RESEARCH DOSSIER
Workplace Training, Skill Development, and Transfer
Abstract. Workplace training, skill development, and transfer concern the organized acquisition, refinement, generalization, and maintenance of occupational competence. This critical narrative monograph reconstructs historical and conceptual foundations of organizational training psychology and integrates primary empirical studies, quantitative syntheses, and methodological reviews of needs analysis, instructional design, learning processes, deliberate practice, motivational and contextual predictors, performance measurement, team training, error management, and transfer. The literature supports meaningful effects of structured training on learning and task performance, while revealing substantial differences among immediate acquisition, on-the-job application, adaptive generalization, and retained capability. The organizational environment shapes whether learned behavior becomes observable practice. Artificial assistance creates additional scientific questions about the distinction between supported productivity and persistent human skill. The synthesis defines the causal and measurement problems required for rigorous evaluation across changing work systems.
Keywords: workplace training; industrial and organizational psychology; organizational learning; training transfer; skill development; expertise; instructional design; deliberate practice; transfer climate; self-regulated learning; error management; knowledge retention; team training; human factors; artificial assistance; Artificial Era.
Scientific Overview and Research Questions
Workplace training is an organized intervention intended to develop knowledge, skill, judgment, and action relevant to work. Skill development is the acquisition and refinement of these capabilities across experience and practice; transfer of training is their application to job tasks and circumstances beyond the instructional environment. The three concepts define related but distinct research problems. An employee can acquire procedural knowledge during an instructional session and subsequently fail to apply it because the actual job differs, the organization prohibits the new behavior, or the capability has decayed.
This monograph examines the scientific explanation of occupational learning from the original training event to durable and adaptive performance. Its questions concern trainable competence, instructional mechanisms, learner readiness, the role of practice and feedback, the effects of organizational conditions, and the measurement of transfer. It also considers the new distribution of work between people and artificial systems. Salas, E., Tannenbaum, S. I., Kraiger, K., & Smith-Jentsch (2012) and Aguinis, H., & Kraiger, K (2009) establish the value of treating training as a connected system of preparation, intervention, and posttraining application.
The object of inquiry is organizational skill intervention, including its assessment and workplace consequences. Learning in general, broad career development, and generalized cognitive ability enter as explanatory resources rather than competing canonical topics. This boundary permits a scientifically coherent treatment of training without equating an organization's training activity with evidence that a new competence exists.
Scientific Method and Evidence Hierarchy
The investigation follows a critical narrative synthesis of original experiments, field studies, validated theoretical frameworks, meta-analyses, and scholarly reviews. Historical and conceptual claims are linked to identifying primary publications; quantitative claims are restricted to outcomes explicitly reported in accessible original abstracts or papers. No registered systematic-review protocol, comprehensive multi-database search, duplicate screening, or new statistical meta-analysis was undertaken. Accordingly, the present synthesis organizes established evidence and unresolved questions rather than claiming the evidentiary completeness of a PRISMA systematic review.
Claims are evaluated according to study design and endpoint. Randomized comparisons can establish the effect of a defined training manipulation under its assigned conditions; observational correlations identify relationships and predictors that may reflect selection or contextual confounding. Longitudinal assessments can speak to persistence, but elapsed time alone does not eliminate confounding. Meta-analytic averages depend on included studies, measures, heterogeneity, and publication processes. Ford, J. K., Baldwin, T. T., & Prasad, J (2018) shows why precision about transfer definitions and measures remains an active methodological problem.
The central methodological commitment is to keep acquisition, generalization, maintenance, and work-system results analytically separate. A knowledge test, a self-report about intended use, an observed new procedure, and a safety outcome are different evidence. Conclusions are proportionate to the weakest causal or measurement link required by the claim.
Historical Foundations of Organizational Training
Institutional apprenticeship and workplace instruction long predate modern experimental research. Industrial psychology, military instruction, personnel development, and human factors progressively introduced systematic task analysis, performance criteria, controlled instructional comparisons, and evaluative methods during the twentieth century. The historical trajectory is a transition from recording training attendance toward testing whether specific behaviors can be learned and used in consequential environments.
Bell, B. S., Tannenbaum, S. I., Ford, J. K., Noe, R. A., & K (2017) reviews a century of training and development research in the Journal of Applied Psychology, documenting the expanding concern with learner differences, work context, technology, and organizational outcomes. An influential turning point was Baldwin, T. T., & Ford, J. K (1988), which characterized transfer by generalization from training to work and maintenance over time. Its organizing distinction among learner characteristics, instructional design, and workplace environment still directs empirical research.
This history also includes repeated critique of evaluation shortcuts. Alliger, G. M., & Janak, E. A (1989) challenged assumptions surrounding a common four-level evaluation framework, and Alliger, G. M., Tannenbaum, S. I., Bennett, W., Jr., Traver (1997) synthesized the relationships among its outcome criteria. Later work refined the meaning of transfer, moved beyond classroom delivery, and examined skill use within teams and technologically mediated work. Historically significant frameworks provide an intellectual map; their specific causal propositions still require valid experiments and measurements.
Defining Skill, Competence, and Expertise
A workplace skill is a capacity to accomplish a class of tasks to a meaningful performance criterion. Competence includes coordinated knowledge, perception, judgment, action, and the ability to identify conditions requiring help or escalation. Expertise implies the development of reliable domain-specific representations and flexible discrimination among ordinary, exceptional, and ambiguous cases. These constructs must be operationalized for the job rather than inferred from a worker's title, confidence, or time in service.
Kraiger, K., Ford, J. K., & Salas, E (1993) distinguished cognitive, skill-based, and affective learning outcomes. This taxonomy clarifies why knowledge recall, observable execution, emotional commitment, and self-efficacy cannot be scored as equivalent evidence of training success. An employee may explain a hazard accurately but respond slowly under pressure; another may perform a rehearsed sequence without knowing when that sequence would be unsafe.
Construct-valid assessments begin with the actual work requirement. For customer service, a competence framework might distinguish policy interpretation, empathetic communication, handling an unfamiliar case, and justified escalation. Each requires representative test cases and an appropriate scoring rubric. A valid monograph of workplace training therefore defines what the individual is expected to do, under which circumstances, and how evidence of success will be distinguished from merely polished performance during instruction.
Needs Assessment and the Organizational Problem
Training needs analysis specifies the performance problem that instruction can plausibly solve. Organizational analysis identifies demands, constraints, incentives, and available resources. Task analysis identifies the actions, judgments, cues, and error modes associated with performance. Person analysis describes individual preparation and specific capability gaps. A failure of competence may arise from deficient knowledge, but also from missing equipment, workload, conflicting incentives, or inaccessible information. Additional training cannot reliably correct every work-system failure.
