testing the test: validation, litigation & risk management kathleen k. lundquist, ph.d. john c....

46
Testing the Test: Validation, Litigation & Risk Management Kathleen K. Lundquist, Ph.D. John C. Scott, Ph.D.

Upload: maegan-dawe

Post on 14-Dec-2015

220 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Testing the Test: Validation, Litigation & Risk Management Kathleen K. Lundquist, Ph.D. John C. Scott, Ph.D

Testing the Test:Validation, Litigation & Risk

ManagementKathleen K. Lundquist, Ph.D.

John C. Scott, Ph.D.

Page 2: Testing the Test: Validation, Litigation & Risk Management Kathleen K. Lundquist, Ph.D. John C. Scott, Ph.D

2 2

About APTMetrics

Global Talent Management Solutions Provider Comprised of:

• Ph.D. industrial/organizational psychologists• Human resource consultants • Information technology specialists

What Sets APTMetrics Apart:• Professional integrity• Evidence-based approach• Technical expertise• Customer service

Diversity Supplier • Certified as a women-owned business

by WBENC• Certified as a women-owned small

business by the US SBA

Global Strategies for Talent Management.

Page 3: Testing the Test: Validation, Litigation & Risk Management Kathleen K. Lundquist, Ph.D. John C. Scott, Ph.D

3 3

Our Areas of Expertise

• Leader Assessment• Employee Selection• Litigation Support• Diversity Strategy &

Measurement• Job Analysis• Competency Modeling• Performance Management• Staffing for Mergers &

Acquisitions• Organizational Surveys

Global Strategies for Talent Management.

Page 4: Testing the Test: Validation, Litigation & Risk Management Kathleen K. Lundquist, Ph.D. John C. Scott, Ph.D

4 4

Our Web-Based Solutions Platform

APTMetrics®

SelectionMetrics®

Employee Selection System

LeadINsm

Leadership Assessment Suite

JobMetrics®

Job Analysis System

360Metrics®

360-Degree Feedback System

SurveyMetrics®

Organizational Survey System

Page 5: Testing the Test: Validation, Litigation & Risk Management Kathleen K. Lundquist, Ph.D. John C. Scott, Ph.D

5 5

APTMetrics’ U.S. Offices

Page 6: Testing the Test: Validation, Litigation & Risk Management Kathleen K. Lundquist, Ph.D. John C. Scott, Ph.D

6 6

• Expert Witness Testimony• Court appointed expert in settlement

• Ford • Abercrombie & Fitch• Morgan Stanley• Sodexo • The Coca-Cola Company

• Invited testimony on testing before EEOC• Negotiations with OFCCP• Proactive HR Process Audits• Development and validation of new selection

systems

Our Background with Testing and Litigation

Page 7: Testing the Test: Validation, Litigation & Risk Management Kathleen K. Lundquist, Ph.D. John C. Scott, Ph.D

7 7

• Are you concerned about a legal challenge to your tests or interviews?• Yes• No• I don’t know

Polling Question #1

Page 8: Testing the Test: Validation, Litigation & Risk Management Kathleen K. Lundquist, Ph.D. John C. Scott, Ph.D

8 8

The Federal Uniform Guidelines on Employee Selection Procedures (1978) define a test as:

Definition

“Any measure…used as a basis for any employment decision.”

Page 9: Testing the Test: Validation, Litigation & Risk Management Kathleen K. Lundquist, Ph.D. John C. Scott, Ph.D

9 9

What is a Test?

Work Simulation Exercises

Cognitive Ability Tests

Job Knowledge Tests

Personality and Interest Inventories

Honesty/Integrity Tests

Assessment Centers

Physical Abilities Tests

Application Form Data

Psychological Assessment

Education/Experience Requirements Reference Checks

Performance Evaluations

Interviews

Page 10: Testing the Test: Validation, Litigation & Risk Management Kathleen K. Lundquist, Ph.D. John C. Scott, Ph.D

10 10

• Is your company currently using written tests?• No, not at all• No, but considering using in the future• Yes, for a small number of jobs• Yes, for a wide range of jobs• I have no idea!

Polling Question #2

Page 11: Testing the Test: Validation, Litigation & Risk Management Kathleen K. Lundquist, Ph.D. John C. Scott, Ph.D

11 11

Validity…and the Law

Page 12: Testing the Test: Validation, Litigation & Risk Management Kathleen K. Lundquist, Ph.D. John C. Scott, Ph.D

12 12

Why are Tests Challenged?

