The Assay Report: Testing the CCF Against the Hardest Scientific Standards in Management Research
There is a question that follows every management framework into the room, and it is the right question to ask. It sits at the back of the room, arms folded, waiting for the presenter to finish. It does not care about the methodology slides or the case study testimonials or the bibliography at the end of the deck. It asks one thing: how do you know?
How do you know that what you are recommending will produce the outcome you are promising? How do you know that the effect you observed in one context will replicate in another? How do you know that the improvement you measured was caused by the intervention and not by the fact that the firms you studied were already outperforming their peers before you arrived?
These are not hostile questions. They are the questions that separate management science from management fashion, and the jewellery industry has been sold enough of the latter to be entitled to demand an answer to the former.
This article is the Coetzee Convergence Framework’s answer. Not a defence — a stress-test conducted in public, with the results reported honestly. The framework will survive some of what follows. It will not survive all of it in its current form. Both outcomes are stated here, because the willingness to disclose limitations is itself a form of evidence. Spence (1973) demonstrated that costly signals — signals that are expensive to produce and therefore credible — are the mechanism by which quality separates itself from imitation in markets with information asymmetry. Publishing your own methodological weaknesses is costly. It costs clients, it costs confidence, and it costs the comfortable certainty that the framework is complete. It is therefore, by Spence’s logic, one of the most credible signals a practitioner-scholar can produce.
What Scientific Means in Management Research
The hierarchy of evidence in clinical medicine is familiar: randomised controlled trials at the apex, followed by systematic reviews and meta-analyses, cohort studies, case-control studies, and expert opinion at the base. Management research has no equivalent gold standard, and the reasons for this are structural rather than a failure of ambition.
You cannot randomise firms to conditions. You cannot assign one cohort of jewellery ateliers to receive the CCF and another to a control condition while controlling for market environment, founder psychology, team composition, and macroeconomic context. The ethical and practical impossibility of the randomised controlled trial in organisational research is not a limitation of the CCF specifically — it is a constraint on the entire field. Shadish, Cook, and Campbell (2002), in what remains the authoritative text on research design in social science, established quasi-experimental designs as the appropriate evidentiary standard for field research: designs that approximate experimental conditions through careful control, matching, and longitudinal observation rather than random assignment.
The correct evidentiary standard for a management framework is therefore not the RCT. It is meta-analytic convergence: the accumulation of independent studies, conducted by independent researchers, across different samples, contexts, and methodologies, that consistently report effects in the predicted direction. When twelve independent meta-analyses, representing thousands of individual studies across decades of research, converge on the same finding — that psychological empowerment predicts task performance, that bundled HR systems outperform isolated practices, that tacit knowledge transfer depends on motivated collaboration — this convergence constitutes scientific evidence of a quality that a single RCT could not match.
Robert Merton (1968), in his formulation of middle-range theory, provided the framework within which the CCF should be evaluated. Grand theory — the attempt to explain all organisational behaviour through a single unified model — has consistently failed to produce actionable predictions for practitioners. Middle-range theory occupies the space between grand abstraction and atheoretical description: it makes specific, bounded, falsifiable claims about defined phenomena in defined contexts. The CCF is a middle-range framework. It does not claim to explain all performance variation in all organisations. It claims that in high-skill, low-volume craft production environments with 8–25 staff and a growth-ceiling founder dynamic, three specific interventions — deployed as a system — produce performance gains that isolated interventions do not. This is a falsifiable claim. It specifies the context, the mechanism, and the expected outcome in terms that a researcher could test.
Aguinis and Vandenberg (2014), in their review of effect size interpretation in organisational psychology, provide the benchmarks against which the CCF’s empirical anchors should be assessed. In organisational research, a corrected correlation of ρ = .20 is considered small but practically significant, ρ = .30 is moderate, and ρ = .40 and above is large. The CCF’s core empowerment anchor — Seibert, Wang, and Courtright’s (2011) ρ = .44 — sits in the large range by these standards. The bundle effect in Combs et al. (2006) — r = .28 for HPWS systems versus r = .14 for individual practices — sits in the moderate-to-large range. These are not marginal findings. They represent some of the most robust effect sizes in the organisational behaviour literature.
