Sources and References

Primary textbook — None assigned

Supplementary texts — Cascio, W. F., & Aguinis, H. (2019). Applied Psychology in Talent Management. Sage.; Gatewood, R. D., Feild, H. S., & Barrick, M. (2016). Human Resource Selection. Cengage.; Catano, V. M. et al. (2020). Recruitment and Selection in Canada. Nelson.

Online resources — SIOP (siop.org), Canadian Human Rights Commission

Assigned readings — Wilmot et al. (2014); Kuncel & Hezlett (2007); Highhouse et al. (2019)

Chapter 1: Introduction to Personnel Psychology

What Is Personnel Psychology?

Personnel psychology is a major subfield of industrial-organizational (I/O) psychology that focuses on the application of psychological principles and research methods to human resource issues in the workplace. While industrial psychology historically addressed topics such as job analysis, selection, training, and performance appraisal, organizational psychology concerned itself with motivation, leadership, group dynamics, and organizational culture. Personnel psychology sits squarely on the “industrial” side of this division, though modern practice blurs these boundaries considerably.

At its core, personnel psychology asks a deceptively simple question: how can organizations identify, select, develop, and retain people who will perform their jobs effectively? Answering that question requires a sophisticated understanding of human individual differences, psychometrics, statistical prediction, legal frameworks, and organizational contexts.

A Brief History of Personnel Psychology

The roots of personnel psychology stretch back to the early twentieth century. Hugo Munsterberg, often called the father of applied psychology, published Psychology and Industrial Efficiency in 1913, advocating for the use of psychological testing in employee selection. During World War I, the U.S. Army commissioned the development of the Army Alpha and Army Beta tests to classify recruits by intellectual ability — one of the first large-scale applications of mental testing.

The interwar period saw the growth of applied psychology in industry. The Hawthorne studies (1924–1932) at Western Electric highlighted the importance of social and psychological factors in productivity, although their primary contributions were to organizational rather than personnel psychology. Meanwhile, personnel selection research advanced through the development of structured interviews, biographical inventories, and aptitude test batteries.

World War II catalyzed further progress. The Assessment Center method, originally developed by the British War Office Selection Boards and the U.S. Office of Strategic Services (OSS), introduced multi-method, multi-assessor evaluation procedures that remain in use today. The postwar decades saw the professionalization of I/O psychology, the founding of the Society for Industrial and Organizational Psychology (SIOP) as Division 14 of the American Psychological Association, and the publication of foundational texts on personnel selection and psychometrics.

In Canada, the development of personnel psychology was shaped by the country’s distinct legal, linguistic, and cultural context. The passage of the Canadian Human Rights Act (1977) and the Employment Equity Act (1986, revised 1995) created a regulatory environment that differs meaningfully from the U.S. framework, with implications for how selection systems are designed, validated, and defended.

The Canadian Context

Personnel psychology in Canada operates within a bilingual, multicultural, and rights-based legal framework. Several features distinguish the Canadian context:

• Human rights legislation exists at both the federal and provincial/territorial levels. The Canadian Human Rights Act prohibits discrimination in federally regulated workplaces on grounds including race, national or ethnic origin, colour, religion, age, sex, sexual orientation, gender identity or expression, marital status, family status, genetic characteristics, disability, and conviction for a pardoned offence.
• Employment equity legislation requires federally regulated employers to identify and remove barriers for four designated groups: women, Aboriginal peoples, persons with disabilities, and members of visible minorities.
• The Canadian Charter of Rights and Freedoms (Section 15) guarantees equality before and under the law, which has implications for employment practices even in the private sector through its influence on human rights legislation.
• The concept of a bona fide occupational requirement (BFOR) — established in landmark cases such as British Columbia (Public Service Employee Relations Commission) v. BCGSEU (the “Meiorin” decision, 1999) — sets the standard for justifying selection criteria that have adverse impact on protected groups.

These legal frameworks mean that Canadian personnel psychologists must be attentive not only to the psychometric quality of their tools but also to their fairness and compliance with human rights principles.

The Scientist-Practitioner Model

Personnel psychology is firmly grounded in the scientist-practitioner model, which holds that effective practice must be informed by rigorous research, and that research questions should be driven by practical problems. This model, articulated at the 1949 Boulder Conference for clinical psychology, has been equally embraced by I/O psychologists.

In practice, the scientist-practitioner model means that personnel psychologists:

1. Use evidence-based methods for selection, appraisal, and training, rather than relying on intuition or tradition.
2. Apply research methodology (experimental design, psychometrics, statistics) to evaluate the effectiveness of HR interventions.
3. Stay current with the research literature and integrate findings from meta-analyses and validity generalization studies into practice.
4. Contribute to knowledge through applied research, often conducted within organizational settings.

This dual commitment to science and practice is what distinguishes personnel psychology from lay approaches to hiring and management. As we will see throughout this course, many common organizational practices — unstructured interviews, graphology, brainteaser questions — lack empirical support, while well-validated methods are often underused.

Overview of the Course

This course covers the major topics in personnel psychology in a logical sequence:

1. Statistical foundations — the tools needed to evaluate personnel research (variance, correlation, regression).
2. Job performance — defining and modeling the criterion we are trying to predict.
3. Performance appraisal — measuring job performance in practice.
4. Job analysis — systematically describing what jobs require.
5. Measurement fundamentals — the principles underlying psychological testing.
6. Reliability and validity — evaluating the quality of measurements.
7. Predictor constructs and methods — what we measure and how we measure it in selection.
8. Selection decision-making — combining information to make hiring decisions.
9. Legal and ethical issues — the regulatory and moral framework for selection.
10. Recruitment — attracting applicants to organizations.
11. Utility analysis — quantifying the economic value of selection systems.