• Adverse impact

• Validity• None

• Inadequate

• Dated

• Reliance on Studies Outside the Company

• Less Adverse Alternatives

• Inconsistent Administration

Page 13: Testing the Test: Validation, Litigation & Risk Management Kathleen K. Lundquist, Ph.D. John C. Scott, Ph.D

13 13

• Disproportionately fewer protected group applicants pass the test than majority group applicants

• Usually determined by:• 4/5ths or 80% rule• Standard deviation test

• The presence of adverse impact may be a given for many types of tests

Adverse Impact Definition

HOWEVER… tests with adverse impact can besuccessfully defended

Page 14: Testing the Test: Validation, Litigation & Risk Management Kathleen K. Lundquist, Ph.D. John C. Scott, Ph.D

14 14

4/5ths Rule Illustration

• 3 out of 5 African-Americans pass the test

• 60% pass rate

• 4 out of 5 Whites pass the test

• 80% pass rate

60% / 80% = 75%

Adverse Impact

Page 15: Testing the Test: Validation, Litigation & Risk Management Kathleen K. Lundquist, Ph.D. John C. Scott, Ph.D

15 15

Trends in Testing Case Law

• More specific evidence of adverse impact required• Beyond 80% rule• Practical and Statistical

Significance• Sample size issues

• Disparate treatment testing cases• Inconsistent

administration• Selective use of test• Jury trial and punitive

damages

• Greater emphasis on “validity”

of passing scores• Rationale• Set at acceptable

performance, not applicant flow

• Plaintiffs’ burden to identify less adverse alternatives• Must be knowable • Must demonstrate would be

less adverse• Must demonstrate

substantially the same validity

Page 16: Testing the Test: Validation, Litigation & Risk Management Kathleen K. Lundquist, Ph.D. John C. Scott, Ph.D

16 16

• Job analyses are conducted

• Procedures have face validity

• They document a search for less adverse alternatives

Employers Tend to be Successful When…

Page 17: Testing the Test: Validation, Litigation & Risk Management Kathleen K. Lundquist, Ph.D. John C. Scott, Ph.D

17 17

• Valid procedures are administered inconsistently

• Cut scores are set too high

Plaintiffs Tend to be Successful When…

Page 18: Testing the Test: Validation, Litigation & Risk Management Kathleen K. Lundquist, Ph.D. John C. Scott, Ph.D

18 18

249 Industrial and Organizational Psychologists answered this and related questions in a recent APT survey (July 2008)• Given a competently conducted criterion-related or content

validation study, how old would the validation study need to be to necessitate a new validation study?

• Average = 5 years old (Blue Collar, Supervisory, Managerial, Executive Tests)

• Technical Tests judged to have shorter shelf life = 3 to 3.5 years old

• Administrative/Customer Service/Sales = 4 to 4.5 years old

What is the “Shelf Life” of a Validation Study?

Page 19: Testing the Test: Validation, Litigation & Risk Management Kathleen K. Lundquist, Ph.D. John C. Scott, Ph.D

19 19

• Assuming a job’s tasks or work behaviors do not change, or change very little, what is the shelf life of a job analysis? That is, after how much time would one need to update the job analysis, even when there are little or no changes to the major tasks or work behaviors of the job?• Average = 5 to 6 years old

• What is the shelf life of the cut score? That is, how long would you recommend using a cut score before conducting additional research to determine if the cut score needs to change? • Average = 3 to 3.5 years old

Other Survey Results

Page 20: Testing the Test: Validation, Litigation & Risk Management Kathleen K. Lundquist, Ph.D. John C. Scott, Ph.D

20 20

• Conditions that shorten the shelf life of a validation study:

• Change in the nature of the job duties and or KSAs

• Conditions that result in the emergence of adverse impact or legal challenges

• Changes in applicant population• Changes in administration mode• Organizational changes (e.g., merger,

downsizing)

Other Survey Results

Page 21: Testing the Test: Validation, Litigation & Risk Management Kathleen K. Lundquist, Ph.D. John C. Scott, Ph.D

21 21

• These are professionally derived “rulesof thumb”

• Use these findings and your professional judgment

• Seek additional professional judgment• Seek legal input

And the Survey Says…

Page 22: Testing the Test: Validation, Litigation & Risk Management Kathleen K. Lundquist, Ph.D. John C. Scott, Ph.D

22 22

What must you do to validate

a test?

