Apa Handbook Io V2

1. 1Measure the construct concretely, not abstractly: adopt Huselid's (1995) operational item — the proportion of the workforce whose job has been subjected to a formal job analysis — and track it as a baseline organizational metric.
2. 2Audit job-description quality using Delery and Doty's (1996) three markers: duties are clearly defined, the description is up to date, and it contains all duties actually performed by employees. Score each role against these.
3. 3Bundle rather than isolate: following Toh, Morgeson, and Campion (2008), count the number of selection systems based on formal work analyses and the number of training programs incorporating systematic work analysis, so context supports HR practices as a coherent set.
4. 4Link the supportive context to organizational outcomes named in the source — reduced turnover, increased sales, profits, and market value (Delaney & Huselid, 1996; Huselid, 1995) — and specify which HPWP bundle drives which outcome.
5. 5Widen the unit of analysis from the individual job to role, team, and organization, because jobs are situated in a larger team and organizational context and focusing only on incumbents leaves an insufficient link to the organization.
6. 6Test context-dependence before importing a practice: check whether a benefit found in one setting (e.g., mentoring in Anglo-Saxon environments) holds in yours, since Bozionelos (2006) shows the same practice can fail on subjective outcomes where trust of authority is low.
7. 7Re-run the audit periodically to counter the static-jobs criticism (Guion, 1993): capture emergent and discretionary task elements, not only prescribed ones, so the supportive context tracks how roles actually change.

A firm wants to strengthen its supportive context and starts with Huselid's (1995) measure: only 40% of its workforce has a job subjected to formal job analysis. Auditing descriptions against Delery and Doty (1996), managers find most fail 'contains all of the duties performed by individual employees' — the descriptions are stale. The firm applies the Toh, Morgeson, and Campion (2008) bundling logic: it rebuilds selection systems on formal work analyses and rewrites training programs to incorporate systematic analysis, treating these as one coherent bundle rather than separate fixes. It then tracks the outcomes the seminal research associates with HPWP — turnover, sales, profit, market value (Delaney & Huselid, 1996). Because the firm operates partly in a non-Anglo-Saxon culture, it does not assume a mentoring program will help subjective outcomes, citing Bozionelos (2006). The result: the supportive context is now measurable, bundled, and outcome-linked rather than aspirational.

• If less than a defined proportion of the workforce has undergone formal job analysis, then treat the supportive context as underbuilt and prioritize analysis coverage before layering new HR practices.
• If a job description fails any of the three Delery-Doty tests (clearly defined, up to date, all duties), then flag the role as an unsupported foundation for selection and training.
• If you are deploying HR practices, then bundle them (selection + training grounded in work analysis) rather than adopting them in isolation, since HPWP effects come from coherent bundles.
• If a practice's benefit was established in one cultural context (e.g., mentoring in Anglo-Saxon settings), then validate it locally before assuming transfer, because subjective benefits can reverse where authority is distrusted.
• If your analysis focuses only on the individual incumbent, then expand it to role, team, and organization or the link between the practice and organizational performance stays insufficient.

• Treating supportive context as a slogan instead of a metric — the tell: no one can state the proportion of the workforce with a formal job analysis.
• Fixing HR practices one at a time — the tell: new selection tests and training programs share no common work-analysis foundation and outcomes don't move.
• Assuming universal transfer of a supportive practice — the tell: importing mentoring or a development program without checking cultural fit, then seeing objective gains but flat or negative subjective/adjustment measures (Bozionelos, 2006).
• Viewing jobs as static — the tell: descriptions capture only prescribed duties and never emergent or discretionary tasks, so context drifts out of date (Guion, 1993).
• Anchoring only on the individual job — the tell: analysis ignores the embeddedness of roles in team and organizational context, breaking the chain to organizational performance.

• ▸Formal job-analysis coverage: % of workforce whose job has had a formal analysis (Huselid metric).
• ▸Description quality score: rate each role on 'clearly defined / up to date / all duties included' (Delery-Doty).
• ▸Selection bundle count: number of selection systems based on formal work analyses.
• ▸Training bundle count: number of training programs incorporating systematic work analysis.
• ▸Unit of analysis check: is the context assessed at individual, role, team, and organizational levels? (Y/N each).
• ▸Emergent-task capture: are discretionary/emergent task elements recorded, not just prescribed ones? (Y/N).
• ▸Cultural-fit flag: was each imported practice validated in the local context before rollout? (Y/N).
• ▸Outcome link: which target outcome (turnover, sales, profit, market value) each bundle is expected to move.
• ▸Re-audit date: next scheduled review to prevent static-job drift.

1. 1Frame applicant reactions as fairness judgments first: the passages state that applicant reactions to selection procedures can be explained largely in terms of perceived fairness or justice. Start every design review by asking which justice dimension (procedural, distributive, interactional) a given step touches.
2. 2Audit each screening step for procedural justice signals. Since applicants generally accept the use of cognitive ability tests, do not assume acceptance problems where none exist — but confirm perceived job-relatedness before defending a procedure that lacks it.
3. 3Separate perceived fairness from actual dropout. The passages note there are sometimes differences in perceived test fairness across demographic groups, yet little evidence that testing causes job seekers to drop out of pools. Measure both perception and behavior; treat them as distinct outcomes.
4. 4For affirmative action or diversity-linked procedures, minimize negative reactions by placing strong emphasis on merit (e.g., affirmative action as a tiebreaker) and explaining the reasons behind the policy — the two moves the passages identify as reducing backlash.
5. 5Segment applicants by quality and experience. High-quality applicants (high grades, many offers) are more critical of recruiting practices such as recruiters and delays; those with greater work experience may be slightly more forgiving. Calibrate expectations and communication cadence to these segments.
6. 6Remove or explain recruiting delays. In campus recruiting contexts delays are flagged as a reaction risk; treat cycle time between stages as a fairness variable, not just an operations metric.
7. 7Manage interactional justice through recruiter behavior: applicants with positive preinterview beliefs perceive recruiter behaviors more positively, so recruiter conduct both shapes and is shaped by prior impressions — brief recruiters accordingly.

A firm uses a cognitive ability test as its first screen and adds an affirmative-action policy for final hiring decisions. Reviewing against the passages, the team confirms procedural acceptance is likely intact — applicants generally accept cognitive ability tests — so they leave the test in place. Because white males tend to react negatively to affirmative action while beneficiaries react positively, the team reframes the policy as a tiebreaker applied only after merit criteria and publishes the rationale in the candidate portal, the two mitigations the source names. They then track two separate metrics: perceived test fairness across demographic groups and actual dropout rate, knowing from the passages that perceived-fairness gaps do not reliably translate into applicants leaving the pool. For their campus channel, they cut the gap between interview and offer because delays and recruiter behavior are the reaction drivers there, and they brief recruiters knowing high-quality candidates with multiple offers judge these practices most harshly.

• If a procedure draws negative reactions, then diagnose which justice type is failing (procedural job-relatedness, distributive outcome, or interactional treatment) before changing the procedure itself.
• If you introduce affirmative action or preference policies, then emphasize merit and explain the rationale — otherwise expect negative reactions concentrated among non-beneficiaries.
• If perceived fairness differs across demographic groups but dropout does not, then prioritize the perception and communication problem rather than assuming you are losing applicants.
• If your applicant pool is high quality (strong grades, multiple offers), then tighten recruiter professionalism and cut delays, because these applicants are more critical of recruiting practices.
• If applicants show positive preinterview beliefs, then expect more favorable recruiter-behavior perceptions — do not over-attribute good reactions solely to the interview itself.

• Assuming acceptance of any test because cognitive ability tests are generally accepted; the tell is defending an unrelated or opaque procedure with 'applicants like tests' rather than evidence of perceived job-relatedness.
• Treating a demographic gap in perceived fairness as proof of dropout; the tell is a policy change justified by perception data with no behavioral attrition data attached.
• Implementing preference policies silently; the tell is a spike in negative reactions from non-beneficiaries because merit emphasis and rationale were never communicated — the two mitigations the source specifies.
• Applying one reaction-management standard to all applicants; the tell is losing top candidates who, being high quality, judge recruiter conduct and delays more harshly than average applicants.
• Ignoring recruiting cycle time as a fairness factor; the tell is stage-to-stage delays treated purely as logistics while campus candidates react negatively to the wait.

