Employee Selection and Placement

HRM and Psychometrics

Information Gathering: Key Principles

• GIGO: "Garbage In, Garbage Out" - the quality of output is determined by the quality of input.
• Multisource/Multimethod: Utilizing various sources and methods to gather information.
• Cost/Benefit Analysis: Weighing the costs of information gathering against the benefits of having that information.
• 100% Certainty is Unlikely: Acknowledging that achieving complete certainty in information gathering is improbable, if not impossible.

Information in HRM

Information is vital for human resource management decisions in areas such as:

• HR planning/forecasting
• Job analysis: Includes Job Specifications (KSAs - Knowledge, Skills, Abilities) and Job Descriptions (TDRs - Tasks, Duties, Responsibilities)
• Recruitment: Using yield ratios.
• Validation studies
• Selection
• Training: Utilizing needs assessment.
• Development
• Performance evaluation: Including ratings and rankings.
• Discipline: Maintaining documentation.
• Compensation: Using job evaluation.
• EEO audits/affirmative action
• Safety data

Areas of HRM Using Job Analysis

• Work redesign
• Human resource planning
• Selection
• Training
• Performance appraisal
• Career planning
• Job evaluation: Determines job pay (Noe et al., 2018, p. 106).

Measurement & Testing in HRM

• Gathering information is a form of measurement.
• Measurement is critical for effective HRM.
• Many HRM activities are tests, such as:
  • Selection: Predicting future job performance.
  • Performance Appraisal: Measuring past performance.
  • Needs assessment: Measuring current KSAs to determine training needs.
• Psychometrics is the science of testing and measurement.

Psychometrics Defined

Psychometrics is the study of educational and psychological measurement, including knowledge, abilities, attitudes, and personality traits.

• It involves:
  • Constructing measurement instruments and procedures.
  • Developing theoretical approaches to measurement.
• Psychometricians may be HR professionals, not necessarily clinical psychologists.

Measurement Considerations

• What can you measure?
• Is the measure reliable?
• Is it accurate?
• Is it meaningful?
• Typically, only a subset of behaviors/performance/outcomes is measured.
• Focusing solely on the measured subset can be dangerous.

Examples of Measurement Pitfalls

• Pete Rose: Holds the record for most career hits in MLB history but also for most outs made.
• Quarterback Contract: A quarterback negotiated a contract focusing on minimizing interceptions but was then sacked more often and had a poor passing record.
• NFL Interception Leaders: Lists career leaders in NFL passes intercepted, including Brett Favre, George Blanda, and others.
• Be careful what you measure and reward.

Selection Process

• Organizations should create a selection process based on job descriptions.
• The selection process should identify candidates with necessary KASOs (Knowledge, Skills, Abilities, and Other characteristics).
• Requires methods to measure the effectiveness of selection tools.
  Figure 6.1 outlines steps in the selection process:

1. Recruitment: Create a large pool of reasonably qualified candidates with a reasonable likelihood of acceptance.
2. Screening Applications and Résumés
3. Testing and Reviewing Work Samples
4. Interviewing Candidates
5. Checking References and Background
6. Selection
7. Making a Selection

Successful Selection Method

• Provides reliable information.
• Provides valid information.
• Can be generalized to apply to candidates.
• Offers high utility.
• Uses legal selection criteria.

Psychometrics: Key Concepts

• Reliability: The extent to which a measurement is free from random error. Does the measurement generate consistent results? (p. 164)
• Validity: The extent to which performance on a measure (test score) relates to what the measure is designed to assess (job performance). Does the measure what it purports to measure? (p. 165)

Reliability in Selection

• Extent to which a measurement is free from random error.
• A reliable measurement generates consistent results.
• Organizations use statistical tests to compare results over time.
  • Correlation coefficients
  • A higher correlation coefficient signifies a greater degree of reliability.

Validity in Selection

• Extent to which performance on a measure (test score) is related to job performance.
• What is being measured?

Reliability & Validity Relationship

• An unreliable measure cannot be valid.
• A reliable measure may not be a valid measure.
• Reliability is a necessary precondition for validity, but it does not guarantee validity.

BMI Example

• BMI (Body Mass Index) is calculated as weight (lbs) / [height (inches)]^2 * 703
• Arnold Schwarzenegger's BMI during his competition days ranged from $30.2$ to $33.4$, which is considered off the charts.

Intelligence Measurement

• Einstein's IQ was $160+$.
• Distribution of IQ scores:
  • Mentally inadequate: $70$
  • Low intelligence: $70-85$
  • Average intelligence: $85-115$ ($34.1$)
  • Above average intelligence: $115-130$ ($13.6$)
  • High intelligence: $130-145$ ($2.1$)
  • Superior intelligence
  • Exceptionally gifted

Validity Types

• Extent to which performance on a measure (test score) is related to job performance.
• Federal government's Uniform Guidelines on Employee Selection Procedures accept three ways of measuring validity:
  • Criterion-related
  • Content
  • Construct

Correlation

• Correlation coefficients measure the degree to which two sets of numbers are related.
• Range from $-1.0$ to $1.0$. Zero indicates no correlation.
• A higher score on an interview should indicate a better future performer.
• Perfect correlation is impossible in HRM. Scores of $0.30$ to $0.50$ may represent important relationships.