The organizational framework in Salas, E., Tannenbaum, S. I., Kraiger, K., & Smith-Jentsch (2012) requires careful diagnosis before curricula are chosen. Direct observation, incident records, critical-event interviews, work-product audits, and expert–novice comparisons can expose features omitted by nominal job descriptions. The result should be a documented causal hypothesis: which capability, developed through which experiences, should change which observable performance measure?
Training objectives must be revisited as work changes. New software, revised regulation, organizational restructuring, and the adoption of artificial assistance may alter the skills required for sound practice. Needs assessment is therefore a continuing scientific inquiry into occupational competence rather than a one-time administrative exercise. Its quality determines whether later training results can be meaningfully interpreted.
Task Decomposition and Competency Architecture
A competency framework translates broad occupational expectations into observable tasks and decision points. Labels such as leadership, accuracy, collaboration, or adaptability are useful only after their operational meaning has been specified. What must the employee notice, decide, communicate, execute, record, and verify? Which variants of the task count as successful, and which kinds of mistake are consequential?
Decomposition supports systematic instruction but creates a risk of teaching isolated fragments without the relations that make them meaningful. For complex occupations, a training architecture should identify prerequisite elements, integrated routines, variable-case judgment, exceptional situations, and recovery after failure. Learners may be competent at predictable cases yet unable to recognize a rare exception. Consequently, an overall pass/fail certificate cannot replace a map of functionally distinct skills.
The map also determines what transfer can mean. Training examples should represent a designed sample of the task domain, while assessment includes cases that differ in significant but principled ways. If all test cases reproduce the same wording or interface, the measure is closer to rehearsal than occupational generalization. A scientifically defensible program specifies the dimensions of variation before declaring the skill mastered.
Learner Differences and Training Readiness
Workers enter a program with differing knowledge, opportunities, language backgrounds, prior practice, motivation, confidence, and experiences of the organization. Readiness describes the relation between a particular learner and instructional demands. It is not a fixed essence of a person. Preassessment is valuable when its purpose is to vary support, task difficulty, or practice rather than to sort employees into permanent categories.
Blume, B. D., Ford, J. K., Baldwin, T. T., & Huang, J. L (2010) synthesized 89 empirical studies of transfer and reported relationships involving trainee cognitive ability, conscientiousness, motivation, and aspects of the work environment. Such observational predictors are informative at the population and research-design level. They do not establish that a particular worker will fail, nor do they authorize treating access to training as a reward for preexisting privilege.
Prior experience may help learners recognize task structure, but entrenched routines can also inhibit adaptation. A veteran employee facing a new protocol may require opportunities to compare old and new causal rules. Novices may need demonstrations and guided practice before independent cases. The relevant scientific question is which adaptation of instruction changes later performance while keeping the assessment criteria equitable and aligned with actual tasks.
Motivation and Goal-Directed Learning
Training motivation involves entering a learning program, sustaining practice, and applying competence afterward. These stages have different causes. An employee may be eager to learn but reluctant to use a new procedure when coworkers disregard it or when managerial incentives favor the former approach. Motivation to transfer should therefore be studied separately from satisfaction with instruction.
Gegenfurtner, A., Veermans, K., Festner, D., & Gruber, H (2009) reviewed the construct of transfer motivation and its antecedents; Grossman, R., & Salas, E (2011) identified motivation, self-efficacy, utility beliefs, and contextual support among salient transfer correlates. These findings point toward explicit relevance and a realistic account of expected behavior at work. Motivation must be joined to opportunity, since a willing employee cannot demonstrate a skill without relevant assignments.
Locke, E. A., & Latham, G. P (2002) synthesized goal-setting research showing the importance of specific, appropriately challenging goals and feedback under suitable conditions. In training, the goal must represent correct performance rather than superficial speed. A numerical target that rewards premature completion can undermine careful reasoning. Good learning goals direct attention to interpretable errors, useful practice, and an accountable standard of performance.
Self-Regulated Learning and Metacognition
Workplace learners plan their practice, allocate attention, choose strategies, monitor success, and adapt when an approach fails. These self-regulatory activities become especially important in digital or asynchronous programs, where the trainee may have substantial responsibility for determining when and how to study. Activity data such as time logged in, completed modules, or number of clicks are only indirect indicators of these processes.
Sitzmann, T., & Ely, K (2011) meta-analyzed self-regulated learning across training and educational settings, connecting a range of cognitive and motivational regulatory processes with outcomes. This body of work motivates task-specific measures of strategy use, calibration, effort, and corrective adjustment. It also reminds investigators to distinguish what employees believe they learned from what independent performance tests show.
An intervention can make self-regulation teachable: learners predict performance, attempt a task, examine feedback, explain mistakes, select a revised strategy, and try an altered case. When uncertainty is explicit, a successful employee may know when to consult a reference or ask for help. Metacognitive improvement is demonstrated through better subsequent decisions, not through the fluent vocabulary of self-reflection alone.
Instructional Design and Scaffolding
Instructional architecture joins task objectives, examples, explanations, practice conditions, feedback, and independent assessment in a coherent sequence. Early demonstrations or worked examples may reduce unnecessary search for novices. As competence grows, guidance can be faded and decision demands expanded. The purpose of support is to build capacity for the conditions in which an employee will eventually work.
Training design must reflect the kind of skill at issue. A policy rule may benefit from explanation and retrieval practice; sensorimotor skill requires action and meaningful performance feedback; complex interpersonal judgment needs varied scenarios and perspective-sensitive evaluation. Salas, E., Tannenbaum, S. I., Kraiger, K., & Smith-Jentsch (2012) treats these choices as part of a larger program that includes learner preparation and workplace support.
Scaffolding has a characteristic failure mode: a trainee performs competently only while prompts, cues, or an instructor supply the decisive cognitive work. A valid design therefore tests performance after support is reduced, including situations with unfamiliar surface features. The relation between training support and independence is a central variable, especially as artificial tutors begin to supply near-continuous guidance.
Modeling, Explanation, and Mental Models
Demonstrations can show what an expert does, but competent performance often depends on why an action is selected and which cues distinguish one circumstance from another. Modeling is most useful when reasoning, alternatives, relevant feedback, and diagnostic checks are made visible. Otherwise learners may memorize surface actions while missing the judgment structure that makes them appropriate.
