User's Manual

For The
Employee Reliability Inventory
Gerald L. Borofsky, PhD

1.800.661.5158
www.psychometrics.com
Revision 6.0 - 11/93
Revision 6.5 CDN -10/2004
 ERI® IS A TRADEMARK OF
PSYCHOMETRICS CANADA LTD.
NO MATERIAL DESCRIBING THE ERI® MAY BE PRODUCED
OR DISTRIBUTED WITHOUT THE WRITTEN PERMISSION OF
PSYCHOMETRICS CANADA LTD.

©
COPYRIGHT 2017
PSYCHOMETRICS CANADA LTD.
ALL RIGHTS RESERVED

Psychometrics Canada Ltd.

Edmonton, AB
 CONTENTS

1. INTRODUCTION 1
1.1 PRE-EMPLOYMENT SELECTION PROGRAMS 1
1.2 UNRELIABLE AND UNPRODUCTIVE BEHAVIOR 2
1.2.1 TYPES OF UNRELIABLE AND UNPRODUCTIVE BEHAVIOR 2
1.2.2 COMMON CAUSES OF UNRELIABLE BEHAVIOR 2
1.2.3 THE MULTIFACTORIAL NATURE OF UNRELIABLE
AND UNPRODUCTIVE BEHAVIOR 2
2. ERI® - EMPLOYEE RELIABILITY INVENTORY 4
2.1 DESCRIPTION 4
2.2 RECOMMENDED USE OF THE ERI® 4
3. PROCEDURES AND INSTRUCTIONS FOR THE USE,
ADMINISTRATION, AND SCORING OF THE ERI® 4
3.1 GENERAL INFORMATION REGARDING THE USE OF THE ERI® 4
3.2 LIMITATIONS TO THE USE OF THE ERI® 8
3.3 ERI® USER AUTHORIZATION 8
3.4 ADMINISTRATION OF THE ERI® 8
4. REPORTING AND INTERPRETING ERI® RESULTS 10
4.1 REPORTING ERI® RESULTS 10
4.2 SOME INTERPRETIVE GUIDELINES 10
5. MAKING USE OF ERI® RESULTS 13
5.1 SOME IMPORTANT PRINCIPLES OF ERI® USE 13
5.2 AN IMPORTANT NOTE REGARDING SMALL
DIFFERENCES IN RESULTS 14
5.3 REVIEWING RESULTS FOR POSSIBLE PROBLEMS 14
5.4 "FAKING GOOD" AND ERI® RESULTS 15
5.5 USING THE ERI® AS A PRE-INTERVIEW QUESTIONNAIRE 16
5.5.1 SOME GENERAL GUIDELINES FOR ASKING
FOLLOW-UP QUESTIONS 16
5.6 TECHNICAL SUPPORT 17
6. CONSTRUCTION AND VALIDATION OF THE ERI® 17
6.1 ORIGINAL CONSTRUCTION OF THE ERI® 17
6.2 ORIGINAL VALIDATION OF THE ERI® 18
6.2.1 CONSTRUCT-RELATED VALIDITY AND THE ERI® 18
6.2.2 CONTENT-RELATED VALIDITY AND THE ERI® 18
6.2.3 CRITERION-RELATED VALIDITY AND THE ERI® 18
6.3 R SCALE 19
6.3.1 CRITERION GROUPS 19
6.3.2 VALIDATION 20
6.3.3 CROSS-VALIDATION 20
7. SUBSEQUENT DEVELOPMENT OF THE ERI® 21
7.1 A SCALE: CONSTRUCTION AND VALIDATION 21
7.1.1 CRITERION GROUPS 21
7.1.2 VALIDATION 22
7.1.3 CROSS-VALIDATION 22
7.2 C SCALE: CONSTRUCTION AND VALIDATION 23
7.2.1 CRITERION GROUPS 23
7.2.2 VALIDATION 23
7.2.3 CROSS-VALIDATION 24
7.3 E SCALE: CONSTRUCTION AND VALIDATION 24
7.3.1 CRITERION GROUPS 24
7.3.2 VALIDATION 25
7.3.3 CROSS-VALIDATION 25
 7.4 F SCALE: CONSTRUCTION AND VALIDATION 26
7.4.1 CRITERION GROUPS 26
7.4.2 VALIDATION 26

7.4.3 CROSS-VALIDATION 26
7.5 H SCALE: CONSTRUCTION AND VALIDATION 27
7.5.1 CRITERION GROUPS 27
7.5.2 VALIDATION 27
7.5.3 CROSS-VALIDATION 28
7.6 Q SCALE: CONSTRUCTION AND VALIDATION 28
7.6.1 CRITERION GROUPS 28
7.6.2 VALIDATION 28
7.6.3 CROSS-VALIDATION 29
7.7 S SCALE: CONSTRUCTION AND VALIDATION 29
7.7.1 CRITERION GROUPS 30
7.7.2 VALIDATION 30
7.7.3 CROSS-VALIDATION 30
8. SOME PSYCHOMETRIC PROPERTIES OF THE ERI® 31
8.1 DESCRIPTIVE STATISTICS 31
8.2 Q UENCY DISTRIBUTIONS FOR ERI® SCALES
FREQ 32
8.3 INTERCORRELATION OF THE ERI® SCALES 33
8.4 TEST-RETEST RELIABILITY OF THE ERI® SCALES 33
9. FAIRNESS OF THE ERI® AND ADVERSE IMPACT 34
9.1 RACE 34
9.2 GENDER 36
9.3 AGE 36
9.4 SUMMARY 36
10. SELECTED REFERENCES 37
APPENDICES
APPENDIX: SAMPLE REPORT
 R

EMPLOYEE RELIABILITY INVENTORY

1, 2
Gerald L. Borofsky, Ph.D.

1. INTRODUCTION • Some people become reliable and

1.1 PRE-EMPLOYMENT SELECTION trusted employees while others do not.

PROGRAMS
• Some people drift from job to job, while
The basic reason that there is a need for pre- others become valued long-term em-
employment selection programs is that people ployees.
are different from one another. The ways in
• Some people perform their work in a
which people differ from one another can be
safe manner while others do not.
critically important to an employer.
Employees who behave on the job in an unreli-
• Some people’s job performance is not
able or unproductive manner can have a
disrupted by alcohol or illegal drug use
negative effect on an organization’s efforts to
while others’ is not.
carry out its objectives. For that reason,
• Some people are courteous and cus- organizations have a need to assess the likeli-
tomer service oriented while others are hood that job applicants will perform on the job
not. in a reliable and productive manner, if hired.
Most employers routinely use pre-employment
• Some people are emotionally mature
selection procedures to:
while others are not.
• Identify applicants who are well suited
• Some people are productive and consci-
for a particular position.
entious workers while others are not.
1 Gerald L. Borofsky, holds a Ph.D. in Clinical Psychology from • Identify applicants who are poorly
Michigan State University. He is a licensed psychologist, a
Diplomate in Clinical Psychology of the American Board of suited for a particular position.
Professional Psychology, and is listed in the National Register of
Health Service Providers in Psychology.
For 15 years, Dr. Borofsky was the Director of Psychology in the Commonly-used pre-employment selection
Department of Psychiatry at the Massachusetts General Hospital.
For over 25 years he was a member of the faculty of Harvard procedures include the use of:
Medical School. Dr. Borofsky is an internationally-recognized
authority on the development and use of assessment systems. He
is also a nationally-recognized authority on the development of
• Employment Applications
human reliability programs and forensic psychology. He has
contributed to the development of Congressional policy in the
• Verification of the Employment Applica-
area of pre-employment screening, and has also served as a
consultant in the area of pre-employment assessment, for law tion (Checking of References, Past
firms, police departments, and the development of governmental Employers, Credit Checks, etc.)
programs and policies.
2 The suggestions and assistance of the following individuals is
gratefully acknowledged: Victor Artese, Allison Betts, Norm • Pre-Employment Interviews
Fujiwara, Veronica Johnson, Arthur F. LeBlanc, Sharyn
MacLean, Bruce McCormick, Mark Palmerino, Sharon Varallo,
Joan Wagner, and Charles Wonderlic, Jr.

1
 • Psychologically-Based Assessment reasons, only remain on the job for a short
Methods (Questionnaires and Tests) period of time (e.g. being fired for violations of
company policy or drifting from job to job).
1.2 UNRELIABLE AND This type of behavior can seriously disrupt an
UNPRODUCTIVE BEHAVIOR
organization’s ongoing effectiveness. Such
1.2.1 TYPES OF UNRELIABLE turnover also results in significantly increased
AND UNPRODUCTIVE BEHAVIOR administrative and training costs associated
with the recruiting, hiring, and training of new
A 1983 study sponsored by the National Insti-
employees.
tute of Justice3 revealed that many organiza-
tions in this country are being negatively 1.2.2 COMMON CAUSES OF
affected by the presence of employees whose UNRELIABLE AND
overall performance on-the-job is characterized UNPRODUCTIVE BEHAVIOR

by unreliable and unproductive behavior. This

In order to minimize the occurrence of Produc-
study also found that such unreliable and
tion Deviance, Property Deviance, or Un-
unproductive behavior can be found in just
planned and Uncontrolled Turnover in the
about all job categories within an organization,
workplace, a comprehensive selection program
ranging from unskilled laborers to upper level
typically assesses job applicants for each of the
managers.
five common causes of unreliable behavior.

When referring to the subject of unreliable and

These are:
unproductive behavior in the work place, a
distinction can be made between three types of • Insufficient training or experience to
behavior: (1) Production Deviance, (2) Property adequately carry out the requirements
Deviance,4 and (3) Unplanned and Uncontrolled of the job
Turnover. • Presence of maladaptive personality
traits
PRODUCTION DEVIANCE refers to behavior
• Presence of adaptive personality traits
which conflicts with an organization’s expecta-
which are nonetheless in conflict with
tions of productivity. Common examples of
the specific requirements of the job
production deviance include failure to follow
• Disruptive use of alcohol or illegal drugs5
standard procedures, frequent unauthorized
• Untrustworthiness
absences, coming to work intoxicated, on-the-
job use of alcohol or illegal drugs, and a higher 1.2.3 THE MULTIFACTORIAL NATURE
than average number of injuries and accidents. OF UNRELIABLE AND
UNPRODUCTIVE BEHAVIOR
PROPERTY DEVIANCE refers to behavior which
is directed against an organization’s property. A question of significance for the field of em-
Common examples of property deviance include ployee selection is whether there are interrela-
theft and vandalism. tionships among these various causes of
unreliable and unproductive behavior.
UNPLANNED AND UNCONTROLLED TURN-
OVER refers to employees who, for a variety of Consider the following example:

5 As used in this Manual, and in all other ERI® documentation

3 John P. Clark and Richard C. Hollinger, Theft by Employees.
and materials, the terms drug use and substance use refer to the
(Lexington, MA: Lexington Books, 1983). current illegal use of drugs.
4 Ibid.