• ▸Selection step: name the procedure or stage being evaluated
• ▸Justice type touched: procedural / distributive / interactional (one line)
• ▸Perceived job-relatedness: is the step's relevance visible to applicants? (yes/no + note)
• ▸Preference-policy present?: if yes, list merit-emphasis and rationale statements used
• ▸Perceived-fairness gap by group: record any demographic differences observed
• ▸Behavioral dropout: actual attrition rate at this step (separate from perception)
• ▸Applicant segment: quality/experience profile most affected by this step
• ▸Cycle-time/delay: elapsed time to next stage and reaction risk flag
• ▸Recruiter/interaction touchpoint: who interacts, and briefing status

1. 1Define both sides of the comparison before assessment: name the individual characteristics you care about (values, personality, goals) and the organizational characteristics they must match (culture, values, structure). Fit is compatibility between these two sets, so specify each.
2. 2Treat fit as multidimensional. Following Mitchell and Lee's (2001) embeddedness logic, separate the compatibility judgment into distinct components — here, 'fit' means compatibility or comfort with both the work AND external environments, so assess job-side and community-side alignment separately.
3. 3Decide whether you are measuring fit as a latent cause of scale scores or as a formative construct defined by its indicators. Job embeddedness (which includes a fit dimension) is defined by its indicators — measures of fit, links, and sacrifice in combination 'cause' embeddedness, unlike traditional constructs where the construct causes the scores.
4. 4Build a screening instrument that surfaces misfit directly. Use a forced-choice format like Villanova et al.'s (1994) Job Compatibility inventory: present item sets, have candidates select the situations most discomforting to them, and score against validated 'correct' descriptors that flag distress with events that actually occur on the job.
5. 5Score and flag: high scorers are 'misfits' and potential leavers because they would feel distressed by the actual discomforting events on the job; turn them away at the front end.
6. 6Validate against turnover and performance, and check adverse impact. The compatibility test moderately predicted turnover (correlations in the .30s) and produced no adverse impact (Bernardin & Russell, 1998) — hold your instrument to the same standard.
7. 7Combine multiple attribute tests rather than relying on one. Selection procedures, especially when multiple tests assessing varied attributes are combined, are far more effective at curbing attrition than realistic job previews.
8. 8Monitor fit after hire, not just at selection. Trevor (2001) implies vigilant monitoring of incumbents' job attitudes; when attitudes worsen, people resign regardless of labor market — so treat fit as a maintained condition, not a one-time gate.

Villanova, Bernardin, Johnson, and Dahmus (1994) built a forced-choice personality inventory to screen quit-prone applicants for motion picture theater personnel. Each item set presented four situations, and candidates selected the two most discomforting to them; two items were validated 'correct' descriptors — 'Frequently Having to Interact with Senior Citizens' and 'Being Forced to Keep My Composure Around People Under Trying Circumstances' — while distractors like 'Frequently Being Bored at the Work I Have to Do' carried no score. Applicants earned a positive score for each valid discomforting item they picked, flagging them as job misfits likely to feel distressed by events that genuinely occur on the job. Those high scorers were turned away. The instrument moderately predicted turnover, with correlations in the .30s, and produced no adverse impact per Bernardin and Russell (1998). It demonstrates how P-O/P-J fit can be operationalized as a scored compatibility measure rather than an abstract judgment.

• If a candidate scores high on a validated discomfort/misfit inventory, then treat them as a potential leaver and screen them out at the front end.
• If you must choose between a realistic job preview and a combined multi-attribute selection battery for reducing attrition, then use the combined battery — it is far more effective at curbing attrition than RJPs.
• If your goal is reducing turnover overall, then weigh job enrichment first: McEvoy and Cascio's (1985) meta-analysis showed it is twice as effective as job previews.
• If you are measuring fit as part of embeddedness, then model it formatively (indicators cause the construct) rather than as a latent cause of scale scores.
• If an incumbent's job attitudes worsen over time, then act on retention regardless of labor market conditions, because they will likely resign anyway (Trevor, 2001).

• Assessing only job-side fit and ignoring the external/community environment. Tell: your turnover model explains leaving from work stress but misses people who quit over relocation, neighborhood, or community mismatch (the 'fit' dimension spans work AND external environments).
• Treating fit as a latent construct that causes its indicators when it actually functions formatively. Tell: you assume being embedded causes marriage or homeownership, when Mitchell and Lee (2001) show the causal arrow runs the other way — indicators cause embeddedness.
• Relying on a single test or a realistic job preview to secure fit. Tell: attrition stays high despite RJPs; you skipped combining multiple attribute tests, which is far more effective.
• Deploying a fit/misfit screen without validating it against turnover, performance, and adverse impact. Tell: you cannot cite a criterion correlation or an adverse-impact check the way Villanova et al. (correlations in the .30s) and Bernardin & Russell (no adverse impact) could.
• Treating fit as a one-time selection gate and never monitoring incumbents. Tell: good hires leave anyway because deteriorating attitudes went unwatched (Trevor, 2001).

• ▸Individual characteristics to assess: list the candidate values, personality traits, and goals in scope.
• ▸Organizational characteristics to match: list the culture, values, and structure attributes the person must fit.
• ▸Forced-choice item sets: write situation clusters; mark which descriptors are validated 'discomfort' items.
• ▸Misfit scoring rule: state how selecting a valid discomforting item earns a positive (misfit) score.
• ▸Flag threshold: define the score at or above which a candidate is treated as a potential leaver.
• ▸Criterion validation evidence: record correlation with turnover/performance (target the .30s or better).
• ▸Adverse impact check: note group comparison results confirming no adverse impact.
• ▸Multi-test combination: list the other attribute tests combined with this screen to curb attrition.
• ▸Post-hire monitoring trigger: specify the attitude decline that prompts a retention response for incumbents.

1. 1Decompose competence into the four descriptor families the source names: knowledge, skills, abilities, and other characteristics (KSAOs). Separate them explicitly, because they behave differently — knowledge and skills improve with training and experience, abilities are relatively stable, and 'other characteristics' captures personality, experience, and licensure/certification.
2. 2Start with a work analysis to identify which specific tasks and work responsibilities exist in the role, since tasks are the collections of work elements (action + object + purpose) that competence must serve. Do not rate people before you have documented the job.
3. 3For each task, apply the 'Required on entry' scale to sort competence by acquisition timing: 1 = not expected immediately but after formal training, 2 = not expected immediately but quickly learned on the job, 3 = should possess immediately. This tells you what to select for versus what to train for.
4. 4Rate each task on importance (1=not important to 5=extremely important), frequency, criticality/consequence-of-error, and difficulty using the standard anchored scales, so competence requirements are weighted rather than treated as a flat list.
5. 5Distinguish skill types: content skills (foundational, e.g., writing, critical thinking) versus cross-functional skills (e.g., problem solving, negotiation) that carry across job contexts — and decide which are trainable in your context.
6. 6Guard the ratings against social and cognitive distortion. Morgeson and Campion (1997) identify 16 sources of inaccuracy split into social (normative/self-presentation pressure) and cognitive (information-processing limits); collect from multiple sources and check for inflation.
7. 7Interpret any performance-linked differences in competence ratings cautiously: better employees may genuinely be assigned different or extra tasks, and low performers may exit, creating range restriction. Do not assume rating differences reflect pure competence differences.

Suppose you are analyzing the stockbroker role. Christal-type research (Ford et al., 1992, as cited) found significant relationships between time-spent ratings and stockbroker performance, so you treat time-spent as an informative signal here. You document tasks, then apply the 'Required on entry' scale: securities licensure scores 3 (should possess immediately, since it is a licensure requirement that falls under 'other characteristics'), while proprietary trading-platform navigation scores 2 (quickly learned on the job). You rate client-facing negotiation as a cross-functional skill of high importance (5) and high consequence-of-error (6), flagging it for structured development. Contrast this with Wexley and Silverman's (1978) retail-manager sample, where importance and time-spent ratings showed no performance differences — a reminder that a competence-to-performance link demonstrated in one job does not transfer automatically. So you validate the KSAO weighting inside the target job rather than importing it.