Chapter 2: Statistical Foundations

Why Statistics Matter in Personnel Psychology

Personnel psychology is fundamentally a science of prediction. We want to predict who will perform well on the job, who will stay with the organization, who will benefit from training. Making these predictions requires quantitative tools. This chapter reviews the statistical concepts that underpin the rest of the course: variance, correlation, regression, effect sizes, and meta-analysis.

Variance

Variance is important in personnel psychology because it tells us how much individuals differ from one another on a given attribute. If there is no variance in job performance — if everyone performs at the same level — then there is no point in trying to select the best candidates. Selection is valuable precisely because people differ, and those differences are consequential.

Range restriction is a common problem in personnel research. When we study the relationship between a predictor (e.g., a cognitive ability test) and job performance, we typically can only observe performance for people who were hired. If hiring decisions were based (even partially) on the predictor, then the hired group will have less variance on the predictor than the full applicant pool. This restriction of range attenuates observed correlations and must be corrected statistically to estimate the true relationship.

Correlation

In personnel psychology, the correlation between a predictor and a criterion is called a validity coefficient. For example, if we correlate scores on a cognitive ability test with supervisor ratings of job performance, the resulting \( r \) tells us how well the test predicts performance.

The square of the correlation coefficient, \( r^2 \), is called the coefficient of determination and represents the proportion of variance in one variable that is accounted for by the other. However, some researchers argue that \( r^2 \) can be misleading as a measure of practical significance, because even modest correlations can have substantial practical value when applied across large numbers of selection decisions.

Interpreting Validity Coefficients

In personnel selection, validity coefficients rarely exceed \( r = .50 \). Cohen’s (1988) benchmarks for small (\( r = .10 \)), medium (\( r = .30 \)), and large (\( r = .50 \)) effects are sometimes used, but these benchmarks were not designed for applied prediction contexts. In personnel selection, a validity of \( r = .30 \) is considered practically useful, and even validities of \( r = .15 \)–\( .20 \) can add value in high-volume selection.

Regression

Multiple regression is the statistical foundation for compensatory selection models, in which a high score on one predictor can compensate for a low score on another. The multiple correlation coefficient \( R \) indicates how well the set of predictors jointly predicts the criterion.

A practical concern in multiple regression is shrinkage — the tendency for the multiple \( R \) to be inflated in the sample used to develop the regression equation and to decrease (shrink) when applied to a new sample. Cross-validation is essential to estimate how well a regression equation will generalize.

Effect Sizes and Practical Significance

Beyond statistical significance (which depends heavily on sample size), personnel psychologists are increasingly interested in effect sizes — standardized measures of the magnitude of a relationship or difference. Common effect size metrics include:

Meta-Analysis

The pioneering meta-analytic work of Schmidt and Hunter demonstrated that much of the variability in validity coefficients across studies was due to statistical artifacts rather than true differences in validity across situations. This finding — known as validity generalization — challenged the earlier view that validity was necessarily situation-specific and had to be established locally for every job and organization.

Key meta-analytic findings that will recur throughout this course include:

• General mental ability (GMA) tests have an estimated operational validity of \( \rho \approx .51 \) for predicting job performance across jobs (Schmidt & Hunter, 1998).
• Structured interviews are substantially more valid than unstructured interviews.
• Conscientiousness is the Big Five personality trait most consistently related to job performance across occupations.
• Work sample tests show high validity but are expensive to develop and administer.

Chapter 3: Job Performance

Defining Job Performance

Before we can select employees effectively, we need to understand what we are trying to predict. Job performance is the criterion — the outcome variable — in personnel selection. Defining and measuring job performance is therefore a foundational concern.

It is important to distinguish performance from related constructs:

• Effectiveness refers to outcomes that may be influenced by factors beyond the individual’s control (e.g., sales revenue depends on market conditions).
• Productivity is typically a ratio of outputs to inputs.
• Utility refers to the value of performance to the organization.

Campbell’s Model of Job Performance

John Campbell and colleagues proposed one of the most influential taxonomies of job performance, identifying eight components that apply across jobs:

1. Job-specific task proficiency — core technical tasks of the position.
2. Non-job-specific task proficiency — tasks that are common across many jobs in the organization.
3. Written and oral communication — the ability to convey information clearly.
4. Demonstrating effort — the willingness to work hard and persist.
5. Maintaining personal discipline — avoiding counterproductive behaviors.
6. Facilitating peer and team performance — helping and cooperating with coworkers.
7. Supervision/leadership — guiding and motivating subordinates.
8. Management/administration — organizing resources and handling administrative duties.

Campbell argued that all jobs can be described in terms of some subset of these components, and that each component has distinct determinants: declarative knowledge (knowing what to do), procedural knowledge and skill (knowing how to do it), and motivation (choosing to do it).

Task Performance vs. Contextual Performance

A seminal distinction in the performance literature is between task performance and contextual performance (Borman & Motowidlo, 1993):

• Task performance refers to activities that contribute to the organization’s technical core, either directly (producing goods or services) or indirectly (supporting the technical core). Task performance varies across jobs and is closely related to job-specific knowledge and skills.
• Contextual performance refers to activities that support the organizational, social, and psychological environment in which the technical core operates. Examples include volunteering for extra duties, helping coworkers, following organizational rules, and endorsing organizational objectives. Contextual performance is more similar across jobs and is more strongly related to personality (especially conscientiousness and agreeableness) than to cognitive ability.

This distinction matters for selection because different predictors may be optimal for predicting different aspects of performance.