Mental models represent the organization of a task or system. In troubleshooting, for example, a technician needs to understand how observable indicators relate to possible causes and which tests can discriminate among them. In interpersonal work, an employee must interpret evolving intentions rather than deliver a fixed script. Training should therefore include explanatory comparisons, counterexamples, and opportunities to describe causal relationships.
Observable competence is the arbiter. An articulate explanation may coexist with poor action under time pressure. Conversely, skilled behavior can develop before verbal theories become comprehensive. An adequate training program aligns explanation with repeated performance in cases that challenge both overgeneralization and unjustified rigidity.
Behavior-Modeling Training
Behavior-modeling training typically combines demonstration, explicit learning points, rehearsal, feedback, and plans for workplace use. It is especially relevant to leadership, customer interactions, conflict management, and other tasks with observable interpersonal components. The mechanism extends beyond imitation: the learner must discriminate which behavior is suitable under which social and operational conditions.
Taylor, P. J., Russ-Eft, D. F., & Chan, D. W. L (2005) meta-analyzed 117 studies of behavior-modeling training. The synthesis found stronger learning outcomes than job behavior or organizational results and identified program features associated with transfer, including the treatment of models, practice, and managerial context. These associations help guide design while leaving open the causal contribution of any single component in particular occupations.
A frequent error is to reward reproducing model language while overlooking contextual appropriateness. An employee who accurately recites a communication script may still misunderstand a customer's need. Assessment should score situational judgment, response to unexpected information, recognition of limits, and the ability to adapt without abandoning the underlying principle. Work-context support remains part of the explanation for whether newly modeled behavior is sustained.
Practice, Spacing, and Long-Term Retention
Practice influences the durability and accessibility of what employees learn. Repeated exposure close together can generate immediate fluency while leaving later recall fragile. Distributed practice introduces intervals that require reconstruction and can improve delayed memory performance. The relevance of spacing to workplace instruction is substantial, though exact timing and magnitude are task dependent.
Cepeda, N. J., Pashler, H., Vul, E., Wixted, J. T., & Rohrer (2006) synthesized 317 experiments with 839 assessments from 184 articles on distributed practice in verbal recall. Their findings show that spacing effects depend partly on the desired retention interval. Most underlying tasks were memory tasks rather than complete occupational performances, so the same optimal intervals cannot be assumed for technical repairs, clinical coordination, or a safety-critical motor sequence.
Workplace training can employ scheduled scenario refreshers, delayed checks, and recurrent exposure to low-frequency decisions. The evidence-based objective is durable capability under relevant task conditions. Evaluations should measure performance after meaningful delays and should record opportunities to use the skill between assessments. Spacing should be treated as a mechanism to test and adjust, rather than a universally fixed training calendar.
Retrieval Practice and the Difference Between Familiarity and Knowledge
Retrieval practice requires learners to reconstruct knowledge from memory rather than only reread or recognize it. Its value becomes visible when delayed performance matters. Roediger, H. L., III, & Karpicke, J. D (2006) found in experiments with prose passages that retrieval could improve later retention even when additional studying seemed more favorable immediately. The distinction between feeling familiar with material and being able to retrieve it later is important for organizational training.
Appropriate workplace retrieval tasks can include identifying an escalation threshold, selecting a procedure in an unfamiliar case, or explaining a diagnostic decision before the correct response is displayed. The test should reflect the decisions employees face. Quizzing irrelevant terminology may establish recall but has little direct meaning for operational competence.
Retrieval is only one component of complex skill. Many work tasks require motor execution, perceptual discrimination, coordination, or skilled use of external resources. The goal is to train what must be remembered, what must be recognized, and when to consult validated information. Delayed and representative assessment is necessary to determine whether retrieval practice produces occupational benefits.
Feedback and Performance Calibration
Feedback provides information about the relation between an attempted action and a performance criterion. Useful feedback identifies the specific error, its significance, and an available correction. Praise may influence motivation, but generic approval cannot replace information about what has been done accurately. In some occupations immediate outcomes are misleading: a risky shortcut may appear to succeed until a rare failure occurs.
A feedback cycle can ask employees to predict performance, execute a task, observe validated results, explain discrepancies, and repeat with changed circumstances. This strengthens calibration between subjective confidence and actual ability. However, extensive instructor feedback can also produce reliance on an external decision maker. Subsequent unsupported trials are needed to establish independent competence.
In controlled evaluations, the benefits of feedback should be separated from those of additional practice time or instructor attention. In organizational settings, the climate for acknowledging mistakes affects whether corrective information is disclosed. Feedback operates within a learning relationship and a work system; both are necessary to understand its causal impact.
Error-Management Training and Adaptive Transfer
Error-management training encourages controlled exploration, mistakes, and learning from correction. It is suited to problems requiring adaptation beyond a rehearsed script. In a meta-analysis of 24 studies involving 2,183 participants, Keith, N., & Frese, M (2008) found an overall Cohen's d of 0.44. The reported moderator analyses showed larger effects for posttraining transfer (d = 0.56) and for structurally different adaptive transfer tasks (d = 0.80) than for more closely matched assessments.
These estimates are properties of the included experimental comparisons. They do not imply that every workplace should encourage uncontained mistakes. In patient care, industrial operations, transportation, and privacy-sensitive work, exploration must be placed in simulation, reversible sandboxes, or carefully supervised conditions. Controlled error experience can develop the recognition, diagnosis, and correction skills necessary for reliable operation.
This evidence challenges an exclusive focus on flawless training-session speed. A learner who explores why a procedure fails may perform less smoothly during rehearsal and more effectively when a novel case arises later. Training studies should select outcomes that can detect this distinction.
Simulation and Safety-Critical Rehearsal
Simulations provide access to expensive, dangerous, or infrequent situations in a controlled environment. Their instructional value is determined primarily by whether relevant information, time pressure, decisions, interpersonal demands, and feedback contingencies match the real task. A visually impressive simulation can still train incorrect habits if the decisive cues or consequences are wrong.
Simulation design should include ordinary and anomalous cases, progressively greater complexity, error recovery, and debriefing that links actions to outcomes. For high-consequence tasks, criteria may include safe stopping, escalation, effective teamwork, and adherence to validated procedures. The setting should permit mistakes during training while protecting people and equipment from preventable harm.
Evidence of simulator mastery requires subsequent validation when meaningful. Employees can perform well with stable equipment, familiar teammates, and known task order yet struggle with an operational system that differs. Training should therefore include structured variation, and assessments should specify which properties of the real setting are reproduced. Functional fidelity and demonstrable transfer matter more than appearance.
