I/O Psychology: Measurement, Statistics, Recruitment, and Validity

Comprehensive vocabulary flashcards covering statistical measurement scales, distribution properties, regression modeling, recruitment strategies, turnover predictors, and test validity concepts in Industrial/Organizational Psychology.

Measurement

The assignment of numbers to properties of objects or events, reflecting specific quantifiable aspects rather than the entire entity itself.

Construct (Latent Variable)

An unobservable psychological concept or attribute (symbolized by circles) assumed to cause variation in observable indicators.

Indicator (Manifest Variable)

An observable, measurable variable (symbolized by squares), such as a survey response score, designed to assess an underlying latent construct.

Nominal Scale

A measurement scale that classifies data into mutually exclusive qualitative categories without any inherent quantitative order (e.g., gender or race).

Ordinal Scale

A measurement scale where numbers reflect rank-ordering (e.g., performance rankings), but intervals between adjacent ranks are unknown or unequal.

Interval Scale

A measurement scale where distances between values are known, equal, and meaningful (permitting addition and subtraction), but without an absolute zero point.

Ratio Scale

A measurement scale featuring equal intervals and an absolute, meaningful zero point, allowing values to be expressed as multiples or ratios of each other.

Standard Deviation (SD)

A measure of variability representing the square root of variance, calculated as $s = \sqrt{\frac{\sum_{i=1}^{N} (X_i - \bar{X})^2}{N - 1}}$, which describes the spread of data around the mean.

Positive Skew

An asymmetrical distribution curve stretched toward the right, where the tail points toward higher values and $\text{Mean} > \text{Median} > \text{Mode}$.

Negative Skew

An asymmetrical distribution curve stretched toward the left, where the tail points toward lower values and $\text{Mean} < \text{Median} < \text{Mode}$.

Kurtosis

The degree of peakedness of a distribution, distinguishing sharply peaked curves with heavy tails (leptokurtic) from flatter curves (platykurtic).

Pearson Product-Moment Correlation Coefficient ($r$)

A standardized statistic ranging from $-1.00$ to $+1.00$ that represents the strength and direction of a linear relationship between two variables.

Range Restriction

A condition occurring when the study sample has less variability than the population of interest, which typically reduces the magnitude of observed correlation coefficients.

Simple Linear Regression

A statistical procedure represented by $Y = b_0 + b_1 X_1 + e$, used to predict values of an outcome variable ($Y$) based on a single predictor variable ($X_1$).

Coefficient of Determination ($R^2$)

The proportion of variance in the outcome variable predictable from the predictor variable(s), calculated as $R^2 = 1 - \frac{Var(e)}{Var(Y)}$.

Multiple Linear Regression

A regression model represented by $Y = b_0 + b_1 X_1 + b_2 X_2 + e$, estimating the unique contribution of each predictor while holding other predictors constant.

p-value

The probability of obtaining an observed sample result or correlation given that the true relationship between variables in the population is zero under the null hypothesis.

Confidence Interval (CI)

A calculated range of values within which the true population parameter is expected to fall with a specified degree of certainty (typically $95\%$).

Recruitment

The organizational process of identifying and attracting a qualified pool of potential job applicants.

Time to Hire

A recruitment efficiency metric tracking the average duration in days across milestones from resume submission to formal job start.

Gross Cost per Hire

An efficiency metric calculated as total recruiting expenditures divided by the total number of individuals hired.

Job Embeddedness

The extent to which an employee's job connects to important non-work aspects of life, comprising organizational/community fit, links, and sacrifices.

Job Satisfaction

A multifaceted work attitude reflecting employee evaluations of pay, promotion opportunities, supervision, coworkers, and the work itself.

Affective Commitment

An organizational commitment component characterized by positive emotional attachment and identification with the organization.

Continuance Commitment

An organizational commitment component driven by the perceived heavy economic, social, or psychological costs associated with quitting.

Normative Commitment

An organizational commitment component defined by perceived moral or ethical obligation to remain with the employer.

Closed Internal Recruitment

An internal staffing system where open positions are kept confidential from general employees and revealed solely to candidates selected by HR.

Open Internal Recruitment

An internal staffing procedure where all job openings are publicized to all employees within the organization.

Yield Ratio Funnel

A graphical representation of applicant throughput tracking the percentage of applicants successfully moving between successive recruitment stages.

Attraction-Selection-Attrition (ASA) Model

Benjamin Schneider's framework explaining how organizations naturally become homogeneous because similar people apply, are chosen, and non-fitting members leave.

Realistic Job Preview (RJP)

Recruitment materials or presentations providing candidates with balanced, accurate information highlighting both favorable and unfavorable aspects of a job.