Organizational Citizenship Behavior (OCB)

Common dimensions of OCB include:

• OCB-I (individual) — behaviors directed toward specific individuals (e.g., helping a coworker who has been absent).
• OCB-O (organizational) — behaviors directed toward the organization as a whole (e.g., defending the organization when others criticize it).
• Altruism — voluntarily helping others with work-related problems.
• Conscientiousness (in the OCB sense) — going beyond minimum role requirements.
• Sportsmanship — tolerating inconveniences without complaining.
• Courtesy — taking steps to prevent problems with others.
• Civic virtue — participating in organizational governance.

Research has shown that OCBs are related to both employee attitudes (job satisfaction, organizational commitment) and personality traits, and that they contribute to organizational effectiveness.

Counterproductive Work Behavior (CWB)

CWBs include theft, sabotage, absenteeism, aggression, substance abuse on the job, and cyberloafing. They can be directed at the organization (CWB-O) or at individuals (CWB-I). Research suggests that CWBs are predicted by personality traits (low conscientiousness, low agreeableness, low emotional stability), situational factors (perceived injustice, poor supervision), and their interaction.

There is debate about whether CWB and OCB are opposite ends of a single continuum or distinct constructs. The evidence suggests they are related but not perfectly inverse — it is possible for an employee to engage in both citizenship and counterproductive behaviors.

The Dynamic Criterion

An important question is whether job performance is stable over time or changes. The dynamic criterion hypothesis suggests that the validity of predictors may change as employees gain experience, because the determinants of performance shift from ability to motivation, or because individuals’ relative standing changes.

Research on this question has produced mixed results. While there is some evidence that validity coefficients can change over time, the changes are generally modest. Nevertheless, the dynamic criterion concept reminds us that performance is not a static attribute but an evolving pattern of behavior.

Chapter 4: Performance Appraisal

Why Appraise Performance?

Performance appraisal is the systematic process of evaluating employees’ job performance. It serves multiple purposes:

• Administrative decisions — promotion, termination, compensation, and placement.
• Developmental feedback — identifying strengths and areas for improvement.
• Criterion data for validation — performance appraisal ratings are commonly used as the criterion in selection validation studies.
• Documentation — creating a record that can be used in legal proceedings if employment decisions are challenged.

Types of Performance Measures

Performance measures can be classified into several categories:

Objective Measures

Objective performance measures include quantifiable outputs such as sales volume, units produced, error rates, and absenteeism. Their advantages include apparent objectivity and ease of quantification. However, they suffer from several problems:

• Criterion deficiency — they often capture only a narrow slice of the performance domain.
• Criterion contamination — they may be influenced by factors beyond the individual’s control.
• Opportunity bias — employees may have unequal opportunities to demonstrate performance.

Subjective Measures (Ratings)

Most performance appraisal in organizations relies on subjective ratings, typically provided by supervisors. Ratings have the advantage of being applicable to virtually any job and can capture the full range of performance dimensions. However, they are susceptible to various biases and errors.

Rating Scale Formats

Several rating scale formats have been developed:

Rater Errors and Biases

A perennial concern in performance appraisal is the tendency for ratings to be distorted by various rater errors:

• Halo error — the tendency to rate an employee similarly across all dimensions based on an overall impression, rather than distinguishing between dimensions.
• Leniency/severity — the tendency to rate all employees too high (leniency) or too low (severity) relative to their true performance.
• Central tendency — the tendency to avoid extreme ratings and rate everyone near the midpoint of the scale.
• Recency effects — giving disproportionate weight to recent performance rather than performance over the entire rating period.
• Contrast effects — rating an employee higher or lower based on comparison with the previous employee rated, rather than against an absolute standard.
• Similar-to-me bias — rating employees more favorably when they share characteristics (demographics, attitudes, background) with the rater.

360-Degree Feedback

Research on 360-degree feedback has found that:

• Different rating sources show only moderate agreement, suggesting they are capturing somewhat different aspects of performance.
• Self-ratings tend to be more lenient and less correlated with other sources than ratings from supervisors or peers.
• Peer ratings can be valid predictors of future performance and are less susceptible to halo error than supervisor ratings.
• The effectiveness of 360-degree feedback depends on how the information is used — it is generally more effective for developmental purposes than for administrative decisions.

Frame-of-Reference Training

Research has consistently shown that FOR training is the most effective approach to improving rating accuracy, outperforming rater error training (which teaches raters about common errors but may simply shift the distribution of ratings without improving accuracy) and behavioral observation training.

Chapter 5: Job Analysis

What Is Job Analysis?

Job analysis is the foundation of virtually every personnel psychology activity:

• Selection — job analysis identifies the KSAOs that selection tools should measure.
• Performance appraisal — job analysis defines the dimensions on which performance should be evaluated.
• Training — job analysis identifies the tasks that must be trained and the KSAOs that trainees need to develop.
• Job evaluation and compensation — job analysis provides the basis for determining the relative value of jobs.
• Legal defense — job analysis provides evidence that selection criteria are job-related (a key requirement under Canadian human rights law).

Job Analysis Methods

Observation and Interview

The most basic methods involve directly observing workers as they perform their jobs and interviewing them (and their supervisors) about what the job entails. These methods are flexible and can capture qualitative detail, but they are time-consuming and may be influenced by the analyst’s biases.

Critical Incidents Technique

The critical incidents technique (Flanagan, 1954) involves collecting descriptions of specific behaviors that represent particularly effective or ineffective job performance. Subject matter experts (SMEs) provide detailed accounts of incidents they have observed, which are then sorted into categories representing performance dimensions.