• If a task's 'Required on entry' rating is 3, then select for that competence at hiring; if it is 1 or 2, then plan to develop it through training or on-the-job learning instead.
• If the attribute is knowledge or a skill, then assume it can improve with training and experience; if it is an ability (cognitive, psychomotor, physical, sensory-perceptual), then treat it as relatively stable and select rather than train.
• If competence ratings differ between high and low performers, then check whether higher performers were assigned different/extra tasks or whether low performers exited (range restriction) before attributing the gap to competence.
• If a task has high criticality/consequence-of-error, then weight its associated competence more heavily even when its frequency is low.
• If ratings come from a single self-reporting source, then treat them as vulnerable to self-presentation and social-influence distortion and corroborate with additional raters.

• Treating all KSAOs as trainable. The tell: a development plan that budgets training hours for stable abilities the source says are 'more stable over time,' producing no improvement.
• Using within-job relative scales (e.g., relative time-spent) and then comparing across jobs. The tell: Harvey's (1991) warning — cross-job comparisons become problematic because the anchors are only relative within one job.
• Reading rating differences between performers as pure competence signals. The tell: ignoring that better employees may hold different tasks and low performers may have left, as in the Conley and Sackett (1987) juvenile-officer null result being over-interpreted elsewhere.
• Assuming a competence-performance link found in one occupation transfers. The tell: importing stockbroker time-spent weighting into a retail-manager role, where Wexley and Silverman (1978) found no such difference.
• Ignoring social and cognitive rating inaccuracy. The tell: inflated, uniformly high self-ratings with no multi-source check, matching the normative/self-presentation pressures Morgeson and Campion catalog.

• ▸Task statement — write the action + object + purpose for one specific task
• ▸Descriptor type — mark Knowledge / Skill (content or cross-functional) / Ability / Other characteristic
• ▸Importance rating — score 1 (not important) to 5 (extremely important)
• ▸Consequence-of-error rating — score 1 to 7 using the criticality scale
• ▸Required-on-entry rating — score 1 (after formal training), 2 (learned on job), or 3 (immediate)
• ▸Select vs. develop decision — write 'Select' if entry=3 or ability; 'Develop' if entry=1/2 and trainable
• ▸Data source(s) — list who provided the rating and how many raters (flag single-source risk)
• ▸Range-restriction note — note whether high/low performers hold different tasks or exit patterns exist
• ▸Validation status — record whether the competence-performance link is confirmed in THIS job or borrowed

1. 1Recognize that motivation is one of the 'other characteristics' descriptors in work analysis (personality and motivational traits like conscientiousness, leadership, initiative) — treat it as a distinct attribute needed for successful role performance, not a byproduct.
2. 2Map the work context that shapes motivation: assess task context (autonomy, task clarity, consequence of error, accountability, resources), social context (interdependence, communication, interpersonal conflict), and physical context, because these situational opportunities and constraints affect the occurrence and meaning of behavior.
3. 3When you rely on incumbents for information about motivation-related tasks, verify their input independently — some incumbents lack the motivation to provide accurate, reliable data, and others inflate the value of their job when an outcome (like compensation redesign) could benefit them.
4. 4Design work as part of high-performance work practices (HPWP): systematic selection, strategic training, performance management, and compensation systems purported to increase knowledge, skills, and motivation for organizational benefit.
5. 5Build the analysis on a formal work analysis base — HPWP research measures 'the proportion of the workforce whose job has been subjected to a formal job analysis,' so ground motivational interventions in documented duties, not assumptions.
6. 6Cross-check attributes across multiple sources (documentation, incumbents, technical experts, supervisors, analysts) because motivational input from any single source is biased by self-presentation processes.

Smith and Hakel (1979) found that incumbents and supervisors displayed considerable response inflation on socially desirable work analysis items compared with analyst ratings. Apply this directly to motivation: when you survey incumbents about how much initiative or leadership their role requires, expect inflated self-reports driven by social desirability — a need for social approval attained through culturally acceptable answers — and impression management, where incumbents create and maintain desired perceptions of themselves. If the survey feeds a compensation redesign, inflation rises further because the outcome benefits them. Counter this by adding analyst ratings as a benchmark, since analysts have little motivation to bias results and produce highly reliable ratings. The gap between incumbent and analyst ratings tells you how much of the reported 'motivational requirement' is genuine versus self-serving. This is the operational reason multiple sources beat a single self-report.

• If the work analysis outcome could benefit incumbents (e.g., compensation redesign), then discount their motivation/importance ratings and weight analyst or supervisor input more heavily.
• If incumbents cannot articulate what they do — from low verbal ability or low motivation to respond accurately — then use supervisors, who have higher verbal ability and less incentive to distort, to fill the gap.
• If you are gathering motivational data in an SME group meeting, then guard against conformity pressures, extremity shifts (group polarization), and motivation loss by also collecting independent individual input.
• If the task context lacks autonomy, task clarity, or resources, then treat weak motivation as a structural design problem, not an individual attribute deficiency.
• If better employees show different work analysis responses, then check whether they were assigned additional or different tasks (as reward or capacity) before concluding the role itself demands more motivation.

• Treating self-reported motivation as fact — the tell is incumbent ratings systematically higher than analyst ratings on socially desirable items (Smith & Hakel, 1979).
• Ignoring range restriction: low performers leave voluntarily or are terminated, so surviving-sample data understates the motivation-performance link — the tell is a suspiciously weak or null relationship between experience and ratings.
• Running SME panels without controls — the tell is rapid consensus on 'this aspect is essential' driven by majority conformity pressure rather than evidence, plus quieter members contributing little (motivation loss).
• Neglecting work context and attributing all motivation to the person — the tell is intervening only on individual traits while autonomy, accountability, and resource conditions go unexamined.
• Omitting work analysis from HPWP design — the tell is training or compensation programs built without a systematic needs analysis or formal job analysis underpinning them.

• ▸Motivational trait assessed: name the specific trait (e.g., conscientiousness, initiative, leadership).
• ▸Incumbent rating: record self-reported importance/frequency using a defined scale.
• ▸Analyst rating: record the analyst's independent rating of the same item.
• ▸Inflation gap: subtract analyst from incumbent rating; flag large positive gaps.
• ▸Outcome stake: note whether the analysis outcome (e.g., pay) could benefit the incumbent — yes/no.
• ▸Task context conditions: list autonomy, task clarity, accountability, and available resources for this role.
• ▸Group-setting risk: note if data came from an SME panel and which bias (conformity/polarization/motivation loss) applies.
• ▸Source triangulation: list which of documentation, incumbents, experts, supervisors, analysts confirmed the attribute.
• ▸Action: state whether to keep, discount, or re-collect the rating and why.

1. 1Locate the construct precisely: proactive work behavior is self-starting, anticipatory, and change-oriented action aimed at improving the current situation. Before labeling any action proactive, confirm all three markers are present — reactive compliance and passive adaptation do not qualify.
2. 2Separate the three targets of improvement the definition names — the work situation, personal effectiveness, and organizational functioning — and assign the behavior you are examining to one. This tells you where the payoff should register and who should notice.
3. 3Build the enabling orientation, not just the behavior. Per the career passages, proactive enhancement behaviors travel with a value-driven, self-directed (protean) orientation and with career insight; develop insight of self, skills, and aspirations first, because behavior without direction produces the 'wanderer' pattern.
4. 4Design conditions that reward anticipation. The 'trapped/lost' profile shows what happens under low proactiveness plus inability to see cross-boundary possibilities: control collapses to luck. Give people visibility of options across role boundaries so anticipatory action has somewhere to go.
5. 5Measure proactivity through forward-thinking indicators, not effort. Track whether the person initiates contact, drives themselves toward opportunities, and originates change — the same self-direction markers that distinguished 'wanderer' (adapts but doesn't drive) from a genuinely proactive actor.
6. 6Tie the behavior to an outcome the passages actually support: proactive career enhancement behaviors correlate strongly with protean orientation (β=.76) and with career insight (β=.87). Use insight-building interventions (e.g., career counseling as in De Vos & Soens, 2008) as the lever.

De Vos and Soens (2008) studied individuals who had received career counseling. They found protean orientation was strongly related to career insight — insight into one's self, skills, and aspirations (β=.87) — and to proactive career enhancement behaviors (β=.76). This is the mechanism in action: the counseling did not directly command proactive acts; it developed the self-directed, value-driven orientation and the insight, which is where the anticipatory, change-oriented behavior originates. Separately, Baruch and Quick (2007), studying retired navy admirals pursuing second careers, found protean orientation associated with shorter time to find the new job (r=.13) and higher overall career satisfaction (β=.63) — proactivity shortening the transition and improving the outcome. The practical read: to raise proactive work behavior, invest in insight and orientation, then expect faster, self-driven pursuit of improvement.