2
Pre-employment selection procedures may unproductive behavior should be sufficient to
indicate that an applicant appears acceptable accurately identify the “bad apples” in an
in terms of his or her training and experience, applicant pool.
and that his/her job performance appears
unlikely to be disrupted as a consequence of An alternative perspective can be called the
his/her pattern of alcohol or illegal drug use. multifactorial view of behavior. According to
On this basis, would it be reasonable for an this point of view, different individuals perform
employer to assume that the applicant is also unproductively or unreliably for different
likely to work cooperatively with others and to reasons. For example, one job applicant may
exercise good judgment on the job? Would it have personality traits which are well suited to
also be reasonable for an employer to assume the requirements of the job, but may be vulner-
that the applicant will make a long term com- able to performing unreliably on the job as a
mitment to the job? consequence of his/her pattern of illegal drug or
alcohol use. Another job applicant may be
Whether the various causes of unreliable and trustworthy but may be vulnerable to perform-
unproductive behavior are intercorrelated has ing unreliably on the job as a result of person-
implications for the design of pre-employment ality traits which conflict with the requirements
selection programs. If there are significant of the job.
interrelationships between the various causes,
then an employer might be able to assess job The multifactorial perspective accepts that
applicants for just one cause, on the assump- there may be some degree of intercorrelation
tion that he or she was concurrently assessing among the various causes of unreliable and
applicants for the other causes as well. On the unproductive behavior. However, unlike the
other hand, if these five causes are relatively “bad apple” viewpoint, the conclusion to be
uncorrelated, it would seem advisable to in- drawn from the multifactorial perspective is
clude procedures which assess applicants for that a comprehensive pre-employment selection
each job relevant cause. program should assess job applicants for all of
the job-relevant causes of unreliable behavior.
Conventional wisdom seems to subscribe to According to the multifactorial view, it is quite
what can be called the “bad apple” view of unlikely that assessing job applicants for just
behavior. According to this point of view, all one cause of unreliable behavior will ad-
unreliable and unproductive people are more or equately call attention to applicants who may
less similar to each other, in that they are all be likely to perform unreliably on the job due to
likely to manifest multiple causes of unreliable other causes.
behavior. For example, the “bad apple” per-
spective would predict that if a job applicant is Preliminary studies, including the correlation
likely to demonstrate maladaptive personality matrix for the scales of the Employee Reliability
traits on the job, he or she is also likely to have Inventory (ERI® ) shown in Section 8.3 of this
disrupted job performance as a consequence of Manual, seem to favor the validity of the multi-
his/her pattern of illegal drug or alcohol use, factorial perspective over the “bad apple” point
and is likely to perform on the job in an un- of view.6
trustworthy manner. For purposes of pre- 6 Evans, M.A., McGee, M.P., and Borofsky, G.L. Psychological
evaluation and illicit drug use in an industrial population.
employment assessment, the implication of the Proceedings. American Academy of Forensic Science. 1986.;
Borofsky, G.L., Friedman, J.,and Pignato, J. C., Interrelationships
“bad apple” perspective is that assessing appli- Among Various Causes Of Unreliable Behavior In The Work-
place, Research Bulletin, Number 400-9. Groton, MA: Human
cants for any single cause of unreliable or Reliability Institute, 1987

3
2. ERI® - EMPLOYEE RELIABILITY completed the employment application and
INVENTORY
before they are interviewed or references are
2.1 DESCRIPTION verified. When used in this manner, the ERI®
can serve as an objective method for assessing
The ERI® is a criterion-keyed, self-adminis- issues related to job performance, which can
tered, “True - False” type of behavior inventory. then be explored further during interviews and
It contains 81 statements, which are worded at reference verification. If an applicant’s score on
a sixth grade reading level. Results for each one or more scales suggests a possible problem
scale are presented in an easy to understand area, subsequent interviews and the verifica-
format designed to add flexibility to your tion process can be used to develop additional
selection process. Results provide an estimate information, which will clarify, confirm or call
of the likelihood that a job applicant will into question the ERI® results. Coordinated
perform, on the job, in a reliable and produc- use of the ERI® with focused inquiries during
tive manner. interviews and verification, should increase the
likelihood of identifying reliable and productive
The ERI® consists of seven (7) separate scales.
individuals prior to making a hiring decision.
Each scale assesses job applicants with respect
to a different dimension of reliable and produc- Because it was validated as a pre-employ-
tive behavior7. The seven (7) ERI® scales are ment questionnaire, the ERI® should not be
described on the following page. used to evaluate current employees or for
any purpose other than as an aid in the pre-
2.2 RECOMMENDED
USE OF THE ERI® employment selection process.

The ERI® was designed and developed to be The ERI® is not a medical examination, nor
used as one part of a company’s pre-employ- should it be administered in a medical setting.
ment selection program. The purpose of the It is not a psychological stress evaluator, is not
ERI® is to assist employers in their efforts to invasive, and does not measure physiological or
hire reliable and productive employees. It was psychological responses in the subject being
designed to be used where production devi- assessed.
ance, property deviance, and unplanned and
3. PROCEDURES AND INSTRUCTIONS
uncontrolled turnover are important job con- FOR THE USE, ADMINISTRATION
cerns. Accordingly, the ERI® should be used AND SCORING OF THE ERI®
where such behaviors are related to job perfor-
3.1 GENERAL INFORMATION
mance in the position for which the applicant is REGARDING THE USE OF
being considered. THE ERI®

The ERI® was designed to be used as a pre- 1. The ERI® user materials include docu-
interview questionnaire. For this reason, it is mentation, technical information,
recommended that the ERI® be administered to business information as well as various
job applicants immediately after they have printed materials, which are collectively
7 Each of the ERI® scales assesses job applicants with respect referred to as “ERI® documentation and
to a different behavioral dimension. The scales estimate the materials”.
likelihood that an applicant, if hired, would perform on the job in
a reliable and productive manner.
However, you should be aware that subsequent changes in a
person’s life and work environment (both positive and negative)
can affect his/her performance on the job. Supervisory knowl-
edge of employees and supervisory observation of on the job
behavior are commonly - used techniques for monitoring and
helping to enhance job performance, once an applicant is hired.
4
 ®
ERI SCALES

SCALE NAME WORK BEHAVIOR SKILLS ASSESSED

Freedom From Disrupted Job Performance

Assesses the likelihood that an applicant's work performance
will be reliable, in that his/her performance will not be dis-
A rupted by behaviors such as inattentiveness, unauthorized
absence/lateness, failing to follow through on assignments, or
other inappropriate work behaviors.
It is important to emphasize that this scale does not assess the
extent of prior or current alcohol or illegal drug use. Similarly,
it is not designed to reveal, nor should it be used for the
purpose of revealing, the existence, nature, or severity of a
disability.

Courtesy
Assesses the likelihood that an applicant's interactions with
C customers/guests will be characterized by a high level of
courtesy and commitment to service.

Emotional Maturity
Assesses the likelihood that an applicant's work performance
will be characterized by mature behavior, and that it will not
E be disrupted due to the presence of maladaptive personality
traits such as irresponsibility, poor judgement, difficulty in
working cooperatively with others, poor frustration tolerance or
poor impulse control.
It is important to emphasize that this scale does not assess,
nor should it be used to assess, for the presence of a mental or
psychological impairment or disorder, or an applicant's
general physical or psychological health.

Conscientiousness

F Assesses the likelihood that a candidate will perform on the

job in a productive and conscientous manner and will not be
fired.

Trustworthiness
H Assesses the likelihood that a candidate will perform in a
trusworthy manner and will not engage in various forms of
property deviant behavior.

Long Term Job Commitment

Q Assesses the likelihood that a candidate will make a long term
commitment to the job and will not quit.

Safe Job Performance

S Assesses the likelihood that a candidate will perform on the job

in a safe manner and will not have a significant on-the-job
accident.

5
2. ERI® documentation and materials 5. Only ERI® scale scores should be used.
should be used solely for pre-employ- Answers to the individual statements
ment purposes. They should not be contained in the ERI® should never be
used to assess existing employees or for used as part of the selection process.
any other purpose.
6. At no time should ERI® results be
3. In order for the ERI® to be effective as a
discussed with applicants, regardless of
pre-interview questionnaire, it is neces-
the selection outcome. For reasons of
sary to insure that you, your employees
security and confidentiality, ERI®
(or representatives), and job applicants,
results should never be discussed with
fully comply with all of the procedures
® unauthorized employees or representa-
and instructions contained in the ERI
tives.
documentation and materials, includ-
ing, but not limited to those contained 7. A job applicant’s failure or refusal to
in this User’s Manual. completely follow the instructions given
to him/her should not be the basis for
ERI® documentation and materials making any inferences regarding that
should not be reproduced or copied, in individual’s future job performance.
whole or in part, except as necessary for
use by you, as authorized in this 8. You and your authorized employees or
representatives should be aware that
Manual.
access to ERI® results which are part of
You and your authorized employees or an employee’s personnel, medical or
representatives should not mark, alter other records, may be governed by
or deface ERI® documentation and specific laws in your state. For this
materials other than in accordance with reason, only authorized individuals
the Procedures and Instructions con- should have access to ERI® results.
tained in this Manual.
9. Psychometrics Canada has developed
4. The ERI® should be used, administered certain trade secrets, confidential and
and scored only by your authorized proprietary information in the
employees or representatives who have development and use of the ERI®
read and are thoroughly familiar with documentation and materi-als,
the procedures and instructions con- including but not limited to, scoring and
tained in this Manual, and in other ERI® results, information proce-dures,
® systems, techniques, forms, methods
ERI documentation and materials.
and the like which are unique to
Psychometrics Canada and its business
and are not of general public knowledge.

8 For your convenience, an additional copy of the appropriate

User Authorization Form is contained in Appendix E of this
Manual. Please make a copy of this, if needed.

6
All such ERI® documentation and
12. Psychometrics Canada reserves the right to
materials furnished to users should be
revoke or cancel, without advance notice, a user’s
considered confidential. Because of the
license to use the ERI® documentation and
nature of Psychometrics Canada's business materials if the user or the user’s employees or
and the nature of the ERI®, disclosure or representatives fail to comply with all of the
dis-semination of the ERI® documentation Procedures and Instructions contained in the ERI®
and materials could damage the effec- documentation and materials, including, but not
limited to those contained in the ERI® User’s
tiveness of the ERI® as a pre-employ-ment
Manual, the ERI® Guide For Computer Scoring, as
questionnaire and/or compromise the
well as any subsequent revisions or updates sent to
competitive position of Psychometrics you by Psychometrics Canada.
Canada in the marketplace.

10. In making the ERI® available to users

and their authorized employees or
representatives, Psychometrics Canada
is granting the user a personal,
nontransferable and non-exclusive
license to use in the United States,
North America and/or South America,
without the right to sublicense, any or
all ERI® documenta-tion and materials,
in whatever form recorded, which are
furnished, includ-ing any revisions or
updates to these materials.

11. It is expected that users and their

authorized employees or representatives
will use their best efforts to ensure that
all ERI® documentation and materials
are treated as confidential. This infor-
mation must be maintained with the
same level of care and discretion as that
used for similar data which users
designate as confidential. Users should
instruct their authorized employees or
representatives regarding the appropri-
ate measures required to safeguard the
confidentiality of this information.

7
3.2 LIMITATIONS TO THE USE OF
THE ERI ® Computer Scoring, as well as any subsequent
1. The ERI® was developed and validated revisions or updates, sent to you by
to be used as a pre-employment assess- Psychometrics Canada.
ment tool. It is to be used for the
In order to document the above requirements, it
assessment of new job applicants.
Under no circumstances should theERI® is necessary that Psychometrics Canada have a

be administered to current employees, or completed and signed User Authorization Form

used for any purpose other than as an (Computer Scoring or Template Scoring version)
on file for all companies using the ERI®. A copy
aid in the pre-employment selection
of the appropriate version of this form is sent to
process.
you with your order. If for any reason you are
2. The ERI® is not designed to reveal, nor unable to locate this form, please make a copy
should it be used for the purpose of of the sample form included as Appendix E of
revealing, the existence, nature, or this Manual, and use this copy to complete and
severity of a disability, as defined under send to Psychometrics Canada.
the Canadian Human Rights Act.3.
3.4 ADMINISTRATION OF THE ERI ®
The decision to hire or not hire a specific
applicant should not be based solely on
the applicant’s ERI® scores. Hiring
THE ERI ® CAN BE COMPLETED THROUGH YOUR
decisions should be based on a review of
CAREERID SITE.
ALL information collected during the
conduct of the total selection process
.