Critical incidents are especially useful for developing behaviorally anchored rating scales (BARS) and for identifying the KSAOs that distinguish good from poor performers.

Position Analysis Questionnaire (PAQ)

The Position Analysis Questionnaire (McCormick, Jeanneret, & Mecham, 1972) is a structured, standardized job analysis instrument that describes jobs in terms of 195 job elements organized into six categories:

1. Information input
2. Mental processes
3. Work output
4. Relationships with other persons
5. Job context
6. Other job characteristics

The PAQ is worker-oriented — it describes the behaviors and processes workers use, rather than the specific tasks of the job. This makes it applicable across a wide range of jobs and allows for quantitative comparisons between jobs.

O*NET

The Occupational Information Network (O*NET), developed by the U.S. Department of Labor, is a comprehensive database of occupational information. While American in origin, O*NET is widely used in Canada and internationally as a starting point for understanding job requirements. O*NET describes occupations in terms of:

• Worker characteristics (abilities, interests, work values, work styles)
• Worker requirements (skills, knowledge, education)
• Experience requirements
• Occupational requirements (work activities, work context, organizational context)
• Occupation-specific information (tasks, tools, technology)
• Workforce characteristics (labor market information)

Canada has developed its own occupational classification system, the National Occupational Classification (NOC), which categorizes occupations based on skill type and skill level.

Competency Modeling

Critics argue that competency modeling lacks the rigor and specificity of traditional job analysis, often producing generic competency lists that could apply to almost any managerial job. Proponents counter that competency modeling is more aligned with organizational strategy, more understandable to non-psychologists, and better suited to rapidly changing work environments.

In practice, the best approaches often combine elements of traditional job analysis (for legal defensibility and psychometric rigor) with competency modeling (for strategic alignment and organizational buy-in).

Chapter 6: Measurement Fundamentals

Classical Test Theory

Key assumptions of CTT include:

• The expected value of error is zero: \( E(E) = 0 \).
• Error is uncorrelated with the true score: \( r_{TE} = 0 \).
• Errors on different measurements are uncorrelated: \( r_{E_1E_2} = 0 \).

From these assumptions, it follows that the variance of observed scores equals the variance of true scores plus the variance of error:

Reliability is defined as the proportion of observed score variance that is true score variance:

Scales of Measurement

Psychologists distinguish four scales of measurement (Stevens, 1946):

Most psychological tests produce scores that are treated as interval-level data, though the degree to which this assumption is met is often debated.

Standardization

Standardization is essential in personnel selection when combining scores from different tests that use different scales. Without standardization, a predictor measured on a larger numerical scale would receive undue weight in a composite.

Norms and Score Interpretation

A test score has limited meaning in isolation. Norms provide the reference group against which individual scores are interpreted. Norm-referenced interpretation compares an individual’s score to the distribution of scores in a relevant normative sample. The quality of norms depends on:

• The size of the normative sample.
• Its representativeness of the population to which the test will be applied.
• Its recency — norms can become outdated (the Flynn effect demonstrates systematic changes in cognitive ability test scores over time).

In personnel selection, norms may be based on the general population, on applicant pools for similar jobs, or on incumbent employees.

Chapter 7: Reliability

What Is Reliability?

Reliability is a necessary (but not sufficient) condition for validity. If a test does not measure consistently, it cannot measure accurately. However, a test can be highly reliable while measuring the wrong thing.

Types of Reliability

Test-Retest Reliability

Test-retest reliability is assessed by administering the same test to the same group on two occasions and correlating the two sets of scores. It estimates the stability of scores over time. A high test-retest correlation indicates that individuals maintain their relative standing across occasions.

Considerations:

• The time interval between administrations affects the estimate — longer intervals typically produce lower correlations.
• Practice effects and memory may inflate test-retest correlations for short intervals.
• Test-retest reliability is most appropriate for constructs assumed to be stable (e.g., cognitive ability) and less appropriate for constructs that may genuinely change (e.g., mood).

Internal Consistency

Internal consistency reliability estimates the degree to which items within a test measure the same construct. The most common index is Cronbach’s alpha (\( \alpha \)):

where \( k \) is the number of items, \( s_i^2 \) is the variance of item \( i \), and \( s_X^2 \) is the variance of total scores.

Alpha ranges from 0 to 1, with values above .70 generally considered acceptable for research purposes and values above .80 or .90 preferred for high-stakes decisions. However, alpha is influenced by test length (longer tests tend to have higher alpha) and is only appropriate for unidimensional scales.

Split-half reliability is a related approach: the test is divided into two halves (e.g., odd and even items), scores on the two halves are correlated, and the Spearman-Brown formula is applied to estimate the reliability of the full-length test:

where \( r_{hh} \) is the correlation between the two halves.

Inter-Rater Reliability

Inter-rater reliability (or inter-rater agreement) assesses the degree to which different raters assign similar scores to the same targets. This is particularly important in personnel psychology for performance appraisal ratings, interview scores, and assessment center ratings.

Common indices include:

• Intraclass correlation coefficient (ICC) — the proportion of total variance attributable to true differences between ratees, rather than differences between raters.
• Cohen’s kappa (\( \kappa \)) — a measure of agreement for categorical ratings that corrects for chance agreement.
• Percentage agreement — simple but does not correct for chance.

Wilmot et al. (2014): Increasing Interrater Reliability Using Composite Performance Measures

Standard Error of Measurement

The SEM has important practical implications in personnel selection. When two candidates’ scores differ by less than one or two SEMs, the difference may not reflect a true difference in the underlying construct. This reasoning underlies the practice of banding in selection decision-making (discussed in Chapter 11).