Process Fairness

Applicant perceptions regarding the procedural equity, job relatedness, consistency, and respectfulness of the selection process.

Outcome Fairness

Applicant perceptions concerning the fairness of the final hiring or rejection decision.

Validity

The degree to which empirical evidence and theoretical rationales support the adequacy and appropriateness of inferences drawn from assessment scores.

Reliability

The consistency and repeatability of test scores across time, raters, or forms; mathematically establishes the upper ceiling for validity.

Content Validity

The extent to which test items accurately represent and theoretically link to the specific behaviors, knowledge, or dimensions required for job performance.

Substantive Validity Coefficient ($C_{SV}$)

An index evaluating content validity: $C_{SV} = \frac{n_c - n_0}{N}$, where $n_c$ is posited construct assignments and $n_0$ is the highest competing construct assignment.

Criterion-Related Validity

The degree of empirical, statistically significant association between predictor test scores and an outcome measure (such as job performance).

Predictive Validity

A criterion validation design where applicant test scores are gathered prior to hiring and subsequently correlated with performance measured after a tenure period (typically $\sim 6\text{ months}$).

Concurrent Validity

A criterion validation design in which predictor scores and performance criterion measures are collected simultaneously from current job incumbents.

Correction for Attenuation

A statistical adjustment formula, $r_{XT_y} = \frac{r_{XY}}{\sqrt{r_{YY}}}$

Used to estimate what criterion-related validity would be if the criterion were measured without error.

Situational Specificity Hypothesis

The view that predictor validity varies idiosyncratically across different organizations and settings, largely refuted by validity generalization meta-analyses.

Validity Generalization

A meta-analytic approach developed by Schmidt and Hunter demonstrating that selection test validity generalizes across settings when statistical artifacts are removed.

Construct Validity

Evidence demonstrating that test scores align with theoretical expectations regarding the internal structure of the construct and its relationships with external constructs.

Convergent Validity

The degree to which a new measure correlates strongly ($r = .50\text{ to }.80$) with existing measures assessing the exact same theoretical construct.

• A new measure of a construct should be strongly correlated with existing measures of the same construct.

Discriminant Validity

The degree to which a test shows lower or non-significant correlations with existing measures of conceptually distinct constructs.

Factor Analysis

A statistical method used to evaluate construct validity by testing whether observed indicators correlate as expected with underlying latent factors.

Factor Loading

The standardized regression coefficient linking a latent factor to an observed indicator, with ideal values typically falling between $0.40$ and $0.80$.

Multitrait-Multimethod Matrix (MTMM)

A Campbell and Fiske framework measuring multiple traits using multiple methods to isolate construct variance from common method bias.

Consequential Validity

The assessment of both intended consequences (e.g., hiring top performers) and unintended consequences (e.g., adverse impact) of test use.

Utility Analysis

The process of translating selection validity coefficients and performance gains into direct bottom-line dollar revenue value for an organization.

Taylor-Russell Approach

A selection model predicting the proportion of successful hires using a test, based on the validity coefficient, selection ratio, and base rate.

Selection Ratio

The mathematical proportion of total applicants hired for a job opening relative to the total applicant pool size.

Base Rate

The proportion of applicants who would perform adequately on the job if hiring decisions were made entirely at random.

Nominal Data Operations

For nominal data, it only makes sense to compute percentages in each category, not means or standard deviations because category order is arbitrary.

Minimum Scale for Statistics

You need at least an interval scale to compute means, standard deviations, and correlations.

Zero Correlation vs. Independence

Zero correlation does not imply independence

Correlation Magnitude vs. Importance

Correlation magnitude does not necessarily equal level of importance (e.g., small medical correlations like aspirin use and heart attack can be critically important).

Cost of a Bad Hire

A single bad hire costs companies an average of $14,900, amounting to roughly 30 percent of the hire's annual salary.

New Hire Break-Even Timeline

It generally takes at least 6 months for companies to break even on new hires.

Internal vs. External Hires (Bidwell 2011)

External hires have more education/experience and higher starting pay, but internal hires perform at a higher level and experience lower turnover.

External Recruitment Sources Rule of Thumb

Informal sources (such as employee referrals) typically outperform formal recruitment methods.

Pay-to-Quit Programs

A retention strategy used by companies that relies on cognitive dissonance.

Standard Corporate Hiring Yield Ratios

Corporate hiring funnels average a 14:1 resume-to-serious-candidate ratio, a 4:1 serious-candidate-to-interview ratio, and a 5:2 interview-to-hire ratio.