4. Laws regarding questionnaires such as

the ERI® vary from province to province.
Users are responsible for the monitoring
of any such laws.

3.3 ERI ® USER AUTHORIZATION

In order to maximize proper use, confidential-

ity, and security, Psychometrics Canada
requires that only properly trained and
authorized individuals have access to ERI®
documentation, materials, and results.
Accordingly, please ensure that all individuals
who will have access to ERI® docu-mentation,
materials, and results have read and have been
fully trained to administer, score, interpret, and
use the ERI® according to the Procedures and
Instructions contained in all of the ERI® 9 See sample contained in Appendix A of this Manual. Users
should make copies of this Scoresheet, as needed
documentation and materials, including, but
not limited to those contained in the ERI® User’s
Manual.

8
 ®
PREPARING THE APPLICANT TO APPLICANT’S COMPLETION OF THE ERI
TAKE THE ERI ®
If applicants have questions about answering a
1. Be sure that the ERI® is administered in particular statement, instruct them to use the
a quiet and well-lit location, which is as guidelines contained in the test instructions,
free from distractions as possible. as the basis for coming up with their answer.
Please be sure that the administration If an applicant states that he/she is unable
of the ERI® is conducted in a profes- to answer a particular statement because it
sional manner and that, among other does not apply to him/er (e.g., statement 60,
things, the applicant is encouraged to because he/she never goes into bars), the
adopt a serious and thoughtful ap- applicant should be told to answer the statement
proach to answering all statements. as being “False”.

Research has shown that use of the You may not tell the applicant how to answer a
ERI® does not result in discrimination on statement. However, if the applicant does not
the basis of race, gender, or age (Please understand the meaning of certain words or
refer to Section 9 of this Manual for expressions, you may explain what the word or
detailed information). Nonetheless, expression means.
please be sure to assess each individual
applicant to determine if cultural, ethnic, INSTRUCT APPLICANTS THAT ALL 81
or language factors are likely to interfere STATEMENTS MUST BE ANSWERED AND
with his/her ability to understand the THAT THEY SHOULD GIVE ONLY ONE AN-
ERI® instructions and statements. SWER FOR EACH STATEMENT.
Needless to say, a failure to accurately
understand the instructions and state-
ments, due to cultural, ethnic, or educa-
tional causes, can significantly affect the
accuracy and usefulness of the ERI®. In
this regard, please note that in addition
®
to English, the ERI is available in
Spanish and French translations.

2. In ALL cases, the applicant should be

advised as to the purpose of the ERI ®,
BEFORE it is administered to him/her.

(E.G. “This questionnaire is being

administered to you as one part of your
application for employment with [Your 10 Completion of information regarding age, sex, and race is
Company’s Name]. I would like you to voluntary and does not affect an applicant’s questionnaire results in
any way. Similarly, the process of scoring of the questionnaire
carefully read and follow all of the does not adjust the scores, use different cutoff scores for, or
otherwise alter the results on the basis of race, color, religion, sex,
directions.”) or national origin.

9
 ®
4. REP0RTING AND INTERPRETING ERI 1. The applicant’s score on each scale is
RESULTS shown as a horizontal line extending
® from left to right. Shorter lines indicate
4.1 REP0RTING ERI RESULTS
a higher likelihood of unreliable behavior.
®
For purposes of communicating ERI results, Longer lines indicate a higher likelihood
scores are reported in terms of a system of four of reliable behavior.
(4) arithmetically equal-sized zones (designated
1 through 4). Each of the four zones is further 2. For each scale, different applicants can
subdivided into two (2) arithmetically equal- be compared, based on the “zone” into
sized zones (A and B). Results for each of the which their results fall. Results in Zone
®
ERI scales are reported as falling into one of 1 indicate there is a higher likelihood
the eight (8) zones, as shown in the diagram the applicant will behave in a reliable
below. manner. Results in Zone 4 indicate
there is a higher likelihood the appli-
cant will behave in an unreliable
manner. Results in Zone 3A indicate
there is a higher likelihood of reliable
behavior than results in Zone 3B.

RELIABILITY

ZONE 4 ZONE 3 ZONE 2 ZONE 1

B A B A B A B A

Appendices C and D of this Manual contains You can also use this system of zones to get an
samples of how this system of zones appears approximate idea of how “low” or “high” the
on the written report of results which is sent to applicant’s score is on each scale, relative to
Template Scoring users, at the end of each day. the range of possible scores that can be ob-
A similar, but less graphically elaborate repre- tained on that scale.
sentation of the eight zones appears on your
®
computer screen, each time you score an ERI 1. The table on the back side of these
using the in-house computer scoring option. forms can also be used to help you
approximate where an applicant’s
4.2 SOME INTERPRETIVE GUIDELINES

other job applicants. A copy of this

PLEASE NOTE: Because of the variability
table is also shown on page 15.
inherent in any type of scores, small differ-
®
ences in ERI results should never be used as
the basis for making decisions about appli-
cants or for comparing applicants.

As you read the following guides for interpret-

®
ing applicant’s’ ERI results, please make
reference to either Appendix C or D at the back
of the Manual.

10
 The table shows the approximate regions of the country, represents all 10
percentage of job applicants who fall Standard Industrial Classification (SIC)
into each zone. The table also shows Code Divisions, 54 Major SIC Groups,
the cumulative percentage of job appli- and a wide range of job categories. In
cants who fall into that zone plus the looking at this table, one can see that
zones to the left of that zone. applicants’ results are distributed
continuously along each of the behav-
The table is based on a group of over ioral - psychological dimensions mea-
60,000 job applicants who completed sured by the ERI .
®
®
the ERI as part of their pre-employ-
ment processing. This normative group
of job applicants is drawn from all

Eight Zone Frequency Distributions For The Seven ERI® Scales

By Percent and Cumulative Percent
Based on a General Group of Job Applicants (N=60670)

Likelihood of Unreliable Behavior

RELIABILITY

ZONE 4 ZONE 3 ZONE 2 ZONE 1

SCALE B A B A B A B A

A (2.3) (4.2) (11.9) (20.1) (24.5) (19.9) (13.0) (4.2)

100.0 97.8 93.6 81.7 61.6 37.1 17.2 4.2

(2.5) (12.5) (4.9) (14.3) (20.2) (28.6) (12.2) (4.8)

C
100.0 97.5 85.0 80.1 65.8 45.6 17.0 4.8

(4.9) (3.6) (4.9) (8.9) (22.3) (34.1) (16.4) (4.8)

E
100.0 95.0 91.4 86.5 77.6 55.3 21.2 4.8

(7.0) (1.1) (1.7) (2.8) (18.7) (36.1) (23.8) (8.8)

F 100.0 93.0 91.9 90.2 87.4 68.7 32.6 8.8

(1.9) (1.6) (3.1) (7.5) (18.8) (32.8) (25.8) (8.6)

H 100.0 98.2 96.6 93.5 86.0 67.2 34.4 8.6

(13.1) (1.4) (1.4) (1.7) (18.1) (36.5) (20.9) (6.8)

Q 100.0 86.9 85.4 84.0 82.3 64.2 27.7 6.8

(6.0) (4.8) (11.5) (7.1) (25.1) (24.0) (14.4) (7.1)

S 100.0 94.0 89.2 77.7 70.6 45.5 21.5 7.1

NOTES:
The upper number in parentheses represents the percentage of job applicants in the normative sample who fall into that
particular zone.
The lower number with no parentheses represents the percentage of job applicants in the normative sample who fall into
that zone, plus the zones to the left of that zone - i.e., they represent the cumulative percentage.
For example on the F scale 2.8% of job applicants in the normative sample fall into zone 3A, and 90.2% of job applicants in
the normative sample fall into zone 3A or one of the lower zones (i.e., zones 1A, 1B, 2A, and 2B). Put slightly differently,
90.2% of job applicants in the normative sample get a score in zone 3A or a better score. Only 9.8% (100.0% - 90.2%) get a
poorer score.

11
The table shown below, like the one on the contrast to the previous table, however, here
previous page, shows the distribution of scores the cells indicate the approximate percentile
® ®
on each of the seven ERI scales, for the same distributions for each of the seven ERI scales.
sample of over 60,000 job applicants. In

Eight Zone Frequency Distributions For The Seven ERI® Scales

By Approximate Percentile
Based on a General Group of Job Applicants (N = 60670)

Likelihood of Unreliable Behavior

RELIABILITY

ZONE 4 ZONE 3 ZONE 2 ZONE 1

SCALE B A B A B A B A

A 0% 2% 6% 18% 38% 63% 83% 96%

C 0% 2% 15% 20% 34% 54% 2% 95%

E 7% 5% 8% 13% 22% 45% 79% 95%

F 0% 7% 8% 10% 13% 31% 67% 91%

H 0% 2% 3% 7% 14% 33% 66% 91%

Q 0% 13% 14% 16% 18% 36% 72% 93%

S 0% 6% 11% 22% 29% 54% 78% 93%

NOTES:
The number in each cell represents the approximate percentage of job applicants in the normative sample who obtained
scores on that scale which were "poorer" than the job applicant's.

As an illustration of how to use this table, please note that in the normative sample, the number 10 appears in zone 3A for
the F scale. This indicates that scores in this zone are at approximately the 10th percentile (i.e., approximately 10% of the
job applicants in the normative sample obtained scores on the F scale that were "poorer" than zone 3A, or put slightly
differently, approximately 10% of the normative sample obtained F scale scores in zones 3B, 4A, or 4B).

12
 ® ®
5. MAKING USE OF ERI RESULTS 3. The seven ERI scales estimate the
likelihood that an applicant, if hired,
5.1 SOME IMPORTANT PRINCIPLES
® would perform on the job in a reliable
OF ERI USE
and productive manner. However, you
®
When making use of an applicant’s ERI should be aware that subsequent
®
results, please remember that the ERI is a changes in a person’s life and work
pre-employment questionnaire which is de- environment (both positive and nega-
signed to help employers systematically identify tive) can affect his/her performance on
reliable and productive individuals prior to the job. Supervisory knowledge of
making a hiring decision. The questionnaire employees and supervisory observation
assesses factors related to job performance of on the job behavior are commonly -
which can then be explored further during used techniques for monitoring and
interviews and reference verification. Please helping to enhance job performance,
adhere to each of the following principles when once an applicant is hired.
®
using the ERI .
4. Because it is a pre-employment ques-
1. In order to make effective and accurate tionnaire, it is recommended that the
® ®
use of an applicant’s ERI results, it is ERI be administered to job applicants
necessary that you be thoroughly immediately after they have completed
®
familiar with what each ERI scale the application for employment and
assesses, the limitations to the use of before they are interviewed or references
®
the ERI , and the manner in which the are verified. When utilized in this
®
ERI was validated. These are covered manner, the ERI® can serve as an
in Sections 2 through 9 of this Manual. objective method for assessing issues
related to job performance, which can
2. The ERI® was developed and validated then be explored further during inter-
as a pre-employment assessment tool. views and reference verification.
For that reason, it is to be used only for
the assessment of new job applicants.
Under no circumstances should the
ERI® be administered to current em-
ployees or individuals other than actual
job applicants, nor should it be used for
any purpose other than as an aid in the
pre-employment selection process.