Chapter 8: Validity

What Is Validity?

Modern validity theory, as articulated in the Standards for Educational and Psychological Testing (AERA, APA, NCME, 2014), emphasizes that validity is a unitary concept — there is one overarching question (are the intended inferences justified?) — but evidence can come from multiple sources.

Sources of Validity Evidence

Content Validity

Content validity refers to the degree to which the content of a test is representative of the domain it is intended to measure. Content validation involves a systematic comparison of the test content with the job content, typically based on a thorough job analysis.

Content validity is established through expert judgment rather than statistical analysis. Subject matter experts evaluate whether test items are relevant to the job, whether the test covers the important aspects of the job domain, and whether the test does not include irrelevant content.

Content validation is particularly appropriate for:

• Work sample tests — where candidates perform actual job tasks.
• Knowledge tests — where the test covers job-related knowledge identified through job analysis.
• Structured interviews — where questions are based on job-related scenarios.

Criterion-Related Validity

Criterion-related validity is established by demonstrating a statistical relationship between test scores (predictors) and measures of job performance (criteria). There are two designs:

• Predictive validity — administer the test to applicants, hire without using the test scores, then collect performance data later and correlate test scores with performance. This is the ideal design but is rarely feasible because it requires ignoring test results in hiring decisions.
• Concurrent validity — administer the test to current employees and correlate their scores with current performance measures. This is more practical but may be affected by range restriction (poor performers may have been terminated) and the effects of job experience.

The resulting correlation coefficient is the validity coefficient. In Canadian human rights law, criterion-related validity evidence is one way to demonstrate that a selection tool is a bona fide occupational requirement (BFOR).

Construct Validity

Construct validity is the most comprehensive form of validity evidence and subsumes content and criterion-related validity as special cases.

Validity Generalization

The VG research program, initiated by Schmidt and Hunter in the late 1970s, had profound implications for personnel selection practice:

1. Organizations may not need to conduct local validation studies if validity has been established meta-analytically for the same type of predictor and similar jobs.
2. Synthetic or transportability strategies — in which validity evidence from other settings is used to justify the use of a test — are scientifically defensible.
3. Situational specificity — the view that validity is unique to each organization and job — is largely unsupported for well-established predictors like cognitive ability tests.

However, VG has been debated. Critics argue that moderators (e.g., job complexity, criterion type) can meaningfully affect validity, and that blanket generalization without attention to context is unwarranted.

Kuncel and Hezlett (2007): Standardized Tests Predict Graduate Students’ Success

Threats to Validity

Several factors can affect the accuracy of validity estimates:

• Range restriction — when the sample has less variability on the predictor than the applicant pool (because selection has already occurred), the observed validity coefficient underestimates true validity.
• Criterion contamination — when the criterion measure is influenced by knowledge of predictor scores.
• Criterion deficiency — when the criterion measure fails to capture important aspects of performance.
• Low reliability — unreliability in either the predictor or the criterion attenuates the observed correlation.

Statistical corrections for range restriction and criterion unreliability are routinely applied in meta-analyses to estimate operational validity.

Chapter 9: Predictor Constructs

Overview

This chapter examines the major psychological constructs used in personnel selection — the “what” of selection, as distinct from the “how” (methods) covered in the next chapter. The constructs reviewed here are cognitive ability, personality, emotional intelligence, and integrity.

General Mental Ability (GMA)

Why GMA Predicts Performance

GMA predicts job performance primarily through its relationship with job knowledge acquisition. People higher in GMA learn job-relevant information more quickly and thoroughly, which translates into better task performance. The validity of GMA is higher for more complex jobs, because complex jobs place greater demands on learning and problem-solving.

The Adverse Impact Problem

A major concern with GMA tests is that they produce adverse impact — substantial score differences between racial/ethnic groups and, to a lesser extent, between sexes. In the North American context, Black applicants score approximately one standard deviation below White applicants on average, with Hispanic applicants falling in between. These group differences create a tension between using the most valid predictor (GMA) and achieving a diverse workforce.

This tension — sometimes called the diversity-validity dilemma — is one of the central challenges in personnel selection and has motivated research on alternative predictors and selection strategies that may reduce adverse impact without sacrificing validity.

The Big Five Personality Traits

Validity of Personality for Predicting Performance

Meta-analytic research has established the following patterns:

Conscientiousness is the most generalizable personality predictor, showing positive validity across occupations and criterion types. It predicts both task performance (through effort and persistence) and CWB (through rule-following and self-control).

Faking and Social Desirability

A major concern with self-report personality measures in selection is faking — applicants may distort their responses to present themselves favorably. Research suggests that:

• Applicants can and do fake on personality tests, especially on conscientiousness and emotional stability.
• The effect of faking on the rank ordering of applicants is modest but not negligible.
• Various strategies (forced-choice formats, warnings, response time monitoring, applicant-report vs. other-report) can mitigate but not eliminate faking.

Emotional Intelligence

The evidence on EI as a predictor of job performance is mixed:

• Ability-based EI shows some incremental validity beyond GMA and personality, particularly for jobs involving emotional labor.
• Trait-based EI overlaps substantially with existing Big Five traits and shows less incremental validity.
• The construct validity of EI measures remains debated, and the field has been criticized for over-promising and under-delivering.

Integrity Tests

Integrity tests are designed to predict counterproductive work behaviors such as theft, absenteeism, and disciplinary problems. Two types exist:

• Overt integrity tests — directly ask about attitudes toward dishonesty and admissions of past dishonest behavior.
• Personality-based integrity tests — use items related to conscientiousness, agreeableness, and emotional stability without directly referencing dishonesty.