• If an action lacks any of self-starting, anticipatory, or change-oriented, then classify it as adaptation or compliance, not proactive work behavior.
• If someone adapts easily but their career is controlled by circumstances, then they are a 'wanderer' — build self-direction and the ability to identify and evaluate opportunities, not more adaptability.
• If control over outcomes depends on luck and options are narrow, then the constraint is inability to see possibilities across boundaries — widen visible options before demanding more initiative.
• If you want to raise proactive enhancement behavior, then develop career insight and protean orientation first, since those show the strongest associations (β=.87, β=.76).
• If the person is driven only by personal values without self-direction ('fortressed'), then success will only appear under stable, values-matched conditions — add opportunity-recognition capability.

• Mistaking busy compliance for proactivity: the tell is that all activity is prompted by others and none anticipates a future need or originates change.
• Building adaptability instead of self-direction: the tell is a 'wanderer' — the person adjusts well but never drives themselves toward opportunities; outcomes stay circumstance-controlled.
• Demanding initiative while options are invisible across boundaries: the tell is the 'trapped/lost' pattern — effort exists but control over results feels like luck.
• Rewarding values-alignment without opportunity-recognition: the tell is 'fortressed' behavior — proactive only inside stable, values-matched settings and paralyzed elsewhere.
• Chasing behaviors directly while ignoring the orientation and insight that generate them: the tell is short-lived initiative that fades once external prompting stops.

• ▸Behavior observed — describe the specific action in one line.
• ▸Self-starting? — yes/no: was it initiated without external prompt.
• ▸Anticipatory? — yes/no: did it address a future need before it became a problem.
• ▸Change-oriented? — yes/no: did it aim to improve rather than maintain.
• ▸Improvement target — work situation / personal effectiveness / organizational functioning.
• ▸Orientation profile — self-directed, value-driven, wanderer, trapped/lost, or fortressed.
• ▸Career insight level — rate insight of self, skills, and aspirations (low/med/high).
• ▸Option visibility — can the person see opportunities across role boundaries (yes/no).
• ▸Development lever — insight-building, self-direction, or opportunity-recognition.

1. 1Define the performance domain before you measure it: run task analysis and critical incidents analysis to generate candidate performance dimensions, then reduce them empirically (Project A used content analyses and principal-components analyses on 9,430 incumbents to fit a five-dimension model).
2. 2Separate task performance from adjacent constructs. Isolate 'core technical proficiency' and 'general soldiering proficiency' as distinct dimensions from effort/leadership, personal discipline, and fitness — do not fold citizenship or effort into the task score.
3. 3Build measures at the task level using the descriptor structure: tasks are collections of specific work elements (action + object of action + purpose/result). For an industrial machinery mechanic, write items like 'disassembles machinery to remove parts and make repairs' rather than the broad responsibility 'repairs and maintains equipment.'
4. 4Use multiple methods per dimension to control method variance. Project A indexed each dimension with rating scales, job knowledge tests, hands-on job samples, and archival records, then modeled rating and paper-and-pencil method factors explicitly.
5. 5Derive worker requirements (KSAOs) from the task list so predictors map to the criterion: knowledge (declarative vs. procedural), skills (basic vs. cross-functional, improvable via training/experience), abilities (relatively stable capacities), and other characteristics.
6. 6Validate the predictor-to-criterion link with criterion-related evidence, then correct for range restriction and criterion unreliability using restricted/unrestricted SDs and reliability estimates, as Project A did to move observed validities up to corrected ranges.
7. 7Situate the role in the broader system: because a role is implicitly multilevel, check how the individual's task performance connects to team interdependencies and organizational business goals so the criterion is not defined in isolation.

Project A defined and measured the task-performance domain for 19 military jobs. Researchers ran extensive task and critical incidents analyses to generate performance dimensions, then built measures for each: rating scales, job knowledge tests, hands-on job samples, and archival records. Performance data from 9,430 incumbents were analyzed with content analyses and principal-components analyses, producing five dimensions — core technical proficiency, general soldiering proficiency, effort and leadership, personal discipline, and fitness/bearing — plus rating and paper-and-pencil method factors. A predictor battery (cognitive ability, perceptual-psychomotor ability, temperament/personality, interest) was aggregated into 24 composites and correlated with the five dimensions. Observed validities ran rs = .07–.47 (Mdn = .15); after correcting for range restriction and criterion unreliability, corrected validities for combined composites ran .37 to .70. The effort shows task performance ('core technical proficiency') can be isolated, measured with multiple methods, and predicted with defensible validity.

• If the criterion mixes technical output with helping or effort, then split it — keep core task proficiency as its own dimension (Project A separated 'core technical proficiency' from 'effort and leadership').
• If you have both restricted and unrestricted predictor SDs plus criterion reliability estimates, then report corrected validities alongside observed ones rather than stopping at the observed median.
• If a competency looks improvable with training and experience on a task, then classify it as a skill; if it is a relatively enduring, stable capacity across many activities, then classify it as an ability.
• If the descriptor is a specific action tied to one work role, then treat it as a task; if it is a generic behavior aggregating related tasks across many roles, then treat it as a responsibility.
• If the role is embedded in interdependent team action, then define the criterion multilevel so individual task scores link to collective and organizational goals.

• Treating the job as a static entity — the tell is a task list that never gets revised and ignores discretionary or emergent task elements.
• Reporting only observed validity — the tell is a low median (e.g., .15) presented as the ceiling when range restriction and unreliability corrections would show .37–.70.
• Measuring each dimension with a single method — the tell is that a rating-based score and a paper-and-pencil score correlate because of shared method, not construct.
• Confusing skill with ability — the tell is expecting a trainable proficiency to be as stable over time as a cognitive or physical capacity, or vice versa.
• Defining the criterion only at the individual-incumbent level — the tell is no link to team interdependencies or business goals, which pushes people toward loose competency modeling instead.

• ▸Work role: name the specific role whose task proficiency you are measuring
• ▸Tasks: list specific work elements as action + object + purpose (role-specific)
• ▸Responsibilities: group related tasks into broad behavior statements
• ▸Performance dimensions: the reduced set from task/critical-incident analysis (e.g., core technical proficiency)
• ▸KSAO requirements: knowledge, skill, ability, other characteristics mapped to each dimension
• ▸Measurement methods: assign 2+ per dimension (rating scale, knowledge test, work sample, archival)
• ▸Method factors: note which factors you will model to separate method from construct
• ▸Validation data: predictor composites, sample size, observed and corrected validities
• ▸Multilevel links: how this role's tasks connect to team interdependencies and organizational goals

1. 1Start from the formal definition: OCB is discretionary behavior not directly or explicitly recognized by the formal reward system that promotes effective organizational functioning. Use this to screen candidate behaviors — if the behavior is contractually mandated or triggers a specific reward, it fails the discretionary test.
2. 2Do not assume OCB stays extra-role. The passages show individuals frequently view OCB as falling within their work-role requirements (Haworth & Levy, 2001; Hui, Lam, & Law, 2000; Morrison, 1994). Measure each person's OCB role definition before treating a behavior as extra-role for them.
3. 3Operationalize OCB role definitions the way extant research does — by administering an instrument that asks incumbents whether specific citizenship acts are part of their job (the approach referenced via Kamdar, McAllister, & Turban, 2006; McAllister et al., 2007).
4. 4Treat OCB as generic work behavior applicable across many work roles (Borman & Penner, 2001), then test that assumption rather than accepting it — the passages state the premise of wide applicability 'has yet to be empirically substantiated.'
5. 5Fold OCB into work analysis by capturing discretionary or emergent task elements, not only prescribed ones (Ilgen & Hollenbeck, 1991; Morgeson & Humphrey, 2008). Use a role frame rather than a static job frame so citizenship behaviors interrelated with other role holders become visible.
6. 6Model instrumentality: because instrumentality beliefs predict OCB (Haworth & Levy, 2001) and OCB can carry instrumental value for promotion (Hui, Lam, & Law, 2000), examine whether employees expect citizenship to pay off, since that expectation drives whether they perform it.
7. 7Account for context as a moderator. Climate moderates the link between leader relationships and content-specific citizenship (Hofmann, Morgeson, & Gerras, 2003, using safety climate), so specify which citizenship content you mean and which climate conditions it operates under.
8. 8Link OCB to the criterion domain deliberately: when expanding performance criteria to include OCB for selection (Organ, Podsakoff, & Podsakoff, Ch. 10), define which citizenship dimensions matter for the role before building or validating any selection measure around them.