Digital, Classroom, and Blended Training
The format of instruction affects reach, pacing, interaction, and cost. It does not automatically determine learning quality. Sitzmann, T., Kraiger, K., Stewart, D., & Wisher, R (2006) synthesized comparisons between web-based and classroom instruction and examined the influence of instructional methods. Differences in learning may arise from the way practice and feedback are organized rather than from the medium itself.
Digital programs can distribute material across shifts, provide adaptive problem sets, and document delayed performance. In-person training can facilitate physical demonstration, immediate dialogue, and team rehearsal. Blended programs combine these capacities when each component has a clear instructional purpose. Using two media to repeat the same lecture may increase exposure without adding relevant learning mechanisms.
A credible comparison must consider equivalence of content, assessment difficulty, instructor attention, learner control, practice time, and the organizational setting. Technology acceptance and employee satisfaction are legitimate secondary outcomes; they do not verify mastery. Inferences about effective delivery must be made against the target competence and the measured transfer criterion.
Informal Learning, Mentoring, and Situated Experience
Employees acquire considerable knowledge outside scheduled courses through observation, peer conversations, feedback from real work, supervised problem solving, and the gradual assumption of responsibility. Noe, R. A., Clarke, A. D. M., & Klein, H. J (2014) reviews how workplace learning extends across formal, informal, and technology-mediated arrangements. These arrangements can complement or displace classroom instruction depending on the job.
Experience alone does not guarantee development. A worker can repeat a narrow task for years and gain little exposure to complex cases. High-quality mentoring offers diagnostically useful feedback, appropriately challenging tasks, opportunities to explain decisions, and support for corrected attempts. It also creates conditions in which previously invisible reasoning can be discussed.
Because access to assignments is uneven, workplace learning can amplify existing inequalities. Researchers should measure what opportunities workers actually encounter and how those opportunities differ. An assessment of informal learning needs independent performance criteria, not merely the number of years employed or the frequency of interactions with a supervisor. The operational work setting is simultaneously an instructional resource and a constraint on what development becomes possible.
Team Training and Collective Performance
Teams need competencies that individuals cannot demonstrate alone: maintaining shared task understanding, coordinating handoffs, monitoring one another's workload, speaking up about hazards, and adapting to changing roles. Joint training may therefore improve collective performance even when every individual already understands the relevant procedure.
Salas, E., DiazGranados, D., Klein, C., Burke, C. S., Stagl (2008) meta-analyzed 93 effect sizes representing 2,650 teams and identified positive outcomes of team-training interventions across examined criteria, with variation related to features such as training content and team characteristics. The results support an evidence-based interest in team training without assuming that generic group activities have equivalent effects.
Transfer at team level requires that communication and coordination persist when real staffing changes. In healthcare, aviation, manufacturing, or digital service, teams may operate with unfamiliar members, remote participation, or changing tools. Evaluation should include actual task coordination under realistic uncertainty, not only survey responses about cohesion. The distinction between individual skill and collective competence must be preserved in analysis, measurement, and intervention design.
The Baldwin–Ford Model of Training Transfer
Baldwin, T. T., & Ford, J. K (1988) established a durable framework for thinking about generalization of learned material to work and maintenance of the resulting skill over time. Its three broad antecedent classes—trainee characteristics, training design, and work environment—explain why exposure to instruction is only one part of successful occupational development.
The model's practical force appears when apparently good training produces little observable change. If the individual never acquired the procedure, workplace support alone is insufficient. If an employee acquired it but lacks relevant tools or opportunities, additional instructional repetition may have limited effect. If work practices reward behavior inconsistent with the new procedure, transfer can be suppressed despite competence and intention.
Ford, J. K., Baldwin, T. T., & Prasad, J (2018) revisited the known and unknown in transfer research, emphasizing its measurement and conceptual problems. This distinction prevents an influential theoretical framework from being treated as proof of every causal link it proposes. Experimental and longitudinal studies must test how the antecedent classes interact in specific jobs and populations.
Near Transfer, Far Transfer, and Adaptive Generalization
Training is most readily evaluated on cases similar to those taught. Such near-transfer tests can be important when a job depends on precise routine execution. Far or adaptive transfer asks whether acquired principles operate under materially changed conditions. The latter is particularly difficult because investigators must justify that the new task shares a relevant structure with the training material.
Generalization should be operationalized using explicit dimensions of change, such as different equipment, participants, interface layouts, information quality, timing, or underlying causal rules. Maintenance adds an independent time dimension. An employee may preserve routine accuracy for months yet fail to diagnose an unusual fault; another may adapt well to new cases but lose speed on an infrequently used procedure.
Evidence from Keith, N., & Frese, M (2008) suggests that training methods can differ more sharply on adaptive than on immediate task performance. The research implication is to design evaluation sets with both familiar and meaningfully changed cases and to test them after appropriate delays. A single posttest score cannot characterize the structure of transfer.
Opportunity to Perform and Workplace Task Allocation
Workplace application is contingent on opportunities to use the trained skill. Employees who receive no relevant assignments cannot demonstrate transfer, regardless of whether they learned successfully. Ford, J. K., Quiñones, M. A., Sego, D. J., & Sorra, J. S (1992) investigated factors affecting employees' opportunity to perform trained tasks, providing a foundation for treating opportunity as a separate explanatory variable.
Task assignment may also be systematically unequal. Managers may direct complex cases to workers who already possess experience, producing further growth for them while depriving novices of practice. Flexible workers, staff with disabilities, remote workers, and employees with different schedules may encounter distinct access to meaningful tasks. Those differences can confound estimates of training effects.
A sound evaluation records which employees had chances to apply the trained behavior, how demanding the tasks were, what support was available, and how performance varied conditional on opportunity. Low recorded application may be a failure of instructional design, motivation, or organizational allocation. Distinguishing these explanations is necessary before deciding whether more training is justified.
Supervisor Support, Peer Norms, and Transfer Climate
Supervisors influence transfer through assignments, feedback, incentives, emotional safety, and tolerance for the temporary inefficiency associated with learning. Peers influence it through local norms about correct practice, acceptable errors, and whether new procedures are socially rewarded or resisted. Grossman, R., & Salas, E (2011) and Burke, L. A., & Hutchins, H. M (2007) synthesize evidence linking transfer to learner, design, and workplace factors.
Support should be operationalized behaviorally. Does the supervisor allocate practice time? Are checklists available when needed? Does the team discuss errors without retaliation? Are correct judgments rewarded even when they require more time? A favorable attitude toward training with no change in the work system is weak evidence of actual support.