13
 5. Under no circumstances should the 2. An applicant who is able to read and
decision to hire or not hire an applicant understand the ERI® statements may
®
be based solely on his/her ERI results. nonetheless, choose to not read the
Hiring decisions should be based on a statements carefully, or to not read the
review of ALL information collected by statements at all, before answering.
you during the applicant evaluation Under these circumstances, as well, the
process. applicant has functionally guessed
when answering the statements.
5.2 AN IMPORTANT NOTE REGARDING
SMALL DIFFERENCES IN RESULTS
Under either of these circumstances, it is
highly likely that the applicant’s results will
Because of the variability inherent in any type
® have a characteristic form to them. Typically,
of scores, small differences in ERI results
should never be used as the basis for making there will be two or more scales which have
scores in Zone 4. Routinely examining each
decisions about applicants or for comparing
applicant’s results for this pattern will provide
applicants.
you with an additional check as to the validity
5.3 REVIEWING RESULTS FOR POSSIBLE of the results. 13

PROBLEMS
If you discover that an applicant’s results have
The failure of an applicant to carefully read and
two or more scores in zone 4, it is necessary to
understand each of the 81 statements can
speak with the applicant, in order to determine
significantly affect the accuracy and usefulness
the specific cause of these scores.
of the ERI®. Invalid results may be produced
under two different sets of circumstances. If, upon inquiry, the applicant acknowledges a
difficulty in reading comprehension, you
1. Although statements are worded at a
should review the questionnaire with the
sixth grade reading level, cultural,
applicant, in order to determine the specific
ethnic, language, or educational factors
statements where this difficulty was encoun-
could theoretically contribute to reading
tered. As noted earlier, you may not tell the
comprehension difficulties for some
applicant how to answer a statement, however,
applicants. In spite of having assessed
if the applicant does not understand the
each individual applicant to determine
meaning of certain words or expressions, you
his/her ability to understand the ERI®
may explain what the word or expression
instructions and statements (as noted
means.
in Section 3.4.2), it is possible that an
applicant may have encountered read- If the applicant states that he/she was able to
ing comprehension difficulties, without understand all of the 81 statements, then you
the administrator being aware of this should urge the applicant to review his/her
fact. Under such circumstances the answers, to insure that each response is based
applicant has had to guess when an- on a careful reading and consideration of the
swering those statements which he/she statement.
did not understand.
®
Once this has been done, the ERI should then
be rescored. Our experience has been that most

13 In most samples of job applicants greaterthan 500, when the proper

administration and scoring procedures are followed, the proportion of
scores with this pattern typically falls between12% and 15%.
14
 ®
of the time, the above-described approach instruments, the ERI does not use a response
resolves the matter. However, if there are distortion scale or “faking” scale. For this
still two or more scores in Zone 4, then reason it is of interest to examine whether
caution should be used when interpreting results on any of the seven scales are signifi-
the results, unless there is corroborating cantly affected by an applicant’s attempts to
information from other sources to support “fake good”.
the hypothesis that the applicant is virtu-
ally certain to perform on the job in an This issue has been examined in some detail. 14
unreliable manner, if hired. In brief, correlation coefficients were computed
®
between each of the seven ERI scales and
5.4 “FAKING GOOD” AND three putative measures of response distortion
®
ERI RESULTS
- the 16PF Motivational Distortion scale (N=420),
and the MMPI Lie , and K scales (N=194). The
So-called “faking good” or response distortion is
results are shown in the following table. As
an issue of relevance to the use of psychologi-
can be seen, these data strongly support the
cal assessment techniques in pre-employment ®
conclusion that all seven ERI scales are free
settings. In contrast to some pre-employment
from the potentially confounding effects of
response distortion.

®
ERI
SCALE

A -.05 -.01 .10

C -.01 -.03 .11

E -.09 -.09 -.29

F .05 .04 -.03

H .01 .04 .04

Q -.05 .01 .15

S .06 .00 .15

16PF MMPI MMPI

FAKE GOOD LIE K

14 Borofsky, G. L. (1992) Assessing the likelihood of reliable

workplace behavior: Further contributions to the validation of the
Employee Reliability Inventory. Psychological Reports, 70, 563-
592.

15
 ®
5.5 USING THE ERI 5.5.1 SOME GENERAL GUIDELINES
AS A PRE-INTERVIEW FOR ASKING FOLLOW-UP
QUESTIONNAIRE QUESTIONS

As noted earlier, use of the ERI® is an objective It should be understood that the sample ques-
method for identifying issues related to job tions being referred to are only intended as
performance which may require further explo- examples of the types of questions that could
ration during interviews and reference verifica- be asked. You should ensure that the particu-
®
tion. Use of the ERI can help you to make the lar wording you choose for your questions does
most effective use of your time during inter- not violate any applicable statutory or regula-
views and reference verification by helping you tory restrictions, including the provisions of the
to selectively focus your questions in those Americans With Disabilities Act (ADA).
specific areas of reliable and productive behav-
ior which are important in the particular job In asking follow-up questions you also should
setting for which the applicant is being consid- keep the following guidelines in mind:
®
ered. Coordinated use of the ERI, with fo-
• Questions should not be accusatory.
cused inquiries during interviews and reference
verification, should greatly increase the likeli- • Questions should be limited to content
hood of your identifying reliable and productive areas which are relevant and necessary
individuals prior to making a hiring decision
. for the performance of the specific job for
which the applicant is applying.
If an applicant’s score on one or more of the
®
ERI scales suggests a possible problem area, • There should be a direct relationship
the applicant can be questioned in greater between the information being sought
detail than usual about their past record of on-
and the specific requirements of the job
the-job performance, in the specific area(s) of
for which the applicant is applying. For
behavior where a question has been raised by
®
example, you should not ask questions
the ERI results. In similar fashion, questions
dealing with religious beliefs, or affilia-
asked of past employers can focus in greater tions, racial matters, sexual behavior,
detail than usual on the applicant’s job perfor-
political beliefs and affiliations, or
mance in the specific area(s) of behavior where
beliefs or opinions regarding unions or
a question has been raised by the results.
labor organizations.

• Questions should be non-discriminatory

and should not be used to disclose a
disability, its nature, or its extent.

This issue is given greater coverage in the

Technical Bulletin, Use of the Employee Reli-
®
ability Inventory (ERI ) as a Pre-Interview
Questionnaire, and in the Manual which accom-
®
panies the ERI Americans With Disabilities
(ADA) Kit.

16
5.6 TECHNICAL SUPPORT

If you have questions about any aspect of ERI In the case of the initial version of the ERI®, the
administration, scoring, interpretation, use, or relevant criterion behaviors were determined to
if you would like to discuss the interpretation be:
®
of a specific applicant’s ERI results, with a
®
member of our staff, please call ERI Technical 1. Impaired on-the-job performanceas a
Support at the numbers listed below. consequence of the person’s pattern of
alcohol or illegal drug use.

PSYCHOMETRICS CANADA 2. Having been found guilty of a theft

1-800-661-5158 offense.

3. Absence of a history of unreliable and

unproductive behavior - that is, no
6. CONSTRUCTION AND VALIDATION OF history of impaired on-the-job perfor-
THE ERI ®
mance as a consequence of the person’s
6.1 ORIGINAL CONSTRUCTION OF THE pattern of alcohol or illegal drug use,
ERI ® and no history of having been found
guilty of theft offenses.
The ERI® was constructed using a criterion-
based methodology for attempting to predict The next step was the writing of an initial pool
the likelihood of the target behaviors, rather of over 500 “True - False” type statements
than attempting to predict them indirectly, which, it was believed, would be likely to
through the assessment of personality traits, differentiate reliable and productive individuals
values, beliefs, tendencies or attitudes. from those who were unreliable and unproduc-
tive.
The first step in the criterion-based method is
the identification of groups of individuals who
have actually exhibited the criterion or target
behaviors that are to be assessed. The Uniform
Guidelines on Employee Selection Procedures
, 15
(Section 14b(3)) indicate that the specific
criterion behaviors being assessed should
represent “important or critical work
behavior(s) or work outcomes.” Several of the
specific examples contained in the Uniform
Guidelines represent common forms of unreli-
able/unproductive behavior.

15
Bureau of National Affairs, Inc., Uniform Guidelines on Employee
Selection Procedures. (Washington, D.C.: BNA Education Systems,
1979), pp.34

17
This initial form of the ERI ® was administered structs. For this reason, the construct-related
to individuals in each of the three criterion approach to validation was not considered to
groups described above. Using Discriminant be a preferred method for validating the ERI ®.
Function Analysis, with the stepwise minimiza-
tion of residuals method, contained in Statisti- It should be noted however, that factor analy-
cal Package For The Social Sciences (SPSS ),
X 16 sis of each of the seven (7) ERI® scales reveals
answers were analyzed to determine which of the presence of an underlying structure to
the over 500 statements were most effective in each scale.18 For each scale, this underlying
differentiating subjects whose behavior was structure is discernible in the form of psycho-
reliable and productive from those whose logical constructs.
behavior was unproductive or unreliable. By
6.2.2 CONTENT-RELATED VALIDITY
this procedure, the number of items in the AND THE ERI®
ERI® was reduced to 81.
Content-related Validity refers to the degree
6.2 ORIGINAL VALIDATION OF THE to which the content of the individual items in
ERI®
a questionnaire or test are representative of
The Uniform Guidelines on Employee Selection what the questionnaire is purporting to assess.
Procedures and accepted scientific standards 17 In the case of the ERI ®, the issue is whether
require that questionnaires such as the ERI® the content of the individual items in the ERI ®
demonstrate what is known as validity. That is representative of, or related to, the predic-
is, there must be scientifically sound evidence tion of reliable and productive behavior. Con-
that the questionnaire actually measures what tent-related validity is usually most relevant
it claims to measure. Approaches to validation during the initial development of a test or
typically fall into one of three categories: questionnaire.
1) Construct-Related Validity; 2) Content-
Typically, judgments of content-related validity
Related Validity; and 3) Criterion-Related
are made on the basis of expert judgments as
Validity.
to what is the appropriate content for predict-
6.2.1 CONSTRUCT-RELATED VALIDITY ing the specific behavior being assessed. Dur-
AND THE ERI® ing the initial development of the ERI®, an
initial pool of over 500 statements was created,
Construct-Related Validity refers primarily to
which, it was believed, were related to the
the assessment of a particular concept or
specific aspects of reliable and productive
construct. Examples of psychological con-
behavior that were being studied.
structs are: need for achievement, affiliative
drive, self esteem, locus of control, and time 6.2.3 CRITERION-RELATED VALIDITY
urgency. As noted earlier, the ERI® does not AND THE ERI®
use the measurement of such constructs in
Criterion-related Validity refers to the degree
assessing job applicants. The ERI® directly
to which a questionnaire or test is able to
assesses the likelihood of specific behaviors,
accurately assess individuals with respect to a
rather than indirectly inferring their likelihood
specific criterion behavior such as reliable or
from the assessment of psychological con-
productive behavior. More specifically, the
ERI® could be said to have demonstrated
16 SPSS. (Norman H. Nie, ed.) SPSSX User’s Guide. Chicago: 1983
18 Borofsky, G. L. (1992) A preliminary investigation into the
17 American Psychological Association, Standards For Educational And structure of reliable and productive workplace behavior: Factor
analysis of the Employee Reliability Inventory. Boston, MA:
Psychological Testing, (Washington, D.C.: Author, 1985). Bay State Psychological Associates.
18
criterion-related validity if it could be shown to This criterion group of subjects was
be empirically accurate in differentiating administered the ERI® at the outset of
reliable and productive individuals from those their admission to a private hospital.
who are not. All subjects in this group were patients
on one of the alcohol/substance treat-
The most common method of demonstrating
ment units. These individuals had been
criterion-related validity is through the use of a
hospitalized as a result of their being
variety of comparison or criterion groups, each
unable to perform effectively on the job,
of which manifests one or more of the specific
as a consequence of their pattern of
behaviors of interest.
alcohol or illegal drug use.