Consider the promotion field quasi-experiment referenced in the passages (Hui, Lam, & Law, 2000). Employees who believed organizational citizenship behavior carried instrumental value for promotion engaged in more OCB, and once the instrumental payoff was in play the behavior shifted away from the purely discretionary, unrecognized category in the definition. This is the core tension: the classic definition says OCB is not explicitly recognized by the formal reward system, yet the moment employees perceive it as a route to promotion, they treat it as role-required and instrumentally motivated. For a practitioner, this means an OCB you observe on the floor may be discretionary for one worker and self-defined as in-role for another, depending on their instrumentality beliefs and role definition. The operational move is to measure both the role definition and the instrumentality belief before labeling the behavior. Failing to do so will misclassify motivated, reward-seeking behavior as spontaneous citizenship.

• If a behavior is explicitly recognized by the formal reward system, then it is not OCB by definition — reclassify it as in-role performance.
• If an incumbent's role definition includes the citizenship act (per instrument responses), then treat it as in-role for that person even if it is extra-role for others (Morrison, 1994; McAllister et al., 2007).
• If you are asserting a citizenship behavior applies across many roles, then require empirical support from work-role requirement data, because the wide-applicability premise is not yet substantiated.
• If instrumentality beliefs about the behavior are high, then expect OCB frequency to rise and interpret motives as partly instrumental, not purely discretionary (Haworth & Levy, 2001; Hui et al., 2000).
• If you are studying content-specific citizenship, then specify the relevant climate as a moderator rather than testing OCB as context-free (Hofmann et al., 2003).

• Treating OCB as fixed extra-role behavior for everyone. Tell: your measure never asks incumbents whether the act is part of their job, contradicting Haworth & Levy, Hui et al., and Morrison findings.
• Assuming OCB is universally applicable across roles. Tell: you cite generic applicability (Borman & Penner, 2001) but present no work-analysis data, ignoring that the premise is empirically unsubstantiated.
• Counting rewarded behaviors as OCB. Tell: the behaviors you label citizenship are tied to explicit incentives, violating the 'not directly or explicitly recognized by the formal reward system' clause.
• Ignoring instrumentality motives. Tell: you interpret all OCB as selfless and cannot explain why citizenship spikes near promotion cycles (Hui et al., 2000).
• Analyzing OCB with a static job frame. Tell: your work analysis captures only prescribed tasks and misses discretionary or emergent elements (Ilgen & Hollenbeck, 1991), so citizenship behaviors are invisible in the data.

• ▸Behavior described: state the specific act in observable terms
• ▸Discretionary test: is it explicitly recognized/rewarded by the formal system? (yes = not OCB)
• ▸Incumbent role definition: does this person report the act as part of their job? (source: administered instrument)
• ▸Cross-role applicability: what work-analysis evidence shows it applies to this role?
• ▸Instrumentality belief: does the employee expect the act to aid promotion/rewards? (high/med/low)
• ▸Citizenship content type: name the specific dimension of citizenship at issue
• ▸Relevant climate moderator: which context/climate conditions affect this behavior (e.g., safety climate)?
• ▸Criterion-domain use: how will this OCB feed selection or performance evaluation (Ch. 10)?
• ▸Classification decision: in-role for this person / extra-role / rewarded-not-OCB — with justification

1. 1Define retention operationally as the inverse of voluntary organizational exit (Chapter 11, Hom), and separate voluntary quitting from involuntary exit modes such as layoff, downsizing, and relocation (Coucke et al., 2007) so you don't count what you can't influence.
2. 2Instrument your recruitment stage with realistic job previews (RJPs): the passages report RJPs produce consistent, though small, increases in retention without causing greater applicant self-selection out of the pool — deploy them and measure the delta rather than assuming a large effect.
3. 3Reduce delays between recruitment phases; in campus contexts delays cause significant dropout, heaviest among applicants with the most opportunities, while in other labor markets dropout hits those needing immediate employment — so time your process to your candidate population.
4. 4Add noncognitive predictors to selection: personality factors predict retention beyond general cognitive ability and with far less adverse impact, so score for retention-relevant traits at hire rather than treating retention as a post-hire-only problem.
5. 5For the retained population, invest in development levers the passages tie to retention — training (Brown & Sitzmann), mentoring (Eby, Kram), and learning orientation linked to organizational commitment and talent retention (D'Amato & Herzfeldt, 2008).
6. 6Track special-population retention separately, e.g., repatriate retention, which is influenced by expatriate and repatriate experiences and by organizational versus career support (Kraimer et al., 2009; Lazarova & Cerdin, 2007).
7. 7Close the loop: attribute exits to their mode (quit vs. downsizing vs. relocation) and feed findings back into RJP content, timing of phases, and selection scoring.

A firm hiring on campus finds strong offers accepted but early quitting high. Following the passages, it does two things. First, it inserts an RJP into the process, expecting only a small but consistent retention gain and confirming that dropout from the applicant pool does not spike — RJPs do not appear to cause greater applicant self-selection out. Second, it audits phase timing and finds a multi-week delay between interview and offer; because delays in campus recruiting cause significant dropout, and dropout is most severe among applicants with the most opportunities, it compresses the timeline. It also adds a personality measure at selection, justified by the finding that personality predicts retention beyond cognitive ability with far less adverse impact. Post-hire, it pairs new entrants with mentors, drawing on the mentoring literature (Kram, 1985) associated with commitment. The result is measured as reduced voluntary exit, cleanly separated from any involuntary reductions.

• If turnover includes layoffs, downsizing, or relocation, then classify those as involuntary exit and exclude them from voluntary-retention metrics (Coucke et al., 2007).
• If your recruitment lacks an RJP, then add one — but forecast only a small retention increase and do not expect it to shrink your applicant pool.
• If you recruit on campus, then minimize inter-phase delays, prioritizing speed for high-option candidates who drop out first under delay.
• If you can add a valid noncognitive predictor, then use it to raise retention prediction while limiting adverse impact relative to cognitive-only selection.
• If you employ expatriates, then manage repatriate experience and organizational support deliberately, since these drive repatriate retention.

• Counting all separations as turnover — the tell is retention numbers that swing with restructuring events rather than employee choice; downsizing and relocation are distinct exit modes.
• Expecting RJPs to be a large fix — the tell is disappointment when the retention gain is modest; the evidence says consistent but small.
• Ignoring process delays — the tell is losing your strongest candidates first in campus pipelines, exactly the group with the most opportunities.
• Selecting on cognitive ability alone — the tell is higher adverse impact and weaker retention prediction than adding personality factors would give.
• Treating repatriates like domestic hires — the tell is post-assignment departures traceable to weak organizational or career support and unmanaged repatriate experience.

• ▸Exit classification: tag each separation as voluntary quit or involuntary (layoff/downsizing/relocation).
• ▸Voluntary retention rate: employees remaining / population over the period, involuntary exits excluded.
• ▸RJP status: present/absent, and measured retention delta since introduction.
• ▸Applicant pool effect: dropout rate before vs. after RJP to confirm no added self-selection.
• ▸Phase-delay days: elapsed time between each recruitment stage.
• ▸High-option dropout: share of most-courted candidates lost to delay.
• ▸Selection predictors: list cognitive vs. noncognitive measures and their retention/adverse-impact notes.
• ▸Development levers active: training, mentoring, learning-orientation programs in place.
• ▸Special-population retention: separate rate for repatriates/expatriates with support notes.

1. 1Fix the ultimate purpose before choosing anything else: selection, training, compensation, or performance management. The purpose dictates which descriptors you collect (tasks and/or skills for training; skills and/or traits for selection instruments).
2. 2Choose descriptor type(s) matched to that purpose, and decide explicitly whether to capture organizational context (noise, lighting, temperature, hazards, physiological demands like sitting/standing/climbing), since context has a pronounced effect on work role requirements (Dierdorff & Morgeson, 2007).
3. 3Consult O*NET first to locate the occupation(s) matching your focal role and pull the descriptors most relevant to your purpose. Treat O*NET as the foundation, not the endpoint — its descriptors are clearly defined and theoretically based, and nearly three quarters of occupations were updated since 2005.
4. 4Select multiple data-collection methods (observation, individual interviews, group meetings, questionnaires) so you can triangulate and capture different perspectives, weighing time, cost, and intended purpose. Direct observation avoids selective-recall and reporting bias from workers, though it is time consuming.
5. 5Verify or augment the O*NET base in-house: run SME or incumbent ratings/rankings to confirm existing descriptors, then add company-specific content (particular software systems, department-specific responsibilities) so results support content-related validity evidence.
6. 6Attach rating scales to tasks or attributes — frequency, importance, criticality/consequences of error, difficulty — using the scale format that fits your comparison needs. Use direct importance estimates rather than always building composites.
7. 7Improve data quality by training raters (frame-of-reference training reduces idiosyncratic variance in analyst and incumbent ratings), then evaluate the effort against reliability, accuracy, validity, utility, acceptance, and cost effectiveness — not just documentation of the process.