Related field evidence from Edmondson, A (1999) links psychological safety with learning behavior in 51 manufacturing teams. The study provides an important social mechanism but does not establish a universal randomized effect across all occupations. The broader scientific claim is that training use depends partly on an environment where questioning, trying new practices, and reporting problems are feasible.
Motivation to Transfer and the Transfer System
The transfer system is a constellation of expectations, incentives, capabilities, supervisor responses, job opportunities, and practical constraints affecting whether learning is put to use. Holton, E. F., III, Bates, R. A., & Ruona, W. E. A (2000)11:4%3C333::AID-HRDQ2%3E3.0.CO;2-P) developed the Learning Transfer System Inventory to measure multiple aspects of this environment rather than treating transfer as a single intention or satisfaction rating. Its psychometric approach highlights the importance of specifying constructs when studying organizational conditions.
Self-reported assessments can identify perceived barriers and expectations, but they do not independently demonstrate actual behavior. A worker may sincerely report managerial encouragement and still encounter few relevant assignments. Conversely, an employee may use a skill routinely without describing high motivation. Triangulation with observed work, supervisor behavior, and task allocation improves interpretation.
Gegenfurtner, A., Veermans, K., Festner, D., & Gruber, H (2009) helps connect motivation to the conditions of transfer. Researchers should test whether motivational changes predict performance beyond skill acquisition, and whether organizational interventions that supply opportunities or remove barriers produce independent effects. A transfer system is an empirical object that can be measured and changed, not merely a metaphor for a successful course.
Posttraining Reinforcement and Maintenance
Training programs often terminate at certification, while the actual process of sustaining a new skill begins when employees return to their tasks. Planned follow-up may include supervisor observation, refresher cases, spaced opportunities to perform, updated job aids, and feedback about early implementation. Their objective is to prevent a gap between instructional mastery and sustained workplace practice.
Saks, A. M., & Belcourt, M (2006) surveyed training professionals in 150 organizations and examined the relation between transfer and activities before, during, and after training. Their reports indicate concern about diminishing application and the importance of pretraining and posttraining support. Frequently repeated transfer-loss percentages from such surveys should not be treated as direct universal measurements of employees' retained skills.
A maintenance study needs independent measures over time and data about what the employee encountered between tests. Where practice opportunities are scarce, low apparent use may reflect assignment patterns. Where error rates rise after a policy change, retraining may be appropriate; where staffing or technology prevents a known procedure from being used, the intervention may require work redesign. The relevant aim is reliable competence in the job, not endless repetition of classroom content.
Training Evaluation and Criterion Separation
A common evaluation framework distinguishes trainees' reactions, what they learned, whether work behavior changed, and organizational results. These categories make different research questions visible. Alliger, G. M., & Janak, E. A (1989) examined the assumptions of this hierarchy, while Alliger, G. M., Tannenbaum, S. I., Bennett, W., Jr., Traver (1997) analyzed how the criteria relate empirically. The categories should not be treated as an automatic causal staircase.
Reaction surveys can reveal course relevance and experience; knowledge and skill tests can measure acquisition; workplace observation can assess application; organizational records may indicate consequences. Positive reactions can coexist with poor learning, and improved organizational outcomes can arise from changes in technology, staffing, demand, or incentives. Researchers must therefore specify their proposed causal chain and test the stages they claim.
An evaluation design should reflect the intervention's intended function. A safety curriculum calls for accurate procedures and risk-sensitive outcomes; a troubleshooting course needs performance on fault scenarios, including novel ones; interpersonal training requires valid observation of real interactions. Outcome significance depends on the costs of errors, not simply on a statistically significant score.
Causal Research Designs and Counterfactuals
The scientific question of whether training caused improvement requires an appropriate counterfactual: how would comparable workers have performed without this intervention or under a credible alternative? Randomized assignment offers a strong design when feasible. In organizational research, allocation may occur at individual, team, or site level depending on contamination and operational constraints.
Field evaluation faces threats from selection, supervisor task assignment, concurrent process changes, attrition, and participants exchanging materials. An observed improvement after a course may reflect increased experience or a modified system. Difference-in-differences, matched comparisons, and interrupted time series can provide evidence where randomization is impractical, but each requires explicit assumptions and sensitivity analyses.
The evaluation plan should specify primary outcomes, assignment unit, intervention fidelity, follow-up period, attrition, clustering, and plausible alternative explanations before results are known. Confidence intervals and task-relevant effect sizes allow practical interpretation. A strong field experiment measures both learning during instruction and occupational performance afterward, thereby revealing which stage the intervention actually changed.
Validity, Reliability, and Performance Measurement
Measures of occupational competence vary in relevance and error. Knowledge tests can sample declarative understanding, simulations can assess structured execution, blinded evaluators can rate work products, and operational logs can capture selected real-world outcomes. Each instrument requires evidence for its intended interpretation.
The theoretical taxonomy in Kraiger, K., Ford, J. K., & Salas, E (1993) helps align measures with cognitive, skill-based, and affective constructs. Reliability can be studied across raters, tasks, alternative forms, and time. Validity requires that assessment cases represent the defined skill rather than peripheral features such as platform familiarity, social approval, or linguistic complexity unrelated to the job. A fast completion metric can reward premature decisions; supervisor ratings can reflect reputation and role visibility.
Transfer measures should record who rated performance, whether the assessor knew training allocation, whether employees had opportunities to use the skill, and how difficulty varied between cases. Measures should sample tasks beyond those rehearsed exactly during instruction. When possible, combine independently scored performance, task opportunity, work products, and carefully interpreted self-report.
What Organizational Training Meta-Analyses Establish
Arthur, W., Jr., Bennett, W., Jr., Edens, P. S., & Bell, S (2003) reported sample-weighted average standardized effects of d = 0.60 for reactions, 0.63 for learning, 0.62 for behavior, and 0.62 for organizational results across the corresponding study subsets. These findings provide substantial evidence that organizational training interventions can have measurable consequences. They are not a universal forecast for any one course or organization.
Meta-analysis requires careful attention to differences among the included studies. The effects summarized under the same broad criterion may concern different jobs, interventions, outcome definitions, controls, and follow-up periods. Outcome heterogeneity and the inclusion of several measures from the same study complicate simple interpretation. Publication and reporting patterns also influence the apparent distribution of results.