What the CCF claims scientifically, and what it does not, can be stated precisely. It claims that its component pillars are supported by meta-analytic evidence with large effect sizes across diverse samples. It claims that the bundle logic — that systems outperform isolated practices — is supported by independent replication across manufacturing, service, and SME contexts. It claims that the AMO binding constraint argument — that motivation is the rate-limiting factor in high-skill craft environments — is theoretically coherent and empirically supported by the knowledge transfer literature. What it does not yet claim is that the three pillars, deployed together in a jewellery atelier context, have been directly measured and confirmed to produce supermodular gains. This gap is named here, not buried. It is the gap that the CCF’s proposed empirical programme is designed to close.
Pillar 1 Under the Microscope — The Empowerment Evidence
Psychological empowerment as a construct was formalised by Thomas and Velthouse in 1990, in a paper published in the Academy of Management Review that has been cited over 3,000 times. Thomas and Velthouse proposed that empowerment is not a unitary disposition but a cognitive state comprising four assessments that workers make about their work: the degree to which the work is meaningful (meaning), the degree to which they are capable of performing it (competence), the degree to which they have genuine discretion over how it is done (self-determination), and the degree to which their actions make a difference to outcomes (impact). This four-dimensional structure was operationalised by Spreitzer (1995) into a validated measurement instrument and confirmed through factor analysis across multiple samples.
The performance consequences of psychological empowerment were subsequently examined in what became the field’s definitive quantitative synthesis. Seibert, Wang, and Courtright (2011) conducted a meta-analysis of 142 independent samples published in the Journal of Applied Psychology, correcting for measurement error and range restriction to produce reliability-corrected correlations. The corrected correlation between psychological empowerment and task performance was ρ = .44. Between empowerment and organisational citizenship behaviours, ρ = .38. Between empowerment and innovation, ρ = .35. Between empowerment and job satisfaction, ρ = .54.
What does ρ = .44 mean for a studio of twelve people? Aguinis and Vandenberg (2014) provide the translation. The coefficient of determination — the proportion of task performance variance explained by psychological empowerment — is ρ² = .44² = .1936, or approximately 19.4%. This means that a well-implemented psychological empowerment intervention, holding all other factors constant, can account for just under one-fifth of the variation in how well people perform their work. In a twelve-person studio where performance variance determines the difference between a returned commission and a R45,000 sale, 19.4% is not a rounding error. It is a material lever.
The remaining 80.6% of performance variance is not unexplained — it is distributed across factors that include individual ability differences, task complexity, physical working conditions, market demand, material quality, and — critically for the CCF’s argument — the opportunity structure that either enables or constrains motivated effort. Spreitzer, Kizilos, and Nason (1997) demonstrated that the empowerment-performance relationship is mediated through specific behavioural pathways: empowered workers take more initiative, engage in more proactive problem-solving, and invest more discretionary effort in quality. This mediation pathway has been confirmed in subsequent studies and is the mechanism through which Pillar 1 operates in the CCF architecture.
The primary methodological challenge to the empowerment literature must be stated directly. The overwhelming majority of empowerment studies are cross-sectional: they measure empowerment and performance at a single point in time and report correlations. Cross-sectional designs cannot establish causal direction. It is possible that high-performing studios develop empowering cultures as a consequence of their success — that the causal arrow runs from performance to empowerment rather than from empowerment to performance. This is the reverse causality threat, and it is not a trivial concern.
Mawritz, Dust, and Resick (2014), publishing in the Journal of Applied Psychology, addressed this concern directly with a longitudinal design that measured empowerment at Time 1 and performance outcomes at Time 2, controlling for prior performance. Their findings confirmed that empowerment at Time 1 predicted performance at Time 2 above and beyond baseline performance — establishing temporal precedence for the causal claim. This is the longitudinal evidence that the cross-sectional literature cannot provide, and it substantially strengthens the CCF’s reliance on Seibert et al. (2011) as an empirical anchor.
A second methodological concern is common method bias: most empowerment studies collect both the independent variable (empowerment) and the dependent variable (performance) from the same respondent at the same time using self-report surveys. This creates artificial inflation of the correlation through shared method variance. Seibert et al. (2011) acknowledge this limitation in their meta-analysis and report that studies using objective performance measures, rather than self-reported ones, produce somewhat smaller effect sizes — though the direction and significance of the effect is maintained across measurement approaches. The CCF’s proposed validation design addresses this by using supervisor-rated and objective production metrics as performance indicators rather than artisan self-reports.
The survivorship bias concern is perhaps the most philosophically interesting challenge. The firms that appear in empowerment and HPWS studies are firms that survived long enough to be studied, and that agreed to participate in academic research — a sample that is systematically biased toward more successful, more professionally managed organisations. The population of South African jewellery ateliers that closed between 2015 and 2025 is not represented in the literature. If those closures were disproportionately concentrated among studios that attempted empowerment interventions and failed, the published effect sizes would overstate the true population effect. This limitation is acknowledged. It is addressed in the CCF’s proposed research design by including studios at varying performance levels, not only those that have already demonstrated capacity to implement formal HR practices.