The initial step in the validation of the ERI®

2. Individuals found guilty of theft of-
was to determine whether or not it was techni-
fenses.
cally feasible (as defined in Section 16U of the
Uniform Guidelines) to conduct criterion-related
This criterion group of subjects was
validity studies. This was done following the
administered the ERI® at a Municipal
procedures contained in Section 14b(1) of the
Court. The procedure was to adminis-
Uniform Guidelines. Using these procedures, it
ter the ERI® immediately after there was
was determined that it was, in fact, technically
a finding of guilt made by the presiding
feasible to utilize the criterion-related method
judge.
for studying the validity of the ERI® .

The results of a number of criterion-related 3. Individuals with no history of disrupted

validation and cross-validation studies are productivity as a consequence of their
reported here, and in Section 7, below. pattern of alcohol or illegal drug use,
and no history of having been found
6.3 R SCALE
guilty of theft offenses.

The original form of the ERI® contained one

For the construction and validation of
scale, which was designated as the R scale.
the R Scale, the subjects in this crite-
The R scale was intended to provide an overall
rion group were administered the
measure of the likelihood that an applicant
questionnaire at a variety of locations.
would perform on-the-job in a reliable and
In all cases the absence of a history of
productive manner.
production deviant behavior as a conse-
6.3.1 CRITERION GROUPS quence of their pattern of alcohol or
illegal drug use and the absence of a
Three criterion groups were used in the devel- history of theft offense convictions was
opment, validation and cross-validation of the confirmed by the subjects’ answers to a
R Scale. These were: questionnaire designed to evaluate these
factors.
1. Individuals who had been unable to
perform effectively on the job as a
For the cross-validation study reported
consequence of their pattern of alcohol
below, the subjects in this criterion
or illegal drug use, and who required
treatment for this condition.

19
 group were administered the ERI ® as curred by chance. This far exceeds the level of
one part of the pre-employment selec- statistical significance (one in twenty) recom-
tion process used by a restaurant mended in Section 14B(5) of the Uniform
chain. In all cases, the absence of a Guidelines.
history of production deviant behavior
was confirmed by the subjects’ answers Where the group membership of each indi-
to questions during the conduct of the vidual (reliable or unreliable) was “blindly”
company’s regular application interview. classified, based on the individuals’ R scale
In addition, the subjects in this group scores, the ERI ® correctly classified group
had been rated by their respective membership in 94% of the cases.
restaurant managers as being among
This result indicates that the R scale was
the top 50% of the restaurant’s employ-
effective in differentiating reliable and produc-
ees, in terms of their actual perfor-
tive individuals from those who are unreliable.
mance on the job.
6.3.3 CROSS-VALIDATION
6.3.2 VALIDATION
To ascertain whether or not there is an “over-
Individuals in criterion groups (1) and (2) statement” of a procedure’s validity, Section
above, were pooled together and operationally 14B(7) of the Uniform Guidelines recommends
defined as unreliable, while individuals in the use of a cross-validation methodology.
criterion group (3) above, were operationally
defined as being reliable. Cross-validation involves conducting a second
study using different individuals. The purpose
Sample sizes for this validation study were as is to determine if the discriminant function
follows: derived in the validation study continues to
differentiate the criterion groups from each
Total Sample Size: N = 117
other, when new groups of subjects are as-
Reliable Subjects: N = 38
sessed.
Unreliable Subjects: N = 79
For the cross-validation of the R scale, the
Discriminant Function Analysis, with the subjects in each criterion group were selected
stepwise minimization of residuals method in the manner described earlier. The specific
X
contained in SPSS , was utilized in this analy- subjects in this study, however, were different
sis. individuals from those participating in the
original validation study.
For the discriminant function derived in this
study, the canonical correlation coefficient was Sample sizes for the cross-validation study
0.9677, the Wilks’ Lambda value was 0.0636, were as follows:
and the Chi Square value was 181.845, with 38
Total Sample Size: N = 77
degrees of freedom. This is statistically signifi-
Reliable Subjects: N = 38
cant at the p < 0.00001 level of significance.
Unreliable Subjects: N = 39

That is, this result has a probability of less than

one in one hundred thousand of having oc-

20
In conformance with standard approaches to be reliable, in that his/her performance will not
cross-validation, the discriminant function be disrupted by behaviors such as inattentive-
weights derived during the original validation ness, unauthorized absence/lateness, failing to
of the R scale were used to score each ERI ® in follow through on assignments, or other inap-
this study. propriate work behaviors. It is not designed to
assess the extent of prior or current alcohol or
Where the group membership of each indi- illegal drug use. Similarly, it is not designed to
vidual (reliable or unreliable) was “blindly” reveal, nor should it be used for the purpose of
classified, based on the individuals’ R scale revealing, the existence, nature, or severity of a
scores, the ERI ® correctly classified group disability.
membership in 90% of the cases.
7.1.1 CRITERION GROUPS
This result indicates that even when the R
scale was subjected to cross-validation, it was For both the validation and cross-validation of
found to be effective in differentiating reliable the A scale, the following criterion groups were
individuals from those who are unreliable. used:

7. SUBSEQUENT DEVELOPMENT OF 1. Individuals who had been unable to

THE ERI® perform effectively on the job as a
As noted above, the original R scale was devel- consequence of their pattern of alcohol
oped and validated as an overall measure of the or illegal drug use, and who required
likelihood that an applicant would perform on- treatment for this condition.
the-job in a reliable and productive manner.
Over time, however, increased understanding of This criterion group of subjects was
the various specific causes of unreliable and administered the ERI ® at the outset of
unproductive behavior, as well as an increased their admission to a private hospital.
understanding of the multifactorial nature of All subjects in this group were patients
workplace behavior, led to the conclusion that in on one of the alcohol/substance treat-
order to be most useful to employers, the ERI® ment units. These individuals had been
should be capable of performing a com- hospitalized as a result of their being
prehensive (i.e., multifactorial) assessment of job unable to perform effectively on the job,
applicants. as a consequence of their pattern of
alcohol or illegal drug use.
In order to accomplish this goal, Psychometrics
Canada initiated an ongoing program of research.
2. Individuals with no history of production
To date, the results of this research have yielded
deviant behavior as a consequence of
the seven (7) scales currently contained in the
their pattern of alcohol or illegal drug
ERI® . These are the A, C, E, F, H, Q and S19
use.
scales described earlier in this Manual.

7.1 A SCALE: CONSTRUCTION AND Each member of this criterion group was
VALIDATION a job applicant who possessed a current
security clearance, enabling him/her to
The A scale was designed to assess the likeli-
have access to information classified Top
hood that an applicant’s work performance will

19 In the current version, the R scale is no longer used.

21
 Secret. In order to gain this level of Where the group membership of each individual
security clearance, each individual was (reliable or unreliable) was “blindly” classified,
subjected to, and successfully passed, a based on the individuals’ A scale scores, the
full field background investigation ERI® correctly classified group membership in
which included, among other things, 99% of the cases.
investigation of prior patterns of unreli-
This result indicates that theA scale was effec-
able / unproductive behavior. The
tive in differentiating reliable and productive
ERI® was administered as part of the
individuals from those whose on-the job perfor-
pre-employment processing of each
mance was impaired as a consequence of their
individual.
pattern of alcohol or illegal drug use.

7.1.2 VALIDATION 7.1.3 CROSS-VALIDATION

Individuals in criterion group (1) above, were For the cross-validation of theA scale, the
operationally defined as being unreliable, while subjects in each criterion group were selected
individuals in criterion group (2) above, were in the manner described earlier. The specific
operationally defined as being reliable. subjects in this study, however, were different
individuals from those participating in the
Sample sizes for this validation study were as
original validation study of the A Scale.
follows:
Sample sizes for the cross-validation
Total Sample Size: N = 111
study were as follows:
Unreliable Subjects: N = 53
Reliable Subjects: N = 58 Total Sample Size: N = 44
Unreliable Subjects: N = 29
Discriminant Function Analysis, with the
Reliable Subjects: N = 15
stepwise minimization of residuals method
X
contained in SPSS , was utilized in this analy- In conformance with standard approaches to
sis. cross-validation, the discriminant function
weights derived during the original validation of
For the discriminant function derived in this
the A scale were used to score each ERI® in
study, the canonical correlation coefficient was
this study.
0.9397, the Wilks’ Lambda value was 0.1170,
and the Chi Square value was 189.851, with 27 Where the group membership of each indi-
degrees of freedom. This is statistically signifi- vidual (reliable or unreliable) was “blindly”
cant at the p < 0.00001 level of significance. classified, based on the individuals’ A scale
That is, this result has a probability of less scores, the ERI® correctly classified group
than one in one hundred thousand of having membership in 84% of the cases.
occurred by chance. This far exceeds the level
of statistical significance (one in twenty) recom- This result indicates that even when theA scale
mended in Section 14B(5) of theUniform Guide- was subjected to cross-validation, it was found
lines. to be effective in differentiating reliable and
productive individuals from those whose on-the

22
job performance was impaired as a conse- • Smiles
quence of their pattern of alcohol or illegal drug
• Consistently makes eye contact as part
use.
of their communication skills
7.2 C SCALE: CONSTRUCTION AND
VALIDATION For both the validation and cross-validation of
the C scale, the following criterion groups were
The C scale was designed to assess applicants used:
with respect to the likelihood that their interac-
tions with customers/guests will be character- 1. Individuals who were rated as being the
ized by a high level of courtesy and commit- best in each Department, in terms of
ment to service. meeting the above definition of courte-
7.2.1 CRITERION GROUPS ous behavior.

“Front of the house” employees (i.e., employees 2. Individuals who were rated as being the
having direct face to face contact with guests as poorest in each Department, in terms
part of their duties) in 32 different departments of meeting the above definition of
of an East Coast resort hotel were used to courteous behavior.
construct two criterion groups. Prior to being
7.2.2 VALIDATION
hired, individuals in each criterion group had
®
been administered the ERI as part of the pre- Individuals in criterion group (1) above, were
employment selection process. operationally defined as being more courteous
Each individual was rated by his/her supervi- in their behavior, while individuals in criterion
sor as to how courteous he/she was when group (2) above, were operationally defined as
interacting with guests. The eight point defini- being less courteous.
tion of courteous behavior, shown below, was
Sample sizes for the validation study were as
used for this purpose.
follows:
• Demonstrates courtesy, constant
politeness and a positive attitude Total Sample Size: N = 112
toward guests More Courteous Subjects: N = 81
Less Courteous Subjects: N = 31
• Presents a genuinely friendly outgoing
manner
Discriminant Function Analysis, with the
• Initiates communication with guests stepwise minimization of residuals method
X
through greetings and cordial conversa- contained in SPSS , was utilized in this analy-
tion sis.