Interpreting Correlation Coefficients

• Test A: Job content sample measuring wasted inputs, correlated with performance one year post-hire, $r = -0.55$.

• Test B: Recommendation letter evaluations correlated with performance one year post-hire, $r = +0.10$.

• The correlation is the 'predictive validation' approach.

• Strength of relationship is based on the absolute value, e.g., |-0.7| > |0.2|.

Correlation vs. Causation

• Correlation does not equal causation.
• Spurious correlations example: Correlation between measles cases and marriage rate.
• Another example: A $0.947$ correlation between per capita cheese consumption in the US and number of people who died by becoming tangled in their bedsheets.

Examples of Spurious Correlations

• Number of people who drowned by falling into a pool correlates with films Nicolas Cage appeared in (r=0.666)
• Per capita consumption of mozzarella cheese correlates with Civil engineering doctorates awarded. (r=0.958648)
• Letters in Spelling Bee winning word correlates with the Number of people killed by venomous spiders

Interpreting Correlations

• Be careful when interpreting correlations; there may be a valid reason for the correlation.
• As the overall population grows (increase in high school graduates), donut consumption may grow.
• There may be a third variable.

Contamination and Deficiency

• Contamination: Aspects measured by the test that are not relevant to the material being tested.
• Deficiency: Relevant aspects of the material being tested that are not measured by the test.

Criterion Deficiency & Contamination Model

• Ultimate (theoretical) criterion: Ideal measure of all relevant job performance aspects.
• Actual criterion: Actual measure of job performance obtained.
• Differences between ultimate criterion and actual criterion represent imperfections in measurement.
  • Deficiency: When the actual criterion is missing information.
  • Contamination: When the actual criterion includes unrelated information.

Examples of Deficiency and Contamination

• Deficiency: An exam lacking questions on the Adidas case study when it should have been included.
• Contamination: A quiz question in Russian that tests the ability to read Russian rather than the subject matter.
• A question contaminated with your ability to understand English, but the impact on your performance will be lessened.

Validation Meaning

• A strong positive (or negative) correlation between a measure and job performance means the measure should be a valid basis for selecting (or rejecting) a candidate.
• Important for identifying the best employees and demonstrating fair employment practices.

Validation Process Details

• Federal government's Uniform Guidelines on Employee Selection Procedures include:
  • Criterion-related validity:
    • Predictive validation
    • concurrent validation
  • Content validity
  • Construct validity
• Validation is complex and expensive, but the information can be invaluable.

Criterion-Related Validity

• A measure of validity based on showing a substantial correlation between test scores and job performance scores.
• Two kinds of research are possible for arriving at criterion-related validity:
  • Predictive Validation:
    Research that uses test scores of all applicants and looks for a relationship between scores and future performance of applicants who were hired.
  • Concurrent Validation:
    Research that consists of administering a test to people who currently hold a job, and then comparing their scores to existing measures of job performance.

Content and Construct Validity

• Content Validity
  Consistency between test items or problems and kinds of situations or problems that occur on the job.

• Construct Validity
  Consistency between a high score on a test and high level of a construct (i.e., intelligence or leadership ability) as well as between mastery of this construct and successful performance of the job.

Additional Validation Considerations

• Cost tradeoffs
• Imperfect (it’s not practical to measure everything)
• GIGO (Garbage In, Garbage Out)
• Multisource/multimethod (e.g., combine selection measures/assessment centers)

Additional Considerations

• Generalizability
• Utility
• Legal defensibility (Griggs v. Duke Power, ADA: essential functions)

Generalizability

• A generalizable selection method applies not only to the conditions in which the method was originally developed (job, organization, people, time period, etc.).
• It also applies to other organizations, jobs, applicants, etc.

Practical Value (Utility)

• Being valid, reliable, and generalizable adds value to a selection method.
• Another consideration is the cost of using the selection method.
• Selection methods should cost significantly less than the benefits of hiring new employees.
• Methods that provide economic value greater than the cost of using them are said to have utility.
• Cost-benefit tradeoffs!

Legal Standards for Selection

• All selection methods must conform to existing laws and legal precedents.
• Three acts have formed the basis for a majority of suits filed by job applicants:
  • Civil Rights Act of 1964 and 1991
  • Age Discrimination in Employment Act of 1967
  • Americans with Disabilities Act of 1991

Legal Defensibility

• Job analysis information is critical for defending against a charge of illegal discrimination because it can provide evidence that a measure is valid
• There are two key defenses against a charge of illegal discrimination (using an inappropriate and illegal, and, hence, prohibited basis for making workplace decisions):
  • BFOQ (bona fide occupational qualification): which has very narrow and limited applicability (p. 73)
  • Business necessity (p. 74): Griggs v. Duke Power Co. (1971)
  • Validity can demonstrate job relatedness/job necessity.

Final Measurement Considerations

• What can you measure?
• Is the measure reliable?
• Is it valid?
• Is it meaningful? Does it have utility?
• Can we defend the use of the measure (e.g., in court)?
• Typically, only a subset of behaviors/performance/outcomes is measured.
• Focusing only on that subset can be dangerous.