A practitioner designing a training program starts by fixing the purpose: improved performance, so tasks and skills are the target descriptors. She opens O*NET OnLine, locates the matching occupation, and extracts the defined task and skill statements as a foundation. Rather than stopping there, she runs an incumbent verification pass, asking subject matter experts to rate each task on relative time spent and to directly estimate importance to the overall role. Because she needs cross-job comparability later, she avoids the relative difficulty scale (where '1/2 tasks are more difficult, 1/2 less' makes cross-job comparisons problematic) and leans on absolute frequency plus a direct importance rating, which research shows is as reliable as a difficulty-plus-criticality composite. She augments the O*NET base with company-specific knowledge about the particular software systems the role uses. The resulting task-and-skill profile becomes the grounded input to the training design and the content-validity record.

• If the purpose is selecting instruments, then collect skills and/or traits plus task-to-KSAO linkage judgments; if the purpose is designing training, then collect tasks and/or skills.
• If you need cross-job comparisons, then avoid relative-difficulty scales and use absolute frequency plus direct importance, because relative scales make cross-job comparison problematic.
• If reporting bias or selective recall by incumbents is a concern, then use direct observation, since an outside observer is not subject to those worker-side biases.
• If starting a new work analysis, then use O*NET as the starting foundation and layer in-house verification or customization on top, rather than treating any single method as the one right way.
• If you want to reduce idiosyncratic variance in ratings, then apply frame-of-reference (or error/dimension) rater training before collecting attribute descriptors.

• Treating O*NET output as the finished product — the tell is a profile with no incumbent/SME verification and no company-specific content (e.g., no mention of the actual software systems used).
• Choosing methods or descriptors before fixing the purpose — the tell is data that doesn't cleanly support the intended HR system (skills collected when tasks were needed for training).
• Using relative-difficulty or relative-frequency scales and then attempting cross-job comparisons — the tell is comparisons that collapse because each rater anchored to a different job baseline.
• Relying on a single collection method with no triangulation — the tell is no cross-perspective check and inability to reconcile conflicting incumbent, supervisor, and observer accounts.
• Judging the effort only by whether the process was documented — the tell is a case-study writeup with no evidence on reliability, accuracy, validity, utility, or acceptance of the resulting system.

• ▸Intended purpose: name the HR system this feeds (selection / training / compensation / performance management).
• ▸Descriptor type(s): list tasks, skills, traits chosen to match the purpose.
• ▸Context descriptors: note relevant conditions (noise, lighting, temperature, hazards, physiological demands) to capture.
• ▸O*NET occupation match: record the occupation code/title used as the foundation and its update year.
• ▸Methods for triangulation: list the 2+ methods chosen (observation, interview, group meeting, questionnaire).
• ▸Rating scales: specify frequency, importance, criticality, difficulty scale(s) and whether absolute or relative.
• ▸Verification/augmentation plan: state how SMEs/incumbents will verify O*NET and what company-specific content is added.
• ▸Rater training: note the training applied (frame-of-reference, error, dimension) to reduce idiosyncratic variance.
• ▸Quality criteria: list evaluation targets (reliability, accuracy, validity, utility, acceptance, cost effectiveness).

1. 1Scope the unit of analysis as a role, not a job: define the expected pattern of behaviors interrelated with others, so you capture tasks, responsibilities, and how the role connects to the broader system of roles in the organization (Katz & Kahn, 1978).
2. 2Separate work requirements from worker requirements: list tasks and general responsibilities (activities) on one side, and the knowledge, skills, abilities, and other characteristics (attributes) needed to perform them on the other, following the activity/attribute distinction (Sackett & Laczo, 2003; Dierdorff & Morgeson, 2007).
3. 3Make the four procedural choices explicitly — what data, how (method), when, and from whom — and document the pros and cons of each, since these are the technical decisions that determine data quality.
4. 4When using questionnaires, choose the rating scale deliberately: decide whether frequency is captured with highly specific time-based anchors ('about once per year' to 'about once each hour or more often') or a less specific estimate, and match scale type to the requirement being measured.
5. 5Limit questionnaire length and respondent load; overlong instruments (several hundred questions, multiple scales) produce unreliable, inaccurate responses.
6. 6Plan for systematic, not just random, error: instead of relying only on sampling and standardized materials, identify rater characteristics (job familiarity, tenure) that predictably shift ratings and design to detect them.
7. 7Evaluate the resulting data against a broad criterion set — including mean elevation/depression of ratings and completeness (comprehensiveness) — rather than reliability alone.
8. 8Where possible, adopt O*NET's content model as the state-of-the-art descriptor framework, and record whether ratings come from role incumbents or analysts, because source is itself a variance term.

Van Iddekinge et al. (2005) tested how much of the error variance in single-rater work analysis ratings was attributable to the rater rather than the work. For knowledge, skill, ability, and other characteristics importance ratings, 21.6% of the error variance in single-rater reliabilities was due to rater idiosyncrasies; for needed-at-entry ratings, the figure rose to 29.1%. This means that a substantial share of what looks like measurement 'noise' is actually systematic and predictable variance tied to who is doing the rating. The practical consequence is direct: standardizing materials and sampling harder — the traditional fixes for random error — do not remove this variance, so ratings stay inaccurate. Harvey (1991) had already warned that 'one cannot simply assume that job analysis ratings will be unaffected by characteristics of the rater.' The corrective is to model rater characteristics explicitly and aggregate across raters whose idiosyncrasies differ.

• If the work is interdependent or team-embedded, then analyze it as a role connected to the system of roles rather than as an isolated job.
• If you need to describe both what is done and who can do it, then keep work requirements (tasks/responsibilities) and worker requirements (KSAOs) in separate columns to preserve the activity/attribute distinction.
• If a questionnaire runs to hundreds of items with numerous scales, then split, shorten, or rotate it, because respondent overload degrades reliability and accuracy.
• If ratings on a requirement come from a single rater, then treat up to ~20–30% of the variance as rater-idiosyncratic and add raters instead of only standardizing materials.
• If you are using O*NET data, then record whether descriptors are incumbent-rated or analyst-rated, since the two sources carry different variance and little of the current database has been re-validated.

• Assuming all measurement error is random — the tell is a research design that fixes error only through sampling and standardized materials while ignoring who rates; systematic rater variance survives untouched.
• Defining the unit as a static 'job' rather than a role — the tell is documentation that lists tasks in isolation with no account of how the role connects to interdependent others or the broader organizational system.
• Collapsing activity and attribute — the tell is a single list mixing tasks with required abilities, so you cannot trace which KSAO supports which task.
• Judging data quality by reliability alone — the tell is no check for inappropriately elevated/depressed mean ratings or for completeness/comprehensiveness of coverage.
• Overloading respondents with an oversized questionnaire — the tell is flat, patterned, or inconsistent responses across long rating blocks.