• Remains courteous even during difficult For the discriminant function derived in this
confrontations with guests study, the canonical correlation coefficient was
0.8928 the Wilks’ Lambda value was 0.2030,
• Demonstrates an awareness of in-
and the Chi Square value was 147.50, with 35
house activities and knows the physical
degrees of freedom. This is statistically signifi-
location of public areas
cant at the p < .00001 level of significance.
• Provides service throughout the entire That is, this result has a probability of less
interaction with a guest

23
than one in one hundred thousand of having was subjected to cross-validation, it was found
occurred by chance. This far exceeds the level to be effective in differentiating individuals who
of statistical significance (one in twenty) rec- performed on the job in a courteous manner,
ommended in Section 14B(5) of the Uniform from those who did not.
Guidelines.
7.3 E SCALE: CONSTRUCTION AND
VALIDATION
Where the group membership of each indi-
vidual (more courteous or less courteous) was The E scale was designed to assess the likeli-
“blindly” classified, based on the individuals’ C
hood that an applicant’s work performance will
scale scores, the ERI ® correctly classified
be characterized by emotionally mature behav-
group membership in 98% of the cases.
ior, and that it will not be disrupted by the
presence of maladaptive personality character-
This result indicates that the C scale was
istics or traits such as irresponsibility, poor
effective in differentiating individuals who
judgment, difficulty in working cooperatively
performed on the job in a courteous manner,
with others, or poor impulse control.
from those who did not.

7.2.3 CROSS-VALIDATION It is important to emphasize that this scale is

not intended to assess, nor does it in fact
For the cross-validation of the C scale, the assess, for the presence of mental or psycho-
subjects in each criterion group were selected logical impairment or disorder, or an
in the manner described earlier. The specific applicant’s general physical or psychological
subjects in this study, however, were different health.20
individuals from those participating in the
7.3.1 CRITERION GROUPS
original validation study of the C Scale.

For both the validation and cross-validation of

Sample sizes for the cross-validation study
the E scale, the following criterion groups were
were as follows:
used:
Total Sample Size: N = 16
1. Individuals who had been unable to
More Courteous Subjects: N = 14
perform effectively on the job and in
Less Courteous Subjects: N= 2
their personal lives due to the presence
In conformance with standard approaches to of maladaptive personality traits.
cross-validation, the discriminant function
This criterion group of subjects was
weights derived during the original validation of ®
administered the ERI at the outset of
the C scale were used to score each ERI® in
their admission to a private hospital.
this study.
The admission note for each person was
Where the group membership of each indi- reviewed, to insure that no individual
vidual (more courteous or less courteous) was with an Axis I disorder would be in-
“blindly” classified, based on the individuals’C cluded in the group.
scale scores, the ERI® correctly classified group
2. Individuals with no history of unreli-
membership in 75% of the cases.
able/unproductive behavior due to the
This result indicates that even when the C scale 20 Borofsky, G. L. (1992) Assessing the likelihood of reliable
workplace behavior: Further contributions to the validation of the
Employee Reliability Inventory. Psychological Reports, 70, 563-
592.
24
 presence of maladaptive personality
significant at the p < 0.00001 level of signifi-
traits.
cance. That is, this result has a probability of
less than one in one hundred thousand of
Each member of this criterion group having occurred by chance. This far exceeds
was a job applicant who possessed a the level of statistical significance (one in
current security clearance, enabling twenty) recommended in Section 14B(5) of the
him/her to have access to information Uniform Guidelines.
classified Top Secret. In order to gain
this level of security clearance, each Where the group membership of each indi-
individual was subjected to, and suc- vidual (reliable or unreliable) was “blindly”
cessfully passed, a full field background classified, based on the individuals’ E scale
investigation, which included, among scores, the ERI® correctly classified group
other things, investigation of prior membership in 92% of the cases.
patterns of unreliable/unproductive
® This result indicates that the E scale was
behavior. The ERI was administered
effective in differentiating reliable and produc-
as part of the pre-employment process-
tive individuals from those who were unable to
ing of each individual.
perform effectively due to the presence of
maladaptive personality traits.
7.3.2 VALIDATION
7.3.3 CROSS-VALIDATION
Individuals in criterion group (1) above, were
operationally defined as being unreliable, while For the cross-validation of the E scale, the
individuals in criterion group (2) above, were subjects in each criterion group were selected
operationally defined as being reliable. in the manner described earlier. The specific
subjects in this study, however, were different
Sample sizes for this validation study individuals from those participating in the
were as follows: original validation study of the E Scale.

Total Sample Size: N = 91 Sample sizes for the cross-validation

Unreliable Subjects: N = 35 study were as follows:
Reliable Subjects: N = 56
Total Sample Size: N = 31
Discriminant Function Analysis, with the Unreliable Subjects: N = 14
stepwise minimization of residuals method Reliable Subjects: N = 17
X
contained in SPSS , was utilized in this analy-
sis. In conformance with standard approaches to
cross-validation, the discriminant function
For the discriminant function derived in this weights derived during the original validation of
study, the canonical correlation coefficient was the E scale were used to score each ERI® in
0.9368, the Wilks’ Lambda value was 0.1224, this study.
and the Chi Square value was 144.954, with
22 degrees of freedom. This is statistically Where the group membership of each indi-
vidual (reliable or unreliable) was “blindly”
classified, based on the individuals’ E scale

259
scores, the ERI ® correctly classified group Sample sizes for this validation study
membership in 84% of the cases. were as follows:

This result indicates that even when the Total Sample Size: N = 87
E scale was subjected to cross-validation, it Unreliable Subjects: N = 10
was found to be effective in differentiating Reliable Subjects: N = 77
reliable and productive individuals from those
who were unable to perform effectively due to Discriminant Function Analysis, with the
the presence of maladaptive personality traits. stepwise minimization of residuals method
X
contained in SPSS , was utilized in this analy-
7.4 F SCALE: CONSTRUCTION AND sis.
VALIDATION

For the discriminant function derived in this

The F scale was designed to assess individuals
study, the canonical correlation coefficient was
with respect to one component of unplanned
0.7544, the Wilks’ Lambda value was 0.4309,
and uncontrolled turnover. The F scale as-
and the Chi Square value was 63.140, with 20
sesses the likelihood that a job applicant will
degrees of freedom. This is statistically signifi-
perform in a conscientious and reliable man-
cant at the p < 0.00001 level of significance.
ner, will not be fired, and will remain on the job
That is, this result has a probability of less
for at least 30 days.
than one in one hundred thousand of having
7.4.1 CRITERION GROUPS occurred by chance. This far exceeds the level
of statistical significance (one in twenty) rec-
For both the validation and cross-validation of ommended in Section 14B(5) of the Uniform
the F scale, the following criterion groups were Guidelines.
used:
Where the group membership of each indi-
1. Individuals who had been fired from vidual (reliable or unreliable) was “blindly”
their job within thirty (30) days of being classified, based on the individuals’ F scale
hired. scores, the ERI® correctly classified group
membership in 95% of the cases.
2. Individuals who neither quit nor were
fired from their job within thirty (30) This result indicates that the F scale was
days of being hired; that is, they had effective in differentiating reliable and produc-
worked at their job for more than thirty tive individuals from those who were fired from
(30) days. their job within 30 days of being hired.

Prior to being hired, each individual was admin- 7.4.3 CROSS-VALIDATION

istered the ERI® as part of the pre-employment
selection process. For the cross-validation of the F scale, the
subjects in each criterion group were selected
7.4.2 VALIDATION in the manner described earlier. The specific
subjects in this study, however, were different
Individuals in criterion group (1) above, were
individuals from those participating in the
operationally defined as being unreliable, while
original validation study of the F Scale.
individuals in criterion group (2) above, were
operationally defined as being reliable.

26 60
Sample sizes for the cross-validation ter the ERI® immediately after there
study were as follows: was a finding of guilt made by the
presiding judge.
Total Sample Size: N = 102
Unreliable Subjects: N=9
2. Individuals with no history of having
Reliable Subjects: N = 93 been found guilty of theft offenses.

In conformance with standard approaches to

cross-validation, the discriminant function Each member of this criterion group

weights derived during the original validation of was a job applicant who possessed a
the F scale were used to score each ERI® in current security clearance, enabling
this study. him/her to have access to information
classified Top Secret. In order to gain
Where the group membership of each indi- this level of security clearance, each
vidual (reliable or unreliable) was “blindly” individual was subjected to, and suc-
classified, based on the individuals’ F scale cessfully passed, a full field background
scores, the ERI® correctly classified group investigation, which included, among
membership in 84% of the cases. other things, investigation of prior
patterns of unreliable/unproductive
This result indicates that even when the
behavior. The ERI® was administered
F scale was subjected to cross-validation, it
as part of the pre-employment process-
was found to be effective in differentiating
ing of each individual.
reliable and productive individuals from those
who were fired from their job within thirty (30)
7.5.2 VALIDATION
days of being hired.
Individuals in criterion group (1) above, were
7.5 H SCALE: CONSTRUCTION AND
VALIDATION operationally defined as being unreliable, while
individuals in criterion group (2) above, were
The H scale was designed to assess the likeli- operationally defined as being reliable.
hood that an applicant will perform in a trust-
worthy manner, and will not engage in various Sample sizes for this validation study
forms of property deviant behavior. were as follows:

7.5.1 CRITERION GROUPS Total Sample Size: N = 73

Unreliable Subjects: N = 19
For both the validation and cross-validation of
Reliable Subjects: N = 54
the H scale, the following criterion groups were
used: Discriminant Function Analysis, with the
stepwise minimization of residuals method
1. Individuals who had been found guilty of X
contained in SPSS , was utilized in this analy-
theft offenses.
sis.

This criterion group of subjects was For the discriminant function derived in this
administered the ERI® at a Municipal study, the canonical correlation coefficient was
Court. The procedure was to adminis- 0.9903, the Wilks’ Lambda value was 0.0194,

271
and the Chi Square value was 179.404, with (reliable or unreliable) was “blindly” classified,
31 degrees of freedom. This is statistically based on the individuals’H scale scores, the
significant at the p < 0.00001 level of signifi- ERI® correctly classified group membership in
cance. That is, this result has a probability of 90% of the cases.
less than one in one hundred thousand of
having occurred by chance. This far exceeds This result indicates that even when theH scale
the level of statistical significance (one in was subjected to cross-validation, it was found
twenty) recommended in Section 14B(5) of the to be effective in differentiating reliable and
Uniform Guidelines. productive individuals from those who were
operationally defined as unreliable, based on
Where the group membership of each indi- their documented record of property deviant
vidual (reliable or unreliable) was “blindly” behavior.
classified, based on the individuals’ H scale
7.6 Q SCALE: CONSTRUCTION AND
scores, the ERI® correctly classified group
VALIDATION
membership in 92% of the cases.
The Q scale was designed to assess individuals
This result indicates that the H scale was
with respect to a second component of un-
effective in differentiating reliable and produc-
planned and uncontrolled turnover. The Q
tive individuals from those who were operation-
scale assesses the likelihood that a job appli-
ally defined as unreliable, based on their
cant will not quit and will remain on the job for
documented record of property deviant behav- at least 30 days.
ior.
7.6.1 CRITERION GROUPS
7.5.3 CROSS-VALIDATION
For both the validation and cross-validation of
For the cross-validation of the H scale, the
the Q scale, the following criterion groups were
subjects in each criterion group were selected
used:
in the manner described earlier. The specific
subjects in this study, however, were different 1. Individuals who had quit their jobs
individuals from those participating in the within thirty (30) days of being hired.
original validation study of the H Scale.
2. Individuals who neither quit nor were
Sample sizes for the cross-validation study fired from their job within thirty (30)
were as follows: days of being hired; that is, they had
worked at their job for more than thirty
Total Sample Size: N = 29
(30) days.
Unreliable Subjects: N = 10
Reliable Subjects: N = 19 Prior to being hired, each individual was admin-
istered the ERI® as part of the pre-employment
In conformance with standard approaches to
selection process.
cross-validation, the discriminant function
weights derived during the original validation of 7.6.2 VALIDATION
the H scale were used to score each ERI® in this
Individuals in criterion group (1) above, were
study.
operationally defined as being unreliable, while
Where the group membership of each individual