Blume, B. D., Ford, J. K., Baldwin, T. T., & Huang, J. L (2010) analyzed associations between transfer and learner or contextual factors rather than randomized manipulations of every predictor. Taylor, P. J., Russ-Eft, D. F., & Chan, D. W. L (2005) and Keith, N., & Frese, M (2008) synthesize particular intervention families with distinct task conditions. A careful reader should therefore ask which comparison produced the estimate, for whom, using which measure, and over what interval before translating an average into an organizational decision.
Expertise and Deliberate Practice
Expertise arises through sustained engagement with domain problems, appropriate feedback, refinement of mental representations, and the ability to respond to variable conditions. Deliberate practice describes structured activity directed toward improvement rather than repetition alone. It provides a useful language for analyzing development but does not support a single hours-based formula for all occupations.
Macnamara, B. N., Hambrick, D. Z., & Oswald, F. L (2014) synthesized deliberate-practice research across several domains and found large differences in the amount of performance variation associated with practice. The reported share of explained variance was especially small in professional domains in the included evidence. This finding challenges the idea that practice quantity is a sufficient account of professional differences. It does not invalidate the causal importance of effective practice within a particular task.
Professional competence involves decision making, interaction with tools, responses to exceptions, and awareness of one's limits. Training should therefore include feedback-rich problems that expose and repair inaccurate rules, with progression from basic operations to adaptive performance. Longitudinal measurement is necessary to explain how expertise develops rather than treating professional status or accumulated years as direct substitutes for measured skill.
Lifelong Learning and Occupational Reskilling
Training requirements change as employees take on new roles, technologies, and responsibilities. A novice may require basic routines and task models; a highly experienced employee may need to revise familiar procedures; an individual moving into management must learn to coordinate rather than directly execute every action. The relevant distinction is the relation between current capability and new task requirements.
Age, tenure, digital familiarity, and disability status can interact with opportunity and instructional design. They should not be used as unquestioned substitutes for learning potential. A scientifically informative study examines prior experience, task-specific knowledge, access to practice, course usability, and actual later performance. Repeated measures can distinguish a learning trajectory from a cross-sectional difference between groups.
Reskilling programs should define how the new competence will be recognized in occupational assignments. Tool adoption without improved understanding, safe action, or effective judgment may be a workflow change rather than skill development. The strongest program makes instruction, task opportunities, revised responsibility, and competence verification part of a coherent transition.
Accessibility and Distribution of Learning Opportunity
Developmental opportunity is unequally distributed through schedules, supervisory sponsorship, job assignments, technology access, and occupational hierarchy. Training design can widen participation by supplying captions, readable interfaces, accessible controls, alternate formats, and reasonable performance accommodations when appropriate. These are questions of valid measurement as well as fair access.
A course that requires inaccessible visual cues may inadvertently test visual access rather than the target technical skill. A workplace assessment that assigns challenging cases only to favored employees may confound competence with opportunity. Equitable evaluation therefore records inclusion criteria, participant characteristics where appropriate, access to practice, dropout patterns, and performance under valid accommodations.
The goal is to examine whether training mechanisms apply across the actual workforce and whether particular institutional barriers suppress their effectiveness. Subgroup analyses require sufficient data, privacy protections, and cautious interpretation of small samples. Work-system design, training opportunity, and competence assessment must be considered together if measured differences are to be explained scientifically.
High Workload, Errors, and Human Factors
Complex work often involves interruptions, competing demands, incomplete information, fatigue, and pressure for speed. Training that succeeds only under calm instructional conditions may fail precisely where it matters. Human-factors analysis asks which portions of performance depend on knowledge and practice and which depend on interface design, staffing, workload, or coordination.
Safety-critical instruction should include escalation, stopping rules, recovery from common errors, and accurate understanding of consequential cues. Simulation can present rare and hazardous cases without requiring preventable real-world harm. A correct response under one scenario does not establish reliable performance across configurations. Task sampling and delayed reassessment improve the claim of transfer.
The organization must also distinguish user error from system failure. Training is sometimes prescribed in response to incidents caused by impossible workload, confusing displays, or missing resources. When the dominant causal factor lies in the working environment, instruction alone cannot eliminate it. Rigorous needs assessment and evaluation protect both employee competence and system safety.
Artificial Assistance and the New Unit of Training Analysis
Artificial systems can draft text, retrieve information, classify cases, provide procedural prompts, and suggest next actions during work. Their presence changes the observable unit of performance: a result may be produced by a worker's unaided skill, by an artificial system, or by their coordinated activity. Immediate output quality cannot therefore be assumed to represent durable development of human capability.
Noy, S., & Zhang, W (2023) randomized access to generative AI during professional writing tasks and found improved productivity and judged output under the experimental conditions. The result concerns tool-assisted task performance. In a field deployment among 5,172 customer-support agents, Brynjolfsson, E., Li, D., & Raymond, L (2025) reported a 15% average increase in issues resolved per hour with AI assistance, with heterogeneous effects across worker groups and evidence consistent with improved learning for some less-experienced employees.
Research must distinguish aided from unaided performance, measure delayed competence, and test responses to misleading or unavailable assistance. Artificial systems introduce trainable skills of verification, calibrated reliance, escalation, and human–artificial coordination. In the Artificial Era, such arrangements become part of occupational psychology while retaining clear distinctions among philosophical claims, operational hypotheses, and measured outcomes.
Artificial Tutors, Coaching, and Skill Continuity
An artificial tutor may generate examples, provide feedback, answer procedural questions, or adapt practice difficulty. These capabilities can lower barriers to training, but their effectiveness depends on reliable content, validated scoring criteria, and the appropriate withdrawal of support. A fluent explanation can be incorrect, and an apparently well-performing employee may be relying on machine reasoning that disappears when assistance is unavailable.
Comparative research should distinguish AI-guided practice, conventional instruction, human coaching, blended support, and suitable active controls. Outcomes should include delayed independent performance, novel task generalization, error detection, and the cost of correcting misleading feedback. In safety-critical domains, external validation and appropriate expert review are especially important.
Two trajectories require testing: augmentation, in which assistance expands and accelerates learning, and dependence, in which frequent assistance suppresses rehearsal of essential skills. Both may occur in different task components within the same job. Use frequency or employee satisfaction alone cannot establish either. The scientific objective is sustainable competence across changes in tools, organizational roles, and task environments.
Economic Outcomes and Organizational Value
Training consumes time, wages, instructional resources, equipment, and opportunities that could otherwise be used for work. Its value depends on the effects achieved relative to these costs and to credible alternatives. Benefits may appear in accuracy, safety, service, coordination, adaptability, or employee opportunities, not only in immediate throughput.
