technology- computer-assisted tecs

a r e increasingly using
administration. Organizations instructions in the
test receives the same
ensure that everyone taking the same
format.

Objectivity test results. For a test to h

to the scoring of the
Objectivity refers primarily test obiain the sam
that everyone scoring the
scored objectively, it is necessary or biases on the
must be free of subjectivejudgments the
results. The scoring process
part of the scorer.
have taken both objective and subjective exami.
In your college career you
as those containing multiple-choice and
Objective tests Tes1s for nations. With objective teats (such no special training
mechanical process that requires
which the scoring pro- true-false items) scoring is a
department, an under
cess is free of personal
judgmeni or bias. o r knowledge. A
clerk in a company's human resources
or a computer program can score
graduate grader in the psychology department,
as a
as long key with the correct answers has
scoring
an objective test accurately
been provided. is more
Tests that Scoring subjective tests (such as those containing essay questions)
Subjective tests be iníluenced by the scorer's personal characteris
contain iems such as essay
difficult, the results can
and
questions. The scoring process who took the test. For a fair and eq-
can be influenced by the prfr tics, including a like or dislike for the person desirable.
a n objective test is more
sonal characieristics and atti uitable assessment of a job applicant, then,
Ludes of the scorer.

Test Norms
a frame of reference o r point ot
To interpret the results of a psychological test,
the performance of one person can De
comparison múst be established so that ot tne
others. This is accomplished by means
compared with the performance of similar in nature
of people
test norms, the distribution of scores of a large group
Test n o m s The distribution standard
of this group, called the
of test scores of a large group o the job applicants being tested. The scores
of peopie similar in nature to
the job applicants being tested.
ization sample, serve as the point of comparison in determining.the relau
Standardization sample Standing of the applicants on the ability being tested.
The group ol subjects used
a high school graduate applies for a job that requires
mechani
Lo establish test norms. Thec skillsSuppose
ánd achieves a score of 82 on a test of mechanical ability. This score, ao
scores of ihe standardization thal
sample serve as the poini of tells us nothing about the level of the applicant's skill, but if we compare
comparison for determining the test
the relative standing of the score of 82 with the iest norms-the distribution of scores on the
persons being iesied. can ascribe some meaning
large group of high school graduates-then we

individual score: -we-

f the mean of ihe test norms is 80 and the standard deviation is d
m o a

know imnmediately that an applicant who scores 82 has only an average o ve

are

erate amount of mechanical ability. With this comparative information, ing

in a better position to evaluate objectively the applicant's chances of suce
on the job relative to the other applicants tested.
 Chapter 4 Psychological Testing 101
used
The most widely
for different aage group%,
sychological tests
have sets of norms
for for men and
w o m e n ,
various raclal and
cation. The adequacy of a test's ethnic groups, and for
levels
selection program. norms can determine its usefulness
in any cmployeo
Reliability
. l t v rrefers
Rellabllity to the consistency or
eagnitive ability test one weck andstability
of
response
achíeves a mean
on a wst
If a group Reliability The consistency
takC test a week later añd catns a mean score
score of 100, and re or stability of a respon on a
peats of 72,
would describe the tést aswould have to con
clude that somethi is wrong. We we piychological test.

t ields inconsistent rements. It is common to find a unreliable because

scores whe. when a test is
retaken at a later date, but if the slight variation in test
amiss with the test or the fluctuation is great, some
thing is scoring method.
Three ways to determine reliability are the test-retest
.
method, and the split-halves method. The test-retest method, the equivalent
administeringa test twice to the same method involves Test-retest method A
group of people and correlating the two to determine test
way
e of
sets scores..
The closer the correlation coefficient reliability
that involves adminísterdng a
(called, in this case, the relia-
hility coefficient) approaches perlect positive correlation of +1.00, the
a new testtwice
to the same
ing theoftwo
liahie the test is considered to be. In choosing a test for more re. goup subjects
e1s ofand corelat-
scotes.
employee selection, the
reliabikty coefficient ideally should exceed +.80, although in practice a value of
approximatelyis +.70 is
considered acceptable. This method has several limitations
however. It uneconomical to ask
employees to take time from their jobs to
take the test twice. Alßo, the effects of learning
(remembering the test questions)
and the influence of other experiences between the two
cause the group to score higher the second time.
testing sessions may -
The equivalent-forms method for
determining reliability also uses a
approach, but instead of using the same test a second time, a similar Equlvalent-forms
method
test-retest
Away to dete mineadminisier
test relia
form of the test is administered. The disadvantage of the bility that involves
equivalent-forms ing similar forms of new test
method is that it is difficult and costly to develop two separate and equivalent
a
to the same group of subjects
tests. and correlating the two sets of
SCores
In the split-halves method, the te_t is taken once, the items are
divided split-halves method A
in half, and the two sets of scores are correlated. This method is way to detemine test reliabil-
less' time- ity that involves administering
consuming than the other methods because only one administration of the test anew test to a group of
sub-
5 required. There is no opportunity for learning or recáll to influence the sec- iecns dividinginhait the totul
ond score. number of items, and correlat
ing the two sets ol scores.