• ▸Unit of analysis: name the role and its expected pattern of interrelated behaviors
• ▸Role connections: list the other roles/system this role links to
• ▸Work requirements: list tasks and general responsibilities (activities)
• ▸Worker requirements: list needed knowledge, skills, abilities, other characteristics (attributes)
• ▸Method + rationale: chosen data-collection method with its pros/cons
• ▸Rating scales: scale type per requirement (e.g., frequency — specific time-based vs. less specific)
• ▸Source(s): who rates — incumbents vs. analysts — and expected rater influences (familiarity, tenure)
• ▸Variance plan: number of raters and controls for systematic (not just random) error
• ▸Quality criteria: reliability, mean-rating elevation/depression check, and completeness check

1. 1Define recruitment as a staged process, not a single event: generate applicant pools, maintain viable applicants, then encourage desired candidates to join (Rynes, 1991; Rynes & Cable, 2003). Map your actions to each stage before choosing tactics.
2. 2Integrate context first, per Barber (1998) and Rynes & Cable (2003). Assess the external labor market — the 'war for talent' scope, sector poaching pressure (e.g., start-ups courting Google/Microsoft employees), and available channels — before designing strategy.
3. 3Segment your target by activity level. Decide whether you are pursuing active applicants or passive candidates, because passive-candidate recruiting requires a relationship-building sequence, not a job posting.
4. 4For passive candidates, use the Cappelli (2001) / Lievens & Harris (2003) sequence: identify targets on networking sites (linkedin.com, doostang.com), develop relationships first, then lure them away from competitors — do not lead with an offer.
5. 5Ground recruitment in work analysis data. Use the formal job analysis to specify the tasks, responsibilities, and worker requirements (knowledge, skill, ability, other characteristics) that define your 'desired candidate' so pool-generation and messaging target real role requirements.
6. 6Tie recruitment to strategy as a high-performance work practice (HPWP). Treat recruitment strategy, HR planning, and selection as an integrated system linked to organizational outcomes (turnover, sales, profits, market value per Huselid, 1995).
7. 7Prioritize HR planning as the forecasting backbone: Combs, Liu, Hall & Ketchen (2006) found HR planning has the largest effects on organizational performance among HPWP, so forecast human capital needs before opening requisitions.
8. 8Study your stages integratively rather than in isolation (Breaugh et al.), tracking whether pool quality is maintained through each transition instead of only measuring volume at the top.

A technology start-up needs senior engineers but the active-applicant pool is thin because the best people already work at Google and Microsoft. Rather than posting openings, the recruiter treats this as passive-candidate recruitment. Following Cappelli (2001) and Lievens & Harris (2003), the recruiter identifies specific engineers on linkedin.com and doostang.com and begins developing relationships with them well before any pitch. Web technology here levels the information playing field and increases the pace of contact. Only after the relationship exists does the recruiter encourage the desired candidate to join — the 'encourage desired candidates to join' stage of the process. The role requirements used to target and message these engineers are drawn from a formal work analysis specifying the tasks and worker requirements of the engineering role.

• If the desired talent is already employed by competitors (passive), then build relationships before making an offer; if the talent is actively searching, then optimize pool-generation channels and application funnel.
• If external market signals a 'war for talent' with recruiter difficulty finding good workers, then front-load context analysis and HR planning rather than expanding posting volume.
• If you cannot state the tasks, responsibilities, and worker requirements from a formal job analysis, then pause recruitment and complete the work analysis first — undefined 'desired candidate' produces an unfocused pool.
• If evaluating which HR system to strengthen for organizational performance, then invest in HR planning first, since it shows the largest performance effects among HPWP.
• If you are only measuring how many practices you use, then shift to measuring how the pool is built and maintained across stages — capturing usage alone misses the mechanism.

• Treating recruitment as a one-shot posting rather than a three-stage process — tell: you measure only applicant volume and have no metric for maintaining viable applicants or converting desired candidates.
• Ignoring context — tell: identical recruitment tactics are deployed regardless of labor-market tightness or sector poaching dynamics, so passive-candidate segments are never reached.
• Leading with an offer to passive candidates instead of building a relationship first — tell: high rejection or non-response from employed targets who were approached cold.
• Recruiting against a vague 'ideal candidate' with no work-analysis basis — tell: no documented tasks or worker requirements, so pools are large but poorly matched to the role.
• Building recruitment in isolation from HR planning and selection — tell: requisitions open reactively without forecasting future or changing role requirements, missing the largest HPWP performance lever.

• ▸Target role work-analysis summary: list tasks, responsibilities, and worker requirements (K/S/A/O) driving the profile.
• ▸Context assessment: describe labor-market tightness, sector poaching pressure, and available channels.
• ▸Candidate segment: mark active vs. passive and justify with market evidence.
• ▸Pool-generation actions: name channels/sites used to generate the applicant pool.
• ▸Maintain-viability actions: state how you keep viable applicants engaged across stages.
• ▸Encourage-to-join actions: specify relationship-building or persuasion steps for desired candidates.
• ▸HR-planning link: note forecasted future/changing role requirements this hire addresses.
• ▸Stage metrics: define one measure per stage (pool, maintenance, conversion) rather than volume alone.
• ▸Strategy alignment: state how this recruitment plan connects to broader HPWP and organizational outcomes.

1. 1Start with work analysis, because it is the foundation for selection system development (Morgeson & Dierdorff): systematically investigate work role requirements (tasks, responsibilities) and worker requirements (knowledge, skill, ability, other characteristics).
2. 2Separate activity-oriented analysis (what tasks are performed) from attribute-oriented analysis (what KSAOs are needed), following the Sackett & Laczo activity/attribute distinction, so you know both the criterion to predict and the predictors to assess.
3. 3For each candidate attribute, apply the point-of-entry rating scale: rate whether the attribute must be present at hiring or can be learned on the job. Route entry-required attributes into selection assessments; route learnable ones into training programs.
4. 4Rate each attribute for importance and required level using behavioral-anchor level scales (as in Fleishman's Ability Requirements Scales and O*NET), so the selection instrument targets the specific level the job demands, not just presence/absence.
5. 5Choose the selection procedure(s) matched to the attributes — the handbook's strategies include cognitive/individual-difference tests, personality assessment, interviews, assessment centers, and situational judgment tests (Part II, chapters 4-8).
6. 6Define the criterion domain you are predicting; consider expanding it beyond task performance to include organizational citizenship behavior (Organ, Podsakoff, & Podsakoff, chapter 10), because that changes what a 'valid' inference must cover.
7. 7Build validation support for the procedures (Schmitt & Sinha, chapter 13): assemble the evidence — including content-related validity anchored in the work analysis — that supports the accuracy of inferences about future performance.
8. 8Evaluate the whole system as a system (Part IV): assess applicant reactions (Truxillo & Bauer), validation support (Schmitt & Sinha), and utility via supply-chain/staffing analysis (Cascio & Boudreau).

Suppose the target attribute is 'reaction time,' defined in O*NET/Fleishman as the ability to quickly respond with hand, finger, or foot to a signal when it appears. In work analysis you first confirm through task and attribute questionnaires that fast response to signals is genuinely required by the role, then rate its importance and the required level. Using the level scale's behavioral anchors, you place the job's requirement: low ='start to slow down the car when a traffic light turns yellow,' moderate ='throw a switch when a red warning light goes off,' high = a still-faster response. You then apply the point-of-entry scale and judge reaction time as an at-hiring attribute rather than trainable, so it belongs in selection assessment, not the training curriculum. This gives you a defensible chain: the work analysis specifies the level, the selection instrument measures at that level, and validation evidence links the measure to the inference about future performance.

• If an attribute can be learned on the job, then route it to training; if it must be present at entry, then route it to a selection assessment.
• If you need content-related validity for a test, then anchor the test's content directly in the work analysis of work and worker requirements rather than in general job descriptions.
• If the role's success depends on behaviors like citizenship rather than only assigned tasks, then expand the criterion domain (OCB) before judging whether inferences are valid.
• If an attribute is rated important but the required level is low, then set the assessment threshold to that level rather than selecting for maximum ability.
• If you are evaluating the system, then judge it on applicant reactions, validation support, and utility together, not on any single procedure in isolation.

• Skipping or shortcutting work analysis; the tell is a selection instrument that cannot be traced back to specific work role requirements, so content validity claims are unsupported.
• Confusing entry attributes with trainable ones; the tell is testing applicants on skills the job actually teaches, needlessly screening out capable candidates.
• Measuring presence/absence instead of the required level; the tell is over- or under-selecting because behavioral-anchor level scales were never applied.
• Defining the criterion too narrowly as task performance; the tell is a 'valid' predictor that misses citizenship behaviors the role depends on (per Organ et al.).
• Treating validation as a one-time paperwork step rather than system evaluation; the tell is no evidence on applicant reactions or utility (Cascio & Boudreau's staffing/utility view).