Aguinis, H., & Kraiger, K (2009) reviews training benefits at individual, team, organizational, and societal levels, making the attribution problem explicit. Productivity can improve after a course because of changed demand, technology, staffing, or incentives. A defensible return-on-investment calculation requires a credible estimate of the counterfactual, an accurate cost boundary, and an outcome meaningful to those affected.
Economic evaluation should include the time required for practice and posttraining reinforcement, the durability of benefits, their distribution among workers, and uncertainty. Standardized learning effect sizes cannot automatically be converted into monetary value. A program that prevents rare catastrophic mistakes may be important despite a modest average speed improvement; a short-term increase in throughput may be unhelpful if quality deteriorates.
Cultural and Institutional Conditions
Learning takes place within institutions that differ in governance, hierarchy, communication norms, language, labor protections, and technological access. A training intervention tested in a single organization may depend on supervisor availability, error-reporting norms, or workplace autonomy that is absent elsewhere. Generalizing across cultures requires investigation of these contextual mechanisms rather than fixed assumptions about people from named populations.
Instructional translation should preserve the meaning of performance criteria and assessment cases as well as vocabulary. A procedure that assumes an employee can question a senior colleague may be ineffective where organizational roles make that behavior costly. Conversely, a localized modification may preserve the underlying training mechanism through a different communication practice.
Cross-site research should report adaptations, organizational resources, workforce composition, language requirements, and measures of training fidelity. Replications are most informative when they explain where the mechanism remains stable and where institutional variation changes its operation. Transfer science includes transfer of the training intervention itself across organizations.
Replication, Reporting Bias, and Measurement Disputes
The field contains multiple operational definitions of learning and transfer, and this variation complicates comparison. Supervisor assessments, trainee self-reports, simulator scores, and objective task metrics may produce different conclusions about the same program. Ford, J. K., Baldwin, T. T., & Prasad, J (2018) and Blume, B. D., Ford, J. K., Baldwin, T. T., & Huang, J. L (2010) emphasize conceptual and methodological issues that remain central to training-transfer research.
Meta-analytic syntheses are limited by their included studies. Publication selection, inconsistent reporting, dependent outcomes, variation in follow-up, and unmeasured workplace conditions can change pooled effects. Replication requires clear descriptions of materials, task demands, comparison groups, scorer instructions, analytic decisions, and the environment in which performance occurred.
Studies should report negative or inconclusive outcomes, changes in opportunities to perform, subgroup uncertainty, and organizational changes during follow-up. Better evidence is produced when investigators measure distinct stages of the proposed causal chain instead of labeling every improvement as training transfer. Independent, multi-organization replications are especially useful for determining whether an instructional method works beyond its original occupational setting.
Unresolved Questions and a Research Agenda
Several difficult questions remain. How do complex skills develop across months when relevant tasks are infrequent? Which components of training—models, feedback, retrieval, error experience, and coaching—are necessary, and which become redundant? Can opportunity to perform be modified independently of motivation and instructional quality? How should transfer be measured for genuinely novel tasks when novelty itself is multidimensional?
Longitudinal experiments could randomize practice schedules and follow-up support, measure objective work opportunities, and assess both near and adaptive transfer. Multi-site trials could examine whether the same instructional mechanism survives different organizational incentives or interface constraints. Better evidence is also needed on equity in access to consequential assignments and to validated assessment.
Artificial assistance expands this research agenda. Investigators must establish whether tool-mediated output improves independent competence, when skill fades through disuse, and how workers adapt as systems are updated or removed. The strongest studies will measure both the human worker and the changing joint work system, while keeping training outcomes distinct from product adoption or short-term productivity.
Integrated Scientific Synthesis
The research literature supports a connected explanation of workplace skill development. Needs assessment defines the occupational problem. Instruction creates task-relevant representations and opportunities for practice. Retrieval, feedback, error correction, and varied examples affect the durability and adaptability of learning. Motivation and self-regulation influence how trainees use these experiences. Work assignment, managerial support, social expectations, and available tools influence whether newly acquired capability becomes real job behavior.
Baldwin, T. T., & Ford, J. K (1988) provides the organizing distinction among learner, training, and environment. Arthur, W., Jr., Bennett, W., Jr., Edens, P. S., & Bell, S (2003) shows that many organizational training interventions produce meaningful outcomes. Blume, B. D., Ford, J. K., Baldwin, T. T., & Huang, J. L (2010), Ford, J. K., Baldwin, T. T., & Prasad, J (2018), and Burke, L. A., & Hutchins, H. M (2007) identify the complexity of predicting transfer. Evidence from behavior modeling, error-management training, and team training shows that instructional mechanisms interact with task and evaluation conditions.
A scientifically adequate evaluation asks whether training caused learning, whether learning generalized to representative tasks, whether the effect endured, and whether workplace outcomes changed for the proposed reason. These questions can have different answers. Their separation makes research cumulative, prevents exaggerated claims, and provides an evidentiary basis for modifying both instruction and the work environment.
Conclusion
Workplace training is a method of developing occupational capability, and its scientific significance is demonstrated by durable performance beyond the instructional encounter. Across theoretical frameworks, controlled studies, original field research, and quantitative syntheses, learning is shaped by instructional design, the characteristics and goals of trainees, and organizational conditions for application. Short-term knowledge, independent skilled execution, adaptive transfer, maintenance, and organizational impact remain distinguishable outcomes.
The strongest research treats skill as an operationally defined capacity, investigates the causes of its development, and measures performance under meaningful variation and time delay. This framework also identifies why training can be insufficient when task allocation, staffing, technological systems, or incentives prevent skilled action.
The emerging Artificial Era adds a further distinction between augmented work performance and persistent competence. Human–artificial systems can change how knowledge and action are distributed, requiring new controlled studies of learning, assistance, verification, and skill continuity. The enduring task of workplace training science is to explain when an intervention establishes reliable capability and when the organization makes that capability effective.