Validity
Validity is the most important requirement for a psychological test or any other valldity The determination
cdon device; that is, the test or selection device must be shown to measure owhether apychological
test or other selection device
i n t e n d e d to measure. I-0 psychologists consider several difierent kinds measures what it is iniended
of validity. to measure.

tuion-Related Validity. Suppose that an 1-0 psychologist working for the

e orce develops a test to meàsure the proficiency of radar operators. The
b e considered valid if it measures those skills needed for competent per-
dnce on the job. One way to determine this is to correlate test scores with
Some me If who score
or criterion, of subsequent job performance. persons
high ure,radar the job, and if
those SCore operator proficiency test also perform.well on
low on the test perform poorly on the job, then the validity
 Resources
Part Two The Development of Human
102
coefficient between test scores and job performance will be high. We will .
that the test truly measures the skills needed to be a good radar operator and
success. Validity coefficients of
+.30 to +.40 are consid.d
validpredictor job
of
acceptable for employee selection tests. is called criterion
This approach to defining and establishing
validity ion-
Criterlon-related validity
the nature or properties of the test it
ALype olvalidiryconcemed related validity. It is not concerned with
the relationshlp between test scores and later measures of iak
self but rather with the relationship between
with
test scores and subseqvent job

performance performance. validity are predictive valid.

Two approaches to establishing criterion-related
involves administering the new
and concurrent validity. Predictive validity
Predictive valldity An ap ity
proach to cstablishing crite test to all job applicants for a specified
period and hiring them all, regardless of
rion-related validity in which measure of job performance, such as
their test scores.At a later date, when some the
can be obtained on each worker,
a new test is administered to
all job applicants, and all ap- figures or supervisor ratings,
production to determine how well
plicantsare hired. regardless criteria are correlated
test scores and the job performance however, top management
of scores Later. wheri
test a
In most organizations,
measurc ofjob performance the test predicted lob success. some of the people hired wil
can be obtained, test scores because obviously
to use this approach
are correlaied with job perior is unwilling
mance to see how well the be poor workers. is
test prediced job success.
turn out to
more pòpular
with management than predictive
Concurrent validity is on the job and correlat
Concurrent validity A way
involves giving the test to
employees already
validity. It this method is that
data. The disadvantage of
to establish criterion-related
validity thal involves adminis- with job performance the better
employees
ing the scores
the validation sample
contains mo[tly
current workrs,
test to on
1ering a
the job and correlating their by testing only have quit or been fired, de.
scores with job performance Poorer workers will probably already
employees. establish by the concurrent va.
data.
transferred. Therefore, it is difficult to
moted, or the test is truly distinguishing
between god and poor
whether
lidity method
workers. for a job and
with concurrent validity is that applicants
Another problem motivations and are likely to per.
on the job have different
employees already tests. Applicants may
be more highly motivated
form differently on psychological
wórkers, who may feel more
employment tests than current
to perform well on
situation.
secure in their work used in the establishment of
criterion most often
The job performance
criterion-related validity is ratings assigned
by a supervisor to an employee's pre
as part of the em
Such ratings are made routinely
sent level of job performance. of jobs. (We discuss
appraisal process for all types
ployee performance
in Chapter 5.)
performance appraisal

I1-0 psychologists arealso interested in rational validity,

Rational Validity. a test
nature, properties, and content of
Rational validity The type
of validity that relates 1o the the kind of validity that relates to the
nature properties, and to measures job of performance. In sóme employ
content ofa jest, indepen- independent of its relationship criterion-related validity, perhaps
not feasible to establish
dent of its relationship to job ment situations it is and time-consuming
perfomance measures.
small to support the expensive
because the company is too
has no precedent. In the
initial selection o
validatian process or because the job had taken
for example, before any space flights
the original U.S. astronauts, correlated wn
there were no measures of job
performance that could be
place,
rational validity ofa test arecontentva
test scores.

Two approaches toestablishing the items are assesst

In content validity, the test
Content valldity A type of lidity and construct validity. test is design
adequately sample the knowledge or skills the
validity that assesses test items ensure that they conducting a job anayo
to ensure that they adequajely
This assessment can be accomplished by
sample the skills the test is de to measure. needed to pe
are related to all
those abilities
signed to measure.
and determining if the test items
 Chapter 4 Psychological Testing 103

torm the job. Wih á word-processing job, for example, test questions about
word-procesing computer software are job-related, whereas questions about
isical abilities may not be. In the classroom, if your professor a:1nounces that
will be tested on the lirst three chapters of this book, then questions about
information from other chapters would not be considered content valid.
Construct validity is an attempt to determine the psychological character Construet validity A lype of
cics measured by a test. How do we know that a new test developed to measure validity that attempts to deter.
aelligence or motivation or emotional stability really does so? One way to mea- mine the psychological char
actenstics measured by a test.
sure construct alidity statistically is to correlate scores on the new test with
cores on established tests that are known to measure these variables. If the cor-
relation is high, then we can nave some confidence that the new test is measur-
it claims.to measure.
ing the trait