• ▸Attribute name: state the KSAO from the work analysis
• ▸Definition: one-line behavioral definition (O*NET/Fleishman style)
• ▸Required level: mark low/moderate/high with its behavioral anchor
• ▸Importance rating: how important this attribute is to role performance
• ▸Point of entry: needed at hiring or learnable on the job
• ▸Assigned procedure: test / interview / assessment center / SJT / training
• ▸Criterion covered: task performance and/or OCB the attribute predicts
• ▸Validity evidence: type of support linking measure to the inference
• ▸System check: applicant-reaction, validation, and utility status

1. 1Anchor the performance management system in a work analysis: before you define what to measure, collect existing documentation (job descriptions, O*NET data, training manuals, operating guides) since this is cost-efficient and compiles known information about the role.
2. 2Choose descriptors that fit the purpose. For performance management design specifically, prioritize identifying the attributes and behaviors needed for effective role performance, and separate what must be present at entry from what can be developed on the job.
3. 3Use multiple data collection methods (observation, individual interviews, group meetings, questionnaires) so you can triangulate and capture different perspectives of the target work role; no single method dominates in research.
4. 4Sequence your sources: review written documentation first, use it to inform data collection from incumbents and technical experts, then have supervisors check and augment that data, then have analysts compile and draw conclusions.
5. 5Set level rating scales with behavioral anchors that illustrate low, moderate, and high proficiency, so feedback and goals reference observable behavior rather than abstract labels.
6. 6Feed the analysis into a continuous cycle: use critical incidents (supervisor-recorded effective/ineffective behaviors) as the raw material for ongoing developmental feedback, not just an annual score.
7. 7Gather input from all relevant stakeholders during design, because doing so both improves accuracy and enhances acceptance of the HR system built from the findings.

To build level-based feedback anchors, follow Fleishman's Ability Requirements Scales logic used in O*NET. Take the attribute 'reaction time,' defined as the ability to quickly respond with the hand, finger, or foot to a signal (sound, light, picture) when it appears. Write three behavioral anchors that illustrate increasing proficiency: low is 'start to slow down the car when a traffic light turns yellow,' moderate is 'throw a switch when a red warning light goes off,' and high is 'hit the brake when a pedestrian steps in front of the car.' In a performance conversation you now point to a concrete behavioral level rather than a vague rating, so the employee can see exactly what higher performance looks like. This is the developmental core of performance management: goals and feedback tied to observable behavior on a defined scale, informed by the same work analysis that grounds selection and training.

• If an attribute can be learned on the job, route it into the development and training focus of the system; if it is required at entry, treat it as a selection/expectation baseline.
• If you have time and cost constraints, still use at least two methods so you can triangulate rather than relying on a single biased source.
• If incumbents cannot articulate what they do (low verbal ability or low motivation to report accurately), rely more on supervisors, who bring higher verbal ability, a broader hierarchical perspective, and less motivation to distort.
• If you need the ideal or technical view of how the work should function, use technical experts (engineers, chemists, lawyers, professors) rather than incumbents.
• If your rating scales for level, importance, and criticality keep correlating in the .80s-.90s, do not assume you are capturing distinct dimensions—workers often lose these distinctions, so simplify or add clearer behavioral anchors.

• Treating performance management as a periodic appraisal only. The tell: the system produces a single annual score with no behavioral anchors or continuous feedback loop.
• Relying solely on incumbent self-report. The tell: descriptions are vague or inflated because some incumbents lack verbal ability or motivation to report accurately.
• Using outdated documentation as ground truth. The tell: job descriptions or manuals lack the depth or breadth needed for the intended purpose and no longer match daily work.
• Confusing conceptually different rating scales. The tell: level and importance ratings correlate in the low .90s, showing raters are not distinguishing them, so your data cannot support the design decisions you claim.
• Designing the system without stakeholder input. The tell: the resulting HR system is technically defensible but resisted or ignored because the people whose work it measures never contributed.

• ▸Work role / target: name the role being analyzed.
• ▸Attribute or behavior: state and define the attribute needed for effective performance.
• ▸Entry vs. developable: mark whether it is required at hiring or can be trained on the job.
• ▸Data sources used: list documentation, incumbents, technical experts, supervisors consulted.
• ▸Methods used: note observation, interviews, group meetings, or questionnaires for triangulation.
• ▸Low anchor: write an observable behavior illustrating low proficiency.
• ▸Moderate anchor: write an observable behavior illustrating moderate proficiency.
• ▸High anchor: write an observable behavior illustrating high proficiency.
• ▸Feedback/development action: state the continuous feedback or developmental step tied to this attribute.

1. 1Anchor the effort in work analysis first: define the target role's work requirements (tasks, responsibilities, activities) and worker requirements (knowledge, skill, ability, other characteristics) before designing any training, because work analysis is the foundation for training program development.
2. 2Choose a data-collection method deliberately. For gathering a large volume of competency data that you can summarize quantitatively, use a structured questionnaire (e.g., task inventory, PAQ, or O*NET); accept the trade-off that long instruments risk overwhelming respondents and producing unreliable data.
3. 3Select rating scales that match what you need. Use time-based frequency scales (from 'about once per year' to 'about once each hour or more often') when task frequency drives training priority, and importance or needed-at-entry ratings when deciding which competencies to develop for current versus future performance.
4. 4Treat competency data as subject to systematic, not just random, error. Do not assume sampling and standardized materials fix everything; account for rater characteristics like job familiarity (Madden) and job tenure (Prien & Saleh) that predictably shift ratings.
5. 5Build the development pathway across the three formal channels the handbook separates: training and employee development for structured skill acquisition, mentoring for relational development, and executive coaching for individualized senior development. Match each competency gap to the channel it fits.
6. 6Evaluate the resulting competency profile against completeness (comprehensiveness) and mean ratings (inappropriately elevated or depressed), so the training targets are not distorted before design begins.
7. 7Close the loop by feeding validated competency requirements into program design, then re-analyze the role periodically because roles connect to a broader system of roles and shift over time.

Suppose you must build training for a customer-service role. You start with work analysis: you distinguish work requirements (handling calls, resolving complaints) from worker requirements (product knowledge, conflict-resolution skill). You deploy an O*NET-style structured questionnaire and ask incumbents to rate each competency on both a frequency scale and a needed-at-entry scale. When you check the data, you notice single-rater importance ratings vary widely; consistent with Van Iddekinge et al. (2005), who found 21.6% of error variance in importance ratings and 29.1% in needed-at-entry ratings traceable to rater idiosyncrasies, you aggregate across more raters rather than trusting one. Product knowledge scores high on needed-at-entry, so you route it to formal training; conflict-resolution scores as developed-on-the-job, so you route it to mentoring. You then verify the competency list is complete before designing modules.

• If a competency is rated high on 'needed-at-entry,' then build it into formal pre-employment or onboarding training rather than developmental coaching.
• If you need to summarize competency data across many employees quantitatively, then use a structured questionnaire; but if respondents face hundreds of items and multiple scales, then split administration to protect reliability.
• If a single rater's data drives your training priorities, then treat it as suspect and add raters, because 20-29% of rating variance can be pure rater idiosyncrasy.
• If a competency gap is relational or judgment-based rather than procedural, then assign it to mentoring or executive coaching instead of a training module.
• If a role connects tightly to a broader system of interdependent roles, then analyze those connections before designing training, so individual development does not misalign with collective action.

• Assuming all rating error is random and 'fixing' it only through sampling and standardization; the tell is stable-looking averages that still shift when you change who rates.
• Designing training from a job description alone instead of a systematic work analysis; the tell is competency targets nobody can trace to actual task or worker requirements.
• Overloading the questionnaire (several hundred items, many scales); the tell is respondent fatigue producing unreliable, inaccurate ratings.
• Ignoring rater characteristics such as job familiarity and tenure; the tell is systematic gaps between novice and veteran ratings that get averaged away.
• Treating training, mentoring, and coaching as interchangeable; the tell is procedural modules used to fix relational or executive-judgment gaps that never close.

• ▸Role/Work requirements: list tasks and responsibilities from work analysis
• ▸Worker requirements: list needed knowledge, skill, ability, other characteristics
• ▸Data method used: name the questionnaire (task inventory / PAQ / O*NET) and why
• ▸Rating scales applied: note frequency (time-based) and importance/needed-at-entry scores
• ▸Rater safeguards: number of raters and check for familiarity/tenure effects
• ▸Completeness check: confirm the competency list is comprehensive (yes/no + gap)
• ▸Development channel: assign each competency to training, mentoring, or coaching
• ▸Current vs. future performance: mark whether the gap serves present or later needs
• ▸Re-analysis date: set when to revisit given role-system connections