Rule for Describing Validity

We should never say a test is globally valid or invalid

Why Validity Matters

It ensures a test measures what it is supposed to, test scores relate to other variables in predictable ways, and test scores are reliable.

Validity Degradation

The validity of test scores drops as the time interval between testing and performance evaluation increases.

Test Motivation in Validation

Job incumbents in concurrent validation studies may not approach tests with the same anxiety or motivation as real job applicants.

Discriminant Validity (Similar Constructs)

A new measure should be moderately correlated ($r = .30$ to $.50$) with similar constructs (e.g., job satisfaction and organizational commitment).

Discriminant Validity (Unrelated Constructs)

A new measure should be weakly correlated ($r = 0$ to $.30$) with unrelated constructs (e.g., job satisfaction and political beliefs).

IQ Score Scale Limitation

IQ scores are measured on an interval scale, which is why it is incorrect to say an IQ of 200 is twice as intelligent as an IQ of 100 because it lacks an absolute, meaningful zero point.

Normal Distribution Percentile

In a normal distribution, a person who is one standard deviation above the mean falls above approximately 84 percent of the people in the sample.

Multiple Regression Interpretation

In multiple regression, the slopes represent the unique contribution of each predictor while holding the other predictors constant.

Big Five Correlations with Job Satisfaction

Job satisfaction correlates with Neuroticism (r = -.29), Conscientiousness (r = .26), Extraversion (r = .25), and Agreeableness (r = .17).

Fit Types and Job Satisfaction Ranking

Person-job fit has the strongest relationship with job satisfaction (r = .56), followed by person-organization and person-supervisor (r = .44).

Job Satisfaction vs. Turnover Intention

Job satisfaction correlates more strongly with turnover intention (r = -.58) than with actual turnover (r = -.22).

Job Embeddedness Subfactors

Comprises fit, links, and sacrifice, evaluated across both community and organization domains.

Reliability Ceiling on Validity

Reliability places a hard ceiling on validity, meaning validity cannot exceed reliability.

Taylor-Russell Inputs

The validity coefficient, the selection ratio, and the base rate.

Predictor

Any manifest measure whose scores represent individual differences on latent knowledge, skills, abilities, or other characteristics linked to effective criterion performance, and which is used as the basis for selection or promotion.

Model of Predictor Response Process

A framework illustrating how latent individual differences, contaminating sources of variance, and test-taking motivation interact during comprehension to produce a response.

Constructs vs. Measurement Methods

A comparison framework looking at how different underlying traits are assessed using various measurement formats.

Psychological Fidelity

The extent to which a predictor taps into the same KSAOs required for criterion performance, which is considered more important than physical similarity.

Physical Fidelity

The extent to which a predictor is physically similar to the criterion, roughly corresponding to the measurement methods used.

Proximal Predictors

Predictors that assess KSAOs most directly related to performance and are more state-like constructs such as motivation and job knowledge.

Distal Predictors

Predictors related to criterion performance but not as immediately, consisting of more stable, trait-like individual differences such as general intelligence and personality.

Four Main Considerations for a Good Predictor

Criterion-related validity, user acceptability, utility, and subgroup differences.

Subgroup Differences

Mean predictor score disparities between demographic categories.

Cohen's d

The standardized effect size calculated as the majority group mean minus the minority group mean divided by the pooled standard deviation.

Cohen's d Effect Size Benchmarks

Small equals 0.20, medium equals 0.50, and large equals 0.80.

Predictor Battery

A selection system consisting of multiple predictors that, when well-developed, always outperforms any single predictor.

Regression-Based Weighting

Combining predictors using regression weights to achieve maximum validity without explicitly considering subgroup differences.

Subjective Weighting

Assigning weights subjectively, which can cause some loss of validity but may help reduce subgroup differences.

Compound Predictors

Variables consisting of multiple traits compounded together yielding a single/composite score, such as emotional intelligence or core self-evaluations.

Pure Top-Down Selection

Ranking all applicants by test scores and selecting from the top down based strictly on the number of open positions.

Cut-Off Scores

Using a regression-derived score threshold where applicants must score above it to be considered.

Multiple Cut-Offs Decision Model

A selection system where applicants take multiple tests and must pass thresholds across all of them, which can be expensive.

Predictors in Multilevel Contexts

Staffing considerations where individual-level KSAOs may not be optimal for group-level staffing, requiring evaluation of whether team homogeneity or heterogeneity impacts performance through team mediating processes like cohesion, mental models, and coordination.

Cultural Influences on Testing

Differences across cultures regarding factor structure (e.g., variations of the Big Five), predictive validity, and user reactions.