Cace Validity. Face validity is not a statistical measure but a

nression of how well the test items seem to be related to the job in subjective im
question. Air- Face valldity A subjecive
impression of how well tes
ine pilots would not think it unusual to take tests about mechanics or items seem to be related to the

because these topics are directly related to the job they expect to perform,navigation
but requirements of a job.
they might balk at being asked if they loved their parents or slept with a light on
in their room. Such questions might be related to emotional
stability, but they do
not appear to be related to flying tn airplane. If a test lacks face
cants may not take it seriously, ar.d this may lower their test
validity, appli-
College students from the United States and from Franceperformance.
(a total of 259 sub-
jects) were asked rate the effectiveness and íairness of several selection
to
dures, to assess the students' degree of favorability toward the proce-
measures. The
selection procedures included personality, ability,
tests, as well as interviews, résumés, biographical data,honesty, and work-sample.
tacts, and handwriting analysis (a selection
references, personal con-
technique popular in
countries). The single strongest correlate of favorable ratings of these European
was the face validity techniques
ofí the measures (Steiner & Gilliland, 1996).
The best psychological tests include in their manuals the results of validation
studies. Without this information, the human
can have little confidence that resources or personnel manager
the tests in the
program are actually measuring the qualities andcompany's employee
abilities being soughtselection
in new
employees. Test validation is expensive, but proper validation
more than pay for themselves. procedures will

Validity Generalization
Until the late 1970s, 1-0
psychologists
ficity," which recommended
followed doctrine of "situational
a
speci
job and validating a test inevery situation-that is, every
every organization-for which it was
Were assumed to be differentially valid. A test chosen as a selection device. Tests
tory technicians in one company was not automaticallyappropriate for selecting labora-
ecting technicians in another considered valid for se-
Conidence for company. Therefore, no test could be used with

8iven
employee selection without first determining its validity in'the
instance, no matter how
valid the test had proven for other,
The idea of
situatiónal specificity or differential similar jobs.
validity generalization. On the basis of validity has been replaced by Validity generalization The
studies, 1-0 meta-analyses of previous validation
psychologists have concluded that tests valid in one situation may
idea that tests valid In one sit
uation may also be valid in
another situation.
 Human Resources
104 Part Two The Development of

alca be valid in another situation. In other words, once establshed, the validity
(Schmidt et al., 1993).
of a test can be generalized
and Organlzational Psychology (SIOP) SuDno
The Society for Industrial
has also been endorsed by the National Academy.
validity generalization., It
Sciences and is included
in the Standards for Educational and Psychological Te
is widely accepted not only for tests hue
ing of the APA. Validity generalization
also for biographical data, assessment centers, interviews, integrity tests, and
and government agencies apply
other selection devices. Many large corporations
selection programs.
the concept of validity generalization in their
for psychological
Validity generalization has important practical implications
as an employee selection technique.
Organizations have realized that if
testing for every job at every
tests no longer require expensive validation proceduresby including tests, while
selection programs
level, then they can improve their
saving time and money.

FAIR EMPLOYMENT PRACTICES

in validity re.
legislation has been an increase
One result of fair employment
If studies show that appli.
test is discriminatory.
search to document whether a on the
score bElow a certain
level on a test perfom poorly
cants of all races who research confirms
race. 1-O psychology
then.the test is not discriminatory by
job, coelficients have relatively
low levels of adverse im.
that tests with high validity test scores with job per
procedures (correlating
Criterion-related validation
pact. the guidelines of the Equal
are required, when leasible, by
formance measures) Rational validation procedures
Employment Opportunity Commission (EEOC).
may also be
used to meet EEOC requirements. to be valid for an-
valid for one group is expected
A test that is found to be test
that racial differences among average
other. Research has demonstrated educational, social, and cul.
from
scores are not the
result of test bias but arise
known scientists issued the follow.
tural differences. A group
of internationally
Journal:
ing statement, published in 1994 in the Wall Street
American blácks or other
Intelligence culturally biased against
tests are not
in the U.S. Rather, 1Q scores pre.
native-born, English-speaking peoples
of race and so
dict equally accurately for
all such Americans, regardless
cial class.
Association's ßoard of Sci
American Psychological
A report prepared by the are not biased against
concluded that cognitive ability tests
entific Affairs in 1996 discriminate against minority
African Americans. In other
words, the tests do not
discrimination that has
Rather, they reflect in quantitative terms the
groups.
time.
been reated by society over not guarantee
that a
demonstration of validity does
However, the empirical use. This
oc*

declared discriminatory and barred from further

test will not be
Curredwith-the-General Aptitude Test Baltery (GATB)usedby
the U.S. Employ
it also assesses manua
The GATB is a test of cognitive functioning:
meni Service.
than 750 validity studies
have confirmed the high va
and finger dexterity. More instrüment for employment.
Test validities 10r
lidity of the test as a screening minority job applicano
were comparable, although
whites and minorities hired.
scores and hence were
less likely to be
showed lower average