28
individuals in criterion group (2) above, were subjects in this study, however, were different
operationally defined as being reliable. individuals from those participating in the
original validation study of the Q Scale.
Sample sizes for this validation study were as
follows: Sample sizes for the cross-validation study
were as follows:
Total Sample Size: N = 126
Unreliable Subjects: N = 22 Total Sample Size: N = 83
Reliable Subjects: N = 104 Unreliable Subjects: N = 17
Reliable Subjects: N = 66
Discriminant Function Analysis, with the
stepwise minimization of residuals method In conformance with standard approaches to
X
contained in SPSS , was utilized in this analy- cross-validation, the discriminant function
sis. weights derived during the original validation of
the Q scale were used to score each ERI ® in
For the discriminant function derived in this this study.
study, the canonical correlation coefficient was
0.6388, the Wilks’ Lambda value was 0.5919, Where the group membership of each indi-
and the Chi Square value was 61.358, with 14 vidual (reliable or unreliable) was “blindly”
degrees of freedom. This is statistically signifi- classified, based on the individuals’ Q scale
cant at the p < 0.00001 level of significance. scores, the ERI® correctly classified group
That is, this result has a probability of less membership in 76% of the cases.
than one in one hundred thousand of having
occurred by chance. This far exceeds the level This result indicates that even when the
of statistical significance (one in twenty) recom- Q scale was subjected to cross-validation, it
mended in Section 14B(5) of the Uniform Guide- was found to be effective in differentiating
lines. individuals who remained on the job for more
than 30 days, from those who quit their jobs
Where the group membership of each indi- within 30 days of being hired.
vidual (reliable or unreliable) was “blindly”
7.7 S SCALE: CONSTRUCTION AND
classified, based on the individuals’ Q scale
VALIDATION
scores, the ERI® correctly classified group
membership in 90% of the cases. The S scale was designed to assess individuals
with respect to one component of production
This result indicates that the Q scale was
deviance. The S scale assesses the likelihood
effective in differentiating individuals who
that a job applicant will perform on the job in a
remained on the job for more than 30 days, safe manner, and will not have a significant on-
from those who quit their jobs within 30 days of
the-job accident in the first four months of
being hired.
employment. 21

7.6.3 CROSS-VALIDATION

For the cross-validation of the Q scale, the

subjects in each criterion group were selected in
the manner described earlier. The specific 21 For present purposes, a significant on-the-job accident is
defined as one in which the costs involved exceed $300.00.

29
7.7.1 CRITERION GROUPS curred by chance. This far exceeds the level of
statistical significance (one in twenty) recom-
For both the validation and cross-validation of
mended in Section 14B(5) of the Uniform Guide-
the S scale, the following criterion groups were
lines.
used:
Where the group membership of each indi-
1. Individuals who had a significant on-
vidual (reliable or unreliable) was “blindly”
the-job accident in the first four
classified, based on the individuals’ S scale
months of employment.
scores, the ERI ® correctly classified group
membership in 100% of the cases.
2. Individuals who did not have a signifi-
cant on-the-job accident in the first four
This result indicates that the S scale was
months of employment.
effective in differentiating individuals who
performed on the job in a safe manner, from
Prior to being hired, each individual was ad-
those who had a significant on-the-job acci-
ministered the ERI® as part of the pre-employ-
dent in the first four months of employment.
ment selection process.
7.7.3 CROSS-VALIDATION
7.7.2 VALIDATION

For the cross-validation of the S scale, the

Individuals in criterion group (1) above, were
subjects in each criterion group were selected
operationally defined as being unreliable, while
in the manner described earlier. The specific
individuals in criterion group (2) above, were
subjects in this study, however, were different
operationally defined as being reliable.
individuals from those participating in the
Sample sizes for this validation study were as original validation study of the S Scale.
follows:
Sample sizes for the cross-validation study
Total Sample Size: N = 59 were as follows:
Unreliable Subjects: N = 14
Total Sample Size: N = 26
Reliable Subjects: N = 45
Unreliable Subjects: N= 5
Discriminant Function Analysis, with the Reliable Subjects: N = 21
stepwise minimization of residuals method
X In conformance with standard approaches to
contained in SPSS , was utilized in this analy-
cross-validation, the discriminant function
sis.
weights derived during the original validation of
For the discriminant function derived in this the S scale were used to score each ERI® in this
study, the canonical correlation coefficient was study.
0.9997 the Wilks’ Lambda value was 0.0005,
Where the group membership of each individual
and the Chi Square value was 253.72, with 48
(reliable or unreliable) was “blindly” classified,
degrees of freedom. This is statistically signifi-
based on the individuals’ S scale scores, the
cant at the p < .00001 level of significance.
ERI® correctly classified group membership in
That is, this result has a probability of less than
one in one hundred thousand of having oc-

30
85% of the cases. results on each scale are distributed continu-
ously along each of the behavioral-psychologi-
This result indicates that even when the S cal dimensions measured by the ERI .
®

scale was subjected to cross-validation, it was

found to be effective in differentiating individu- 8.1 DESCRIPTIVE STATISTICS
als who performed on the job in a safe manner,
To facilitate the use of these descriptive statis-
from those who had a significant on-the-job
tics when examining an individual applicant’s
accident in the first four months of employ-
results, they are reported in terms of the eight
ment.
zone system of scores, rather than the dis-
8.0 SOME PSYCHOMETRIC criminant scores. In reading this table, please
®
PROPERTIES OF THE ERI note that for the mean, median, and mode, the
values of the numbers to the left of the decimal
The data reported below are based on a group
point refer to the zone number in the eight
of job applicants (N = 60,670) who completed
® zone system shown below.
the ERI as part of their pre-employment
processing. This normative group of job appli- For example, one can see that the mean for the
cants is drawn from all regions of the country, A scale is 4.074. This indicates that the mean
represents all 10 Standard Industrial classifi- for this scale is in Zone 2B, but only slightly
cation (SIC) Code Divisions, 54 Major SIC over the line from zone 2A (seven-one hun-
Groups, and a wide range of job categories. As dredths). Similarly, the mean for the
can be seen in Section 8.2 below, applicants’ Q scale
Q is 3.572. This means that it is roughly

RELIABILITY

ZONE 4 ZONE 3 ZONE 2 ZONE 1

B A B A B A B A

A C E F H Q S

Mean 4.074 4.041 3.678 3.274 3.159 3.572 3.937

Median 4.000 4.000 3.000 3.000 3.000 3.000 4.000
Mode 4.000 3.000 3.000 3.000 3.000 3.000 4.000

Standard Deviation 1.584 1.750 1.650 1.698 1.440 1.949 1.816

Variance 2.509 3.061 2.723 2.884 2.074 3.800 3.297
Standard Error 0.006 0.007 0.007 0.007 0.008 0.008 0.007

31
six-tenths of the way through zone 2A. The 8.2 FREQUENCY DISTRIBUTIONS
FOR ERI ® SCALES
median E scale value of 3.000 means that the
median falls in zone 2A.
The frequency distributions shown below are
reported in terms of discriminant scores. The
Likewise, the standard deviation of the A scale
frequency and percentile distributions for the
is 1.584. This means that the standard devia-
ERI® scales, using the eight zone system, are
tion for this scale is roughly 1.6 zones of the
reported in Section 4.2 of this Manual.
eight zone system.

A Scale C Scale

E Scale F Scale

H Scale Q Scale

S Scale

32
8.3 INTERCORRELATION OF THE
®
ERI SCALES 22,23

As dis cussed in Section 1.2.3 of this Manual, below appears to provide some support for the
the correlation matrix of ERI ® scales shown multifactorial view of behavior.

SCALE

A 1.00

C -0.05 1.00

E -0.35 -0.18 1.00

F 0.29 -0.00 -0.13 1.00

H 0.41 0.27 -0.23 0.35 1.00

Q 0.10 -0.11 0.09 0.07 0.05 1.00

S 0.19 0.25 -0.25 -0.01 0.13 -0.17 1.00

A C E F H Q S

8.4. TEST-RETEST RELIABILITY OF These results demonstrate that the scales of

THE ERI® SCALES the ERI® showed an acceptable degree of test-
retest reliability. In all cases, the degree of
Test-retest reliability is the term used to de-
test-retest reliability was statistically signifi-
scribe the consistency of an individual’s results
cant at the p < 0.0001 level. That is, this
over the course of separate administrations of a
result has a probability of less than one in one
questionnaire or test. A procedure is consid-
hundred thousand of having occurred by
ered to have good test-retest reliability if it gives
chance. This far exceeds the level of statisti-
roughly the same score or results for an indi-
cal significance (one in twenty) recommended
vidual each time. With specific reference to the
in Section 14B(5) of the Uniform Guidelines.
scales on the ERI®, test-retest reliability pro-
vides a measure of whether a given individual’s
ERI® results consistently reflect his or her
22 The apparently inverse relationships between some of the
actual likelihood of reliable behavior or whether
scales and the E & Q scales is an artifact of the scoring weights
the ERI® results from each administration are assigned by the discriminant function. In reality, these
simply due to random variation, random error, relationships are positive ones, as would be expected.
23 Some of the correlation coefficients appear to be signifi-
or transient fluctuations in mood. cant, given the normal null hypothesis assumption of zero
association between variables. However, it can be argued that
such an assumption is inappropriate when working with
Using the Pearson product-moment correlation
behavioral variables, such as those being assessed here, since
coefficient (r), the test-retest reliability of each there is invariably some degree of association between variables
which assess specific aspects of human behavior. In any event,
of the ERI® scales was computed, with the even in the case of the largest correlation above (that between
results shown on the following page. the A and H scales), the coefficient of determination (r 2) is less
than 0.25. Accordingly, these data are taken as providing
support for the multifactorial perspective described earlier.