Page 23: Testing the Test: Validation, Litigation & Risk Management Kathleen K. Lundquist, Ph.D. John C. Scott, Ph.D

23 23

Test Validity Looks Like…

People who score high on the test are also high performers on the job

People who score low on the test are also low performers on the job

X

Y

Jo

b P

erf

orm

an

ce

Test Scores

Page 24: Testing the Test: Validation, Litigation & Risk Management Kathleen K. Lundquist, Ph.D. John C. Scott, Ph.D

24 24

• Selection procedures provide samples of behavior which allow us to make inferences about:• What basic abilities a person possesses• What the person knows• What the person can do• What a person is willing to do• How a person will behave in the future

The Validity of Selection Procedure Inferences

Page 25: Testing the Test: Validation, Litigation & Risk Management Kathleen K. Lundquist, Ph.D. John C. Scott, Ph.D

25 25

• Validity refers to the degree to which test scores are job related

• The process of validation involves accumulating evidence to provide a sound scientific basis for the proposed use of the test

Validation Defined

Page 26: Testing the Test: Validation, Litigation & Risk Management Kathleen K. Lundquist, Ph.D. John C. Scott, Ph.D

26 26

• Evidence Based Upon Test Content (Content Validity)• Demonstration that the content of a test is representative of

important aspects of performance on the job

• Evidence Based on Relations to Other Variables (Criterion Validity)• Statistical demonstration of a relationship between scores on a

test and job performance of a sample of employees

• Evidence Based on Internal Structure (Construct Validity)• Demonstration that test measures a construct (something

believed to be an underlying human trait or characteristic, such as conscientiousness) and the construct is important for successful job performance

Sources of Validity Evidence

Page 27: Testing the Test: Validation, Litigation & Risk Management Kathleen K. Lundquist, Ph.D. John C. Scott, Ph.D

27 27

Content Validity Study

1. Job Analysis

2. Test Development

3. SME Validation

4. Set Passing Scores

Profile job by identifying essential functions (WABs), knowledge, skills, abilities, and performance standards

Develop representative samples of performance domains - job sample tests, job skill tests, or job knowledge tests

Use subject matter expert judgment to document relationships between test content and performance domains

Use incumbents and subject matter experts to establish passing scores

Page 28: Testing the Test: Validation, Litigation & Risk Management Kathleen K. Lundquist, Ph.D. John C. Scott, Ph.D

28 28

• A comprehensive job analysis

• Competence in test construction

• Test content related to job’s content

• Test content representative of job’s content

• Examination of less adverse alternatives

• A passing score that selects those who can

better perform the job

Key Issues in Content Validity

Page 29: Testing the Test: Validation, Litigation & Risk Management Kathleen K. Lundquist, Ph.D. John C. Scott, Ph.D

29 29

Criterion-Related Validity Study

Establish Administrative Use

Job Analysis

Develop or Acquire Tests

Try Out/Pilot

Collect Test Data (Applicants or Employees)

Relate Test Scores & Performance Measures

Collect Performance Data

Try Out/Pilot

Develop Performance Measures

Page 30: Testing the Test: Validation, Litigation & Risk Management Kathleen K. Lundquist, Ph.D. John C. Scott, Ph.D

30 30

• Adequacy of job performance criteria• The psychometric quality of test and

criterion measure• Degree of correlation necessary to

establish validity• Examination of less adverse alternatives• Appropriateness of the passing score

Key Issues in Criterion-Related Validity

Page 31: Testing the Test: Validation, Litigation & Risk Management Kathleen K. Lundquist, Ph.D. John C. Scott, Ph.D

31 31

• No adverse impact• Transporting validity from another job

or location• Generalizing validity from other

studies of similar jobs

When Can You NOT Validate?

Page 32: Testing the Test: Validation, Litigation & Risk Management Kathleen K. Lundquist, Ph.D. John C. Scott, Ph.D

32 32

• Validation is the joint responsibility of test developer and test user

• When the use of a test differs from that supported by the test developer, the test user bears special responsibility for validation

Responsibility for Validation

Page 33: Testing the Test: Validation, Litigation & Risk Management Kathleen K. Lundquist, Ph.D. John C. Scott, Ph.D

33 33

• Job Analysis will identify what to assess

• It is not necessary to measure every important KSAO!