Meta-analyses show that integrity tests have moderate validity for predicting CWB (\( \rho \approx .32 \)) and modest validity for predicting overall job performance (\( \rho \approx .21 \)). They produce minimal adverse impact, making them an attractive supplement to cognitive ability tests.

Chapter 10: Predictor Methods

Overview

While the previous chapter discussed the constructs measured in selection, this chapter covers the methods — the specific tools and procedures used to assess candidates. The key distinction is that a single construct (e.g., interpersonal skills) can be measured by multiple methods (e.g., interview, assessment center exercise, situational judgment test), and a single method can measure multiple constructs.

The Interview

The employment interview is the most widely used selection method worldwide. However, its effectiveness depends critically on how it is conducted.

Unstructured Interviews

In an unstructured interview, the interviewer has broad latitude to ask whatever questions come to mind, follow up on responses at will, and evaluate candidates holistically. Meta-analytic research shows that unstructured interviews have relatively low validity (\( \rho \approx .20 \)) and poor inter-rater reliability. They are susceptible to:

• First impressions and confirmation bias.
• Similarity bias — preferring candidates who are similar to the interviewer.
• Contrast effects — evaluating candidates relative to previous interviewees rather than against a standard.
• Halo effects — letting one positive or negative attribute color the entire evaluation.

Structured Interviews

Two common types of structured interview questions:

• Behavioral description interviews (BDI) — ask candidates to describe what they actually did in past situations (e.g., “Tell me about a time when you had to deal with a difficult customer.”).
• Situational interviews (SI) — present hypothetical job-related scenarios and ask candidates what they would do (e.g., “What would you do if a coworker was not pulling their weight on a team project?”).

Meta-analytic validity for structured interviews is substantially higher than for unstructured interviews (\( \rho \approx .44 \) vs. \( .20 \)), and structured interviews produce less adverse impact than cognitive ability tests.

Assessment Centers

Assessment centers have good criterion-related validity (\( \rho \approx .36 \) for predicting job performance, higher for predicting advancement). However, research has raised questions about their construct validity: ratings tend to be more strongly associated with exercises than with dimensions, suggesting that assessors may be evaluating exercise-specific performance rather than underlying competencies.

Work Sample Tests

Work sample tests require candidates to perform tasks that are actual components of the job. For example, a typing test for an administrative position or a coding exercise for a software developer position. Work sample tests have high validity (\( \rho \approx .33 \)–\( .54 \)), high face validity (candidates perceive them as fair and job-related), and lower adverse impact than cognitive ability tests. Their main disadvantage is that they are expensive to develop, require job-specific knowledge, and cannot be used for entry-level positions where candidates are not expected to have the relevant skills.

Situational Judgment Tests (SJTs)

SJTs have moderate validity (\( \rho \approx .26 \)), can be administered efficiently to large groups, and produce less adverse impact than cognitive ability tests. The validity of SJTs depends on the constructs they are designed to measure and the response instructions (knowledge-based: “What should you do?” vs. behavioral tendency: “What would you do?”).

Biographical Data (Biodata)

Biographical data (biodata) instruments collect information about candidates’ past experiences, behaviors, and accomplishments through structured questionnaires. The underlying principle is that past behavior is the best predictor of future behavior. Biodata instruments have shown good validity (\( \rho \approx .35 \)) but can raise concerns about:

• Transparency — items may seem invasive or irrelevant.
• Fakability — candidates may exaggerate or fabricate experiences.
• Fairness — some items may reflect socioeconomic or cultural differences rather than job-relevant characteristics.

Highhouse et al. (2019): Dark Motives and Elective Use of Brainteaser Interview Questions

Chapter 11: Selection Decision-Making

Combining Predictor Information

When multiple predictors are used in selection, organizations must decide how to combine the information to make hiring decisions. There are several major approaches:

Multiple Regression (Compensatory Model)

In a compensatory model (typically implemented through multiple regression), a high score on one predictor can compensate for a low score on another. Predictor scores are weighted (by their regression weights) and summed to produce a composite score, and candidates are ranked by their composite.

Advantages: maximizes predicted performance. Disadvantages: may allow candidates with dangerously low scores on a critical predictor to be selected.

Multiple Cutoff (Non-Compensatory Model)

In a multiple cutoff approach, minimum scores are set on each predictor, and candidates must meet or exceed all cutoffs to be considered. Candidates who pass all cutoffs may then be ranked on one or more predictors or selected randomly.

Advantages: ensures minimum qualifications on all critical dimensions. Disadvantages: requires setting cutoff scores, which involves judgment and has implications for adverse impact.

Multiple Hurdle

The multiple hurdle approach is a sequential, non-compensatory model in which candidates must pass each predictor stage before being assessed on the next. This approach is efficient because it allows organizations to screen out unqualified candidates early using cheaper methods (e.g., application blanks, biodata) before investing in more expensive assessments (e.g., interviews, assessment centers).

Profile Matching

In profile matching, the ideal profile of predictor scores is established based on job analysis or the profiles of successful incumbents, and candidates are selected based on the similarity of their profiles to the ideal.

Banding

Banding has been controversial. Proponents argue that it acknowledges measurement error and allows organizations to pursue diversity goals without completely abandoning merit-based selection. Critics argue that it undermines the meritocratic principle and may reduce the predicted performance of the selected group.

Adverse Impact

The four-fifths (80%) rule is a common guideline (originating in U.S. law but used informally in Canada): adverse impact may be indicated when the selection rate for a protected group is less than 80% of the selection rate for the majority group.