33
 SAMPLE SIZE: N = 30
TEST - RETEST INTERVAL: 7 to 21 Days

A r = 0.89 p<.01

C r = 0.68 p<.01

E r = 0.77 p<.01

F r = 0.75 p<.01

H r = 0.73 p<.01

Q r = 0.85 p<.01

S r = 0.83 p<.01

relative selection rates and impact ratios have

been compared for whites, blacks, “other races”,
9. FAIRNESS OF THE ERI® AND
and persons of Hispanic origin. A typical set of
ADVERSE IMPACT
results is shown on the following page.
Research has also been conducted to ascertain
TOTAL SAMPLE SIZE: N = 1350
if use of the ERI ® results in adverse impact as
Whites N = 800
defined in Section 4D of the Uniform Guide-
Blacks: N = 400
lines. This research has examined the relative
Other Races: N = 150
selection rates and impact ratios for each of
Hispanic Origin: N = 104
the seven ERI® scales, over a wide range of
industry types and job categories, in terms of
Using this method, for each of the seven ERI ®
race, 24 gender, and age.
scales, it has also been consistently found that
the impact ratios conform to the requirements
It should be noted that in order to conduct
of the “four-fifths rule of thumb” contained in
these analyses, two basically unacceptable
the Uniform Guidelines. On this basis, it has
assumptions must be made - neither of which
® also been concluded that use of the ERI ® does
occur in the actual course of using the ERI as
not result in adverse impact with respect to
part of the selection process: (1) A fixed cut off
race.
score must be set for each of the seven ERI ®
®
scales; and (2) Each of the applicant’s ERI
scale scores must be considered as the sole
basis on which a selection decision is made.
24 For purposes of categorizing the data, four (4) racial
groupings are used: White, Black, Other races, and Hispanic
9.1 RACE origin. This classification system was chosen because it is the
one used by the Department of Commerce, Bureau of the Census.
Using the method described in theUniform It should be noted that persons of Hispanic origin may be of any
race.
Guidelines on Employee Selection Procedures, the
34
SCALE IMPACT RATIOS %

BLACK / WHITE 83%

OTHER RACES / WHITE 96%
A WHITE / HISPANIC ORIGIN 92%
BLACK / HISPANIC ORIGIN 76%
BLACK / OTHER RACES 87%

BLACK / WHITE 94%

OTHER RACES / WHITE 100%
C HISPANIC ORIGIN / WHITE 95%
BLACK / HISPANIC ORIGIN 89%
BLACK / OTHER RACES 94%

BLACK / WHITE 94%

OTHER RACES / WHITE 87%
E HISPANIC ORIGIN / WHITE 89%
HISPANIC ORIGIN / BLACK 94%
OTHER RACES / BLACK 92%

WHITE / BLACK 96%

OTHER RACES / WHITE 100%
F WHITE / HISPANIC ORIGIN 99%
HISPANIC ORIGIN / BLACK 97%
OTHER RACES / BLACK 96%

BLACK / WHITE 100%

OTHER RACES / WHITE 89%
H HISPANIC ORIGIN / WHITE 93%
HISPANIC ORIGIN / BLACK 93%
OTHER RACES / BLACK 89%

BLACK / WHITE 94%

OTHER RACES / WHITE 96%
Q HISPANIC ORIGIN / WHITE 99%
BLACK / HISPANIC ORIGIN 95%
BLACK / OTHER RACES 98%

BLACK / WHITE 97%

OTHER RACES / WHITE 97%
S WHITE / HISPANIC ORIGIN 98%
BLACK / HISPANIC ORIGIN 95%
BLACK / OTHER RACES 100%

35
9.2 GENDER TOTAL SAMPLE SIZE:N = 1350
UNDER 40 N = 1060
The relative selection rates and impact ratios
40 AND OLDER N = 290
have also been compared for females and males
using the above-described method. The results ®
Using this method, for each of the seven ERI
for the same sample described in the previous
scales, it has also been consistently found that
section are shown below.
the impact ratios conform to the requirements
of the “four-fifths rule of thumb” contained in
Total sample size: N = 1350
the Uniform Guidelines. On this basis, it has
Males N = 899
also been concluded that use of the ERI ® does
Females N = 451
not result in adverse impact with respect to age.

Using this method, for each of the seven ERI ®

SCALE IMPACT RATIOS %
scales, it has also been consistently found that
the impact ratios conform to the requirements A > = 40 / < 40 80%
of the “four-fifths rule of thumb” contained in
C < 40 / > = 40 94%
the Uniform Guidelines. On this basis, it has
also been concluded that use of the ERI ® does E > = 40 / < 40 92%
not result in adverse impact with respect to
gender.
F > = 40 / < 40 96%

H > = 40 / < 40 97%

SCALE IMPACT RATIOS % Q > = 40 / < 40 96%

A FEMALE / MALE 87% S > = 40 / < 40 96%

C FEMALE / MALE 93%

9.4 SUMMARY
E MALE / FEMALE 93%

F MALE / FEMALE 93% In summary, when comparing the relative

selection rates and impact ratios for each of
H MALE / FEMALE 93% the seven ERI® scales, over a wide range of
industry types and job categories, it has been
Q MALE / FEMALE 98%
consistently found that use of the ERI ® does
S FEMALE / MALE 93% not result in adverse impact with respect to
race, gender or age.
9.3 AGE

The relative selection rates and impact ratios

have also been compared for individuals
younger and older than forty (40) years of age
and males using the above-described method.
The results for the same sample described in
Section 9.1 are shown in the following table.

36
10. SELECTED REFERENCES

Borofsky, G. L., (1991) A preliminary investigation into the structure of reliable and productive
workplace behavior: factor analysis of the Employee Reliability Inventory. Boston, MA: Bay
State Psychological Associates.
Borofsky, G. L., (1992a) Assessing the likelihood of reliable workplace behavior: further contribu-
tions to the validation of the Employee Reliability Inventory. Psychological Reports, 70, 563-
592.
®
Borofsky, G. L. (1992b) Psychometric properties of the Employee Reliability Inventory (ERI ).
Boston, MA: Bay State Psychological Associates.
Borofsky, G. L. (1992c) Americans With Disabilities Act user’s manual addendum. Boston, MA:
Bay State Psychological Associates.
Borofsky, G. L. (1992d) Training Syllabus for the Employee Reliability Inventory . Boston, MA:
Bay State Psychological Associates.
Borofsky, G. L. (1992e) Training Guide for administering and scoring the Employee Reliability
Inventory. Boston, MA: Bay State Psychological Associates.
Borofsky, G. L. (1992f) Training Guide for interpreting and using the Employee Reliability Inven-
tory. Boston, MA: Bay State Psychological Associates.
Borofsky, G. L. (1994a) User’s manual for the Employee Reliability Inventory (Rev. 6.2). Boston,
MA: Bay State Psychological Associates.
Borofsky, G. L., (1994b) Use of the Employee Reliability Inventory as a pre-interview question-
naire. Boston, MA: Bay State Psychological Associates, .
Borofsky, G. L., (1996) Employee Reliability Inventory computer scoring system guide . (Rev. 6.2)
Boston, MA: Bay State Psychological Associates.
Borofsky, G. L., Alexander, J., Coleman, R., Reimers, C., Wackenheim, G., & McCormick, B.
(1995) Enhancing courteous job performance: The contribution of a pre-employment screen-
ing inventory. Psychological Reports, 77, 43-50.
Borofsky, G. L., Bielema, M., & Hoffman, J. (1993) Accidents, turnover, and the use of a pre-
employment screening inventory: further contributions to the validation of the Employee
Reliability Inventory. Psychological Reports, 73, 1067-1076.
Borofsky, G. L. & Garely, L. (1995) Assessing the effects of treatment for substance abuse: A
further contribution to the validation of the Employee Reliability Inventory. Psychological
Reports, 76, 1043-1049.
Borofsky, G. L., Green, J., Burzichelli, D., & Paludi, L. (1995) Predicting terminations for cause
and failure to complete successfully a 90-day probationary period of employment: Contribu-
tion of a pre-employment screening inventory. Psychological Reports, 77, 1031-1040.
Borofsky, G. L. & Klein, H.J. (1998) The use of a pre-employment screening system as part of a
comprehensive asset protection program: examining the incremental effects on inventory
shrinkage. Security Journal, 10, 23-29.
Borofsky, G. L., Klein, H.J., & Davis, W. (1993) Pre-employment screening for unreliable work
behaviors: an opportunity to work cooperatively with human resource managers. Security
Journal, 4 (4), 185-192.

37
Borofsky, G. L. & Smith, M. (1993) Reductions in turnover, accidents and absenteeism: the
contribution of a pre-employment screening inventory. Journal of Clinical Psychology, 49 (1),
109-116.
Borofsky, G. L. & Wagner, J. (1993) Terminations for cause and job tenure: the contributions of
a pre-employment screening inventory. Psychological Reports, 72, 591-599.
Borofsky, G. L., Wagner, J., & Turner, S. (1995) Sustained reductions in turnover and accidents
associated with the ongoing use of a pre-employment screening inventory: Results of a three-
year longitudinal study. Psychological Reports, 77, 195-204.
Borofsky, G. L. & Watson, R. (1994) Prediction of Early Voluntary Turnover and Job Performance:
The contribution of a pre-employment screening inventory. Psychological Reports, 74, 819-
826.

38
APPENDICES
 EMPLOYEE RELIABILITY INVENTORY
Applicant Name: Donald Sample
ID: ERITwos
Company: Natcon
Date Scored: 2 Mar 2012
ERI Number: 1053449

Results
Freedom from Disrupted Job Performance (A)

Courtesy (C)

Emotional Maturity (E)

Conscientiousness (F)

Trustworthiness (H)

Long Term Job Commitment (Q)

Safety (S)

The seven ERI® scales assess the likelihood that -

Freedom from Disrupted Applicant's activities outside of work will not disrupt his/her performance and productivity through behaviors such as
Job Performance (A) inattentiveness, unauthorized absence/lateness, failing to follow through on assignments, or other inappropriate work
behaviors.

Courtesy (C) The applicant's interactions with customers/guests will be characterized by a high level of courtesy and commitment to
service.

Emotional Maturity (E) The applicant's performance and productivity will not be disrupted due to the presence of maladaptive personality
traits, such as irresponsibility, difficulty in working cooperatively with others, poor judgment, or poor impulse control,
etc.

Conscientiousness (F) The applicant will perform on the job in a productive and conscientious manner, and will not be fired in the first 30
days of employment.

Trustworthiness (H) The applicant will perform on the job in a trustworthy manner and will not engage in various forms of untrustworthy
behaviour.

Long Term Job The applicant will show a long term commitment to the job and will not quit within the first 30 days of employment.
Commitment (Q)

Safety (S) The applicant will perform on the job in a safe manner, and will not have a significant on-the-job accident in the first 4
months of employment.
Further interpretive information:
Under no circumstances should the decision to hire or not hire an applicant be based solely on his/her ERI® results. Hiring
decisions should be based on a review of ALL information collected by you during the applicant evaluation process.

Because of the variability inherent in any type of scores, small differences in results should never be the basis for making
decisions about applicants or for comparing applicants.

The following table can be used to help you approximate where an applicant's results fit, relative to scores obtained by other job applicants.
This table shows the approximate percentage of job applicants who obtain poorer scores on that particular scale. The table is based on a
group of job applicants (N=60,670) who completed the ERI® as part of their pre-employment processing. This normative group represents all
10 Standard Industrial Classification (SIC) Code Divisions, 54 Major SIC Groups, and a wide range of job categories.

Zone 4 Zone 3 Zone 2 Zone 1

SCALE
B A B A B A B A

A 0% 2% 6% 18% 38% 63% 83% 96%

C 0% 2% 15% 20% 34% 54% 83% 95%

E 0% 5% 8% 13% 22% 45% 79% 95%

F 0% 7% 8% 10% 13% 31% 67% 91%

H 0% 2% 3% 7% 14% 33% 66% 91%

Q 0% 13% 14% 16% 18% 36% 72% 93%

S 0% 6% 11% 22% 29% 54% 78% 93%

NOTES:

The number in each cell represents the approximate percentage of job applicants in the normative sample who obtained scores on that scale
which were poorer than the job applicant's.

As an illustration of how to use this table, please note that in the normative sample, the number 10 appears in zone 3A for the "F" scale. This
indicates that scores in this zone are at approximately the 10th percentile (approximately 10% of the job applicants in the normative sample
obtained scores on the F scale that were poorer than zone 3A, or put slightly differently, approximately 10% of the normative sample
obtained F scale scores in zones 3B, 4A or 4B).

For Help: If you have questions regarding the administration, scoring, or interpretation of the ERI® please call Psychometrics Canada:
1-800-661-5158.

Employee Reliability Inventory (ERI®). Copyright 2017, Psychometrics Canada Ltd. All Rights reserved.