• It is necessary that every KSAO measured be important!

What Can You Test For?

Page 34: Testing the Test: Validation, Litigation & Risk Management Kathleen K. Lundquist, Ph.D. John C. Scott, Ph.D

34 34

Job Analysis is the Foundation

Job

Job

Job

Job

Analysis

Work Activities Performed

Scope and Effect of Work

Technical Skills Required

Competencies Required

Education Requirements

Experience Needed

Minimum & Preferred Qualifications

Test Specifications

Uniform Standardsfor Promotions

Structured Interviews

Performance Standards

Page 35: Testing the Test: Validation, Litigation & Risk Management Kathleen K. Lundquist, Ph.D. John C. Scott, Ph.D

35 35

• Does your company use formal job analyses as the basis for selection procedures?• No• Yes

Polling Question #3

Page 36: Testing the Test: Validation, Litigation & Risk Management Kathleen K. Lundquist, Ph.D. John C. Scott, Ph.D

36 36

Example of Test Specification Matrix

Page 37: Testing the Test: Validation, Litigation & Risk Management Kathleen K. Lundquist, Ph.D. John C. Scott, Ph.D

37 37

Types of Tests to Consider

ENTRY LEVEL

• Cognitive Ability Testing• Non-cognitive Measures• Interview

PROMOTIONAL

• Knowledge Testing• Performance Assessment• Interview

Page 38: Testing the Test: Validation, Litigation & Risk Management Kathleen K. Lundquist, Ph.D. John C. Scott, Ph.D

38 38

FaceValidity

Page 39: Testing the Test: Validation, Litigation & Risk Management Kathleen K. Lundquist, Ph.D. John C. Scott, Ph.D

39 39

• “High Fidelity” measurement tools (work samples, video, assessment centers) are more acceptable to candidates

• Once test specifications have been developed, decide:

• Custom design or identify commercially available test?

• What is appropriate testing medium?• Higher Fidelity = More Costly

• Whether custom or commercially available, must validate for your jobs

Selecting a Test

Page 40: Testing the Test: Validation, Litigation & Risk Management Kathleen K. Lundquist, Ph.D. John C. Scott, Ph.D

40 40

Combine Assessments

Cognitive

Job Performanc

e

Non-Cognitive

Structured Interview

Page 41: Testing the Test: Validation, Litigation & Risk Management Kathleen K. Lundquist, Ph.D. John C. Scott, Ph.D

41 41

• Cut-off scores should:• Be consistent with normal expectations of

proficiency within the workforce• Permit the selection of qualified applicants• Allow an organization to meet affirmative

action goals• Have a documented rationale

Criteria For Establishing Appropriate Cut-off Scores

Page 42: Testing the Test: Validation, Litigation & Risk Management Kathleen K. Lundquist, Ph.D. John C. Scott, Ph.D

42 42

What Makes Tests Fair and Defensible?

Validity

Based on job analysis

Standardized

Consistent

Validated

Implementation

Training

Ongoing monitoring

Appeals process

Communication

Page 43: Testing the Test: Validation, Litigation & Risk Management Kathleen K. Lundquist, Ph.D. John C. Scott, Ph.D

43 43

Ongoing Monitoring• Adverse Impact

• Test Content

• Administration Issues

Page 44: Testing the Test: Validation, Litigation & Risk Management Kathleen K. Lundquist, Ph.D. John C. Scott, Ph.D

44 44

1. “Aren’t tests harder to defend than other selection procedures?”

2. “The vendor says: ‘The test has no adverse impact.’ I have no problems, right?”

3. “The vendor says: ‘The test is validated – Trust us!’ I have no problems, right?”

4. “Candidates can take this new test at home over the Internet…so we can reduce overhead. I have no problems, right?”

5. “Once the test is validated, we’re done, right?”

Top Five Questions About Testing

Page 45: Testing the Test: Validation, Litigation & Risk Management Kathleen K. Lundquist, Ph.D. John C. Scott, Ph.D

45 45

For a copy of the results of APT’s recent Test Validity Shelf Life Survey email your request to:

[email protected]

Test Validity Survey