Strategies for Reducing Adverse Impact

Several strategies have been proposed to reduce adverse impact while maintaining validity:

1. Use predictor composites — combining cognitive and non-cognitive predictors (e.g., GMA + personality + structured interview) can reduce adverse impact relative to using GMA alone.
2. Use alternative predictors — SJTs, structured interviews, and biodata produce less adverse impact than cognitive ability tests.
3. Banding — treating similar scores as equivalent.
4. Within-group norming — comparing candidates to their own demographic group (note: this is legally prohibited in the U.S. but the Canadian legal landscape is more nuanced).
5. Adjusting the selection ratio — hiring more candidates from the top of the score distribution reduces the impact of group differences.

Chapter 12: Legal and Ethical Issues

The Canadian Legal Framework

Personnel selection in Canada operates within a multi-layered legal framework:

Canadian Charter of Rights and Freedoms

Section 15(1) of the Charter guarantees that every individual is equal before and under the law and has the right to the equal protection and equal benefit of the law without discrimination. While the Charter applies directly only to government action, it has influenced the interpretation of human rights legislation that applies to private employers.

Canadian Human Rights Act

The Canadian Human Rights Act (CHRA) prohibits discrimination in employment by federally regulated employers on the following grounds: race, national or ethnic origin, colour, religion, age, sex, sexual orientation, gender identity or expression, marital status, family status, genetic characteristics, disability, and conviction for which a pardon has been granted or a record suspension has been ordered.

Provincial and territorial human rights codes provide similar protections for provincially regulated workplaces, with some variation in the specific grounds covered.

Employment Equity Act

The Employment Equity Act (1995) requires federally regulated employers and federal contractors to:

1. Identify and eliminate employment barriers against persons in four designated groups: women, Aboriginal peoples, persons with disabilities, and members of visible minorities.
2. Institute positive policies and practices to achieve a degree of representation in the workforce that reflects the representation of these groups in the Canadian workforce or in relevant segments of the workforce.
3. File annual reports on workforce composition.

Employment equity differs from affirmative action (as practiced in the U.S.) in its emphasis on systemic change and removal of barriers rather than on numerical targets or quotas.

Bona Fide Occupational Requirement (BFOR)

The Supreme Court of Canada established the modern test for BFOR in British Columbia (Public Service Employee Relations Commission) v. BCGSEU (1999), commonly known as the Meiorin decision. The three-part test requires the employer to show:

1. Rational connection — the standard was adopted for a purpose rationally connected to the performance of the job.
2. Good faith — the standard was adopted in an honest and good-faith belief that it was necessary for the fulfilment of that legitimate work-related purpose.
3. Reasonable necessity — the standard is reasonably necessary to the accomplishment of that work-related purpose. The employer must show that it is impossible to accommodate the affected individual or group without imposing undue hardship on the employer.

Duty to Accommodate

The third prong of the Meiorin test incorporates the duty to accommodate — the obligation of employers to make reasonable adjustments to workplace rules, policies, or conditions to eliminate or reduce the discriminatory effect on individuals with protected characteristics, short of undue hardship.

Factors considered in assessing undue hardship include financial cost, disruption to operations, and health and safety risks. The duty to accommodate is a fundamental principle of Canadian human rights law and has direct implications for selection: if a selection test has adverse impact on a protected group, the employer must consider whether accommodations (e.g., alternative testing formats for persons with disabilities, scheduling accommodations for religious observances) can reduce the impact.

Ethical Considerations

Beyond legal requirements, personnel psychologists are guided by professional ethics codes, including the Canadian Psychological Association’s Canadian Code of Ethics for Psychologists and the SIOP Principles for the Validation and Use of Personnel Selection Procedures. Key ethical principles include:

• Informed consent — candidates should understand what they are being assessed on and how the information will be used.
• Confidentiality — test results should be shared only with authorized decision-makers.
• Fairness — selection procedures should be job-related and should not unfairly disadvantage any group.
• Competence — personnel psychologists should work within the boundaries of their training and expertise.
• Right to feedback — candidates should be able to receive information about their performance on selection assessments.

Privacy Legislation

Canadian privacy legislation — including the Personal Information Protection and Electronic Documents Act (PIPEDA) at the federal level and provincial equivalents — governs the collection, use, and disclosure of personal information in the employment context. Employers must:

• Collect only information that is necessary and relevant to the employment decision.
• Obtain consent for the collection and use of personal information.
• Protect personal information from unauthorized access.
• Allow individuals to access their own personal information.

These requirements affect practices such as background checks, drug testing, social media screening, and the retention of application and assessment data.

Chapter 13: Recruitment

What Is Recruitment?

Recruitment Sources

Organizations use a variety of channels to reach potential applicants:

• Internal recruitment — posting positions to current employees, encouraging internal transfers and promotions. Advantages include lower cost, knowledge of the candidate’s track record, and positive effects on employee motivation. Disadvantages include a limited pool and potential for inbreeding of ideas.
• External recruitment — job postings on websites, social media, job fairs, campus recruiting, employment agencies, professional associations, and employee referrals.
• Employee referrals — current employees recommend candidates from their personal networks. Research shows that referred applicants tend to have lower turnover and, in some studies, better performance than applicants from other sources, possibly because referrers provide realistic information about the job and because social networks create accountability.

Realistic Job Previews (RJPs)

The theoretical rationale for RJPs is based on the met expectations hypothesis: when employees’ pre-hire expectations match their post-hire experiences, they report higher satisfaction and are less likely to quit. By giving applicants realistic information upfront, RJPs allow those who would be a poor fit to self-select out, reducing later turnover.