References
- Aguinis, H., & Kraiger, K. (2009). Benefits of training and development for individuals and teams, organizations, and society. Annual Review of Psychology, 60, 451–474. https://doi.org/10.1146/annurev.psych.60.110707.163505
- Alliger, G. M., & Janak, E. A. (1989). Kirkpatrick's levels of training criteria: Thirty years later. Personnel Psychology, 42(2), 331–342. https://doi.org/10.1111/j.1744-6570.1989.tb00661.x
- Alliger, G. M., Tannenbaum, S. I., Bennett, W., Jr., Traver, H., & Shotland, A. (1997). A meta-analysis of the relations among training criteria. Personnel Psychology, 50(2), 341–358. https://doi.org/10.1111/j.1744-6570.1997.tb00911.x
- Arthur, W., Jr., Bennett, W., Jr., Edens, P. S., & Bell, S. T. (2003). Effectiveness of training in organizations: A meta-analysis of design and evaluation features. Journal of Applied Psychology, 88(2), 234–245. https://doi.org/10.1037/0021-9010.88.2.234
- Baldwin, T. T., & Ford, J. K. (1988). Transfer of training: A review and directions for future research. Personnel Psychology, 41(1), 63–105. https://doi.org/10.1111/j.1744-6570.1988.tb00632.x
- Bell, B. S., Tannenbaum, S. I., Ford, J. K., Noe, R. A., & Kraiger, K. (2017). 100 years of training and development research: What we know and where we should go. Journal of Applied Psychology, 102(3), 305–323. https://doi.org/10.1037/apl0000142
- Blume, B. D., Ford, J. K., Baldwin, T. T., & Huang, J. L. (2010). Transfer of training: A meta-analytic review. Journal of Management, 36(4), 1065–1105. https://doi.org/10.1177/0149206309352880
- Brynjolfsson, E., Li, D., & Raymond, L. (2025). Generative AI at work. The Quarterly Journal of Economics, 140(2), 889–942. https://doi.org/10.1093/qje/qjae044
- Burke, L. A., & Hutchins, H. M. (2007). Training transfer: An integrative literature review. Human Resource Development Review, 6(3), 263–296. https://doi.org/10.1177/1534484307303035
- Cepeda, N. J., Pashler, H., Vul, E., Wixted, J. T., & Rohrer, D. (2006). Distributed practice in verbal recall tasks: A review and quantitative synthesis. Psychological Bulletin, 132(3), 354–380. https://doi.org/10.1037/0033-2909.132.3.354
- Edmondson, A. (1999). Psychological safety and learning behavior in work teams. Administrative Science Quarterly, 44(2), 350–383. https://doi.org/10.2307/2666999
- Ford, J. K., Baldwin, T. T., & Prasad, J. (2018). Transfer of training: The known and the unknown. Annual Review of Organizational Psychology and Organizational Behavior, 5, 201–225. https://doi.org/10.1146/annurev-orgpsych-032117-104443
- Ford, J. K., Quiñones, M. A., Sego, D. J., & Sorra, J. S. (1992). Factors affecting the opportunity to perform trained tasks on the job. Personnel Psychology, 45(3), 511–527. https://doi.org/10.1111/j.1744-6570.1992.tb00858.x
- Gegenfurtner, A., Veermans, K., Festner, D., & Gruber, H. (2009). Motivation to transfer training: An integrative literature review. Human Resource Development Review, 8(3), 403–423. https://doi.org/10.1177/1534484309335970
- Grossman, R., & Salas, E. (2011). The transfer of training: What really matters. International Journal of Training and Development, 15(2), 103–120. https://doi.org/10.1111/j.1468-2419.2011.00373.x
- Holton, E. F., III, Bates, R. A., & Ruona, W. E. A. (2000). Development of a generalized learning transfer system inventory. Human Resource Development Quarterly, 11(4), 333–360. https://doi.org/10.1002/1532-1096(200024)11:4%3C333::AID-HRDQ2%3E3.0.CO;2-P
- Keith, N., & Frese, M. (2008). Effectiveness of error management training: A meta-analysis. Journal of Applied Psychology, 93(1), 59–69. https://doi.org/10.1037/0021-9010.93.1.59
- Kraiger, K., Ford, J. K., & Salas, E. (1993). Application of cognitive, skill-based, and affective theories of learning outcomes to new methods of training evaluation. Journal of Applied Psychology, 78(2), 311–328. https://doi.org/10.1037/0021-9010.78.2.311
- Locke, E. A., & Latham, G. P. (2002). Building a practically useful theory of goal setting and task motivation: A 35-year odyssey. American Psychologist, 57(9), 705–717. https://doi.org/10.1037/0003-066X.57.9.705
- Macnamara, B. N., Hambrick, D. Z., & Oswald, F. L. (2014). Deliberate practice and performance in music, games, sports, education, and professions: A meta-analysis. Psychological Science, 25(8), 1608–1618. https://doi.org/10.1177/0956797614535810
- Noe, R. A., Clarke, A. D. M., & Klein, H. J. (2014). Learning in the twenty-first-century workplace. Annual Review of Organizational Psychology and Organizational Behavior, 1, 245–275. https://doi.org/10.1146/annurev-orgpsych-031413-091321
- Noy, S., & Zhang, W. (2023). Experimental evidence on the productivity effects of generative artificial intelligence. Science, 381(6654), 187–192. https://doi.org/10.1126/science.adh2586
- Roediger, H. L., III, & Karpicke, J. D. (2006). Test-enhanced learning: Taking memory tests improves long-term retention. Psychological Science, 17(3), 249–255. https://doi.org/10.1111/j.1467-9280.2006.01693.x
- Saks, A. M., & Belcourt, M. (2006). An investigation of training activities and transfer of training in organizations. Human Resource Management, 45(4), 629–648. https://doi.org/10.1002/hrm.20135
- Salas, E., DiazGranados, D., Klein, C., Burke, C. S., Stagl, K. C., Goodwin, G. F., & Halpin, S. M. (2008). Does team training improve team performance? A meta-analysis. Human Factors, 50(6), 903–933. https://doi.org/10.1518/001872008X375009
- Salas, E., Tannenbaum, S. I., Kraiger, K., & Smith-Jentsch, K. A. (2012). The science of training and development in organizations: What matters in practice. Psychological Science in the Public Interest, 13(2), 74–101. https://doi.org/10.1177/1529100612436661
- Sitzmann, T., & Ely, K. (2011). A meta-analysis of self-regulated learning in work-related training and educational attainment: What we know and where we need to go. Psychological Bulletin, 137(3), 421–442. https://doi.org/10.1037/a0022777
- Sitzmann, T., Kraiger, K., Stewart, D., & Wisher, R. (2006). The comparative effectiveness of web-based and classroom instruction: A meta-analysis. Personnel Psychology, 59(3), 623–664. https://doi.org/10.1111/j.1744-6570.2006.00049.x
- Taylor, P. J., Russ-Eft, D. F., & Chan, D. W. L. (2005). A meta-analytic review of behavior modeling training. Journal of Applied Psychology, 90(4), 692–709. https://doi.org/10.1037/0021-9010.90.4.692