Meta-analytic evidence shows that RJPs have a modest but reliable effect on reducing turnover (\( d \approx .12 \)). The effect is stronger when:

• The RJP is given early in the recruitment process.
• The RJP uses a vivid medium (e.g., video or job shadowing rather than a written brochure).
• The information is credible and specific.

Employer Branding

Employer branding refers to the organization’s efforts to promote itself as a desirable place to work. Drawing on marketing principles, employer branding involves communicating the organization’s values, culture, benefits, and career development opportunities to potential applicants.

Research on employer branding draws on signaling theory — the idea that applicants use observable attributes of the organization (its reputation, recruitment practices, website quality) as signals of unobservable attributes (what it would be like to work there). Organizations that convey a clear, positive, and authentic employer brand attract more and better applicants.

Applicant Reactions

Research has found that applicant reactions affect:

• Applicant withdrawal — candidates who perceive the process as unfair are more likely to drop out.
• Offer acceptance — negative perceptions reduce the likelihood that selected candidates will accept job offers.
• Organizational attractiveness — applicants who have a positive experience are more likely to recommend the organization to others and to reapply in the future.
• Legal challenges — candidates who perceive unfairness may be more likely to file complaints or lawsuits.

Selection methods perceived as most fair include interviews, work sample tests, and resumes. Methods perceived as least fair include graphology, drug tests, and personality tests (although these perceptions vary across cultures and contexts).

Chapter 14: Utility Analysis

What Is Utility Analysis?

The basic logic of utility analysis is straightforward: if a selection system is valid, it enables the organization to hire employees who perform at a higher level than those who would be hired without the system. This improved performance has economic value that can be estimated in dollar terms.

The Taylor-Russell Model

The Taylor-Russell model (Taylor & Russell, 1939) was one of the earliest approaches to utility analysis. It estimates the proportion of successful employees (those who perform above a defined criterion) as a function of three variables:

1. Validity (\( r_{xy} \)) — the correlation between the predictor and the criterion.
2. Selection ratio (\( SR \)) — the proportion of applicants who are hired.
3. Base rate (\( BR \)) — the proportion of applicants who would be successful if selected randomly.

The model uses the Taylor-Russell tables to determine the success ratio — the proportion of selected applicants who will be successful — given these three inputs.

Key insights from the Taylor-Russell model:

• Utility increases as validity increases.
• Utility increases as the selection ratio decreases (more selective hiring).
• Utility is greatest when the base rate is moderate (around .50) — when almost everyone or almost no one would succeed, selection adds little value.

Limitations: the model requires a dichotomous criterion (success/failure) and does not estimate utility in dollar terms.

The Brogden-Cronbach-Gleser Model

The Brogden-Cronbach-Gleser (BCG) model estimates the dollar value of a selection system using the following equation:

where:

The critical and most controversial parameter is \( SD_y \) — the standard deviation of job performance expressed in dollars. Several estimation methods have been proposed:

• Cost accounting — directly measuring the dollar value of output for each employee (feasible only for some jobs).
• The 40% rule (Schmidt & Hunter) — estimating \( SD_y \) as 40% of the average salary for the job.
• Global estimation — asking supervisors to estimate the dollar value of performance at different percentile levels.

Practical Issues in Utility Analysis

Despite the compelling logic of utility analysis, its adoption by organizations has been limited. Several practical and conceptual issues contribute to this gap:

1. Estimation of \( SD_y \) — estimates vary dramatically depending on the method used, and decision-makers may distrust the resulting dollar figures.
2. Assumptions — the BCG model assumes linear relationships, no interactions among predictors, and stable parameters over time.
3. Managerial skepticism — decision-makers may find the large dollar figures produced by utility analysis implausible, leading to rejection of the analysis rather than adoption of the selection system.
4. Sensitivity to assumptions — small changes in input parameters can produce large changes in estimated utility.

Research has suggested that utility analysis may be more persuasive when results are presented in terms of effect sizes or productivity gains rather than raw dollar figures, and when the analysis is supplemented with other forms of evidence (e.g., face validity, stakeholder support).

ROI of Selection Systems

More broadly, the return on investment of a selection system depends on:

• The validity of the system (higher validity = greater ROI).
• The selection ratio (more selective hiring = greater ROI per hire, but potentially higher recruitment costs to generate a larger applicant pool).
• The cost of the selection system relative to the value it adds.
• The base rate of success (selection adds more value when the base rate is moderate).
• Tenure (the benefits accrue for as long as the employee remains with the organization).

These factors interact, and a comprehensive analysis must consider all of them. The key takeaway is that valid selection is not just a fairness issue — it is an economic imperative. Organizations that use evidence-based selection methods gain a substantial competitive advantage through improved workforce quality.

Summary and Integration

Personnel psychology provides a scientific foundation for managing the human side of organizations. The topics covered in this course — from job analysis and performance measurement to selection, legal issues, recruitment, and utility analysis — form an integrated system:

1. Job analysis defines what the job requires.
2. Performance criteria define what success looks like.
3. Measurement principles (reliability, validity) ensure that our tools are sound.
4. Predictor constructs and methods assess candidates on job-relevant attributes.
5. Decision-making models combine predictor information to make hiring decisions.
6. Legal and ethical frameworks ensure that selection is fair and compliant.
7. Recruitment generates the applicant pool.
8. Utility analysis quantifies the economic value of the system.

Each component depends on the others. A selection system built on a poor job analysis will measure the wrong things. A valid test administered unfairly will create legal liability. A brilliant selection system paired with poor recruitment will fail because the applicant pool is weak. The scientist-practitioner model — with its dual commitment to rigorous evidence and practical application — ties these elements together and provides the intellectual foundation for the field.