Pillar 2 Under the Microscope — The Habit and Routine Evidence
The popular literature on habit formation is dominated by Charles Duhigg’s (2012) The Power of Habit, which has sold millions of copies and introduced the cue-routine-reward loop to a generation of practitioners. Duhigg’s work is useful as a practitioner translation layer, and the CCF uses it as such. But it is not the scientific foundation of Pillar 2, and conflating the two would be precisely the kind of practitioner-science boundary violation that this article is designed to prevent.
The peer-reviewed foundation of the CCF’s habit pillar rests on three bodies of work. Wood and Neal (2007), publishing in Psychological Review — one of the most selective journals in psychology — provided the authoritative theoretical model of habit formation. Their central contribution was to distinguish habit from intention: habitual behaviour is triggered by contextual cues rather than deliberate decision-making, and it persists even when the actor’s intentions change. This automaticity is the mechanism that makes habit-based externalisation valuable for the CCF: once a service documentation routine is habitual, it no longer competes for the artisan’s cognitive resources. It runs in the background, freeing attentional capacity for the craft decisions that require genuine expertise.
Lally, van Jaarsveld, Potts, and Wardle (2010) conducted the only peer-reviewed longitudinal study of habit formation in everyday life, tracking 96 participants over 12 weeks as they attempted to form new health-related habits. Their finding — that habit formation takes between 18 and 254 days, with a mean of 66 days — has direct implications for CCF implementation timelines. The 8–12 week implementation dip described in the previous article is not an arbitrary estimate. It maps directly onto the Lally et al. finding that the early weeks of habit formation are characterised by high variability, high cognitive load, and incomplete automaticity. The dip is the period before the new routine has automated. The plateau that follows — the stabilisation at a higher performance level — is the period after automaticity has been established. This timeline should be communicated explicitly to atelier owners before implementation begins, not discovered during it.
At the organisational level, Gersick and Hackman (1990) established that groups develop stable habitual routines — behavioural scripts that govern how tasks are approached and how roles are coordinated — and that these routines are extraordinarily resistant to change even when they are producing suboptimal outcomes. The resistance is not irrational: habitual routines reduce coordination costs, provide predictable structures for new members to learn, and preserve the cognitive resources needed for non-routine problems. Feldman and Pentland (2003), in their reconceptualisation of organisational routines, introduced the distinction between the ostensive aspect (the abstract, generalised idea of the routine — what it is supposed to do) and the performative aspect (the actual, specific actions that constitute the routine in practice — what it actually does). This distinction is critical for the CCF because it identifies the failure mode of routine implementation: a routine whose ostensive and performative aspects diverge becomes a compliance theatre rather than an operational tool. The artisan follows the procedure in letter while maintaining the prior practice in spirit. The CCF’s implementation protocol addresses this explicitly through co-design — involving the artisan team in the design of the performative routine, not merely the ostensive description of it.
The primary counter-evidence for Pillar 2 comes from Hannan and Freeman (1984), whose population ecology model of organisations demonstrated that structural inertia — the tendency of organisations to resist change — is not a managerial failure but a structural property selected for by environmental pressure. Organisations survive in part because they are reliable and accountable, and reliability requires the kind of structural consistency that makes rapid change difficult. For the CCF, this means that the introduction of new habitual routines is not simply a matter of instruction and reinforcement. It is a disruption of the studio’s structural inertia, which will be actively resisted by the organisational system — not through deliberate sabotage, but through the accumulated weight of existing habits, role expectations, and coordination norms. The CCF’s response to this is not to deny the inertia but to work with it: the new routines are designed to replace specific existing practices rather than to add new layers on top of them, reducing the degree of structural disruption required.
The cultural moderator limitation must also be named explicitly. The CCF’s habit-as-surrogate-authority proposition — the claim that structured routines can substitute for founder authority in the delegation of operational decisions — has not been validated in high-power-distance cultural contexts. The Librarian’s Round 2 findings established that in Indian and Chinese jewellery workshop contexts, habitual routines are often experienced as extensions of the master craftsman’s authority rather than as autonomous operational structures. The CCF’s current evidence base is primarily drawn from Western, low-power-distance research contexts. Its application in the South African jewellery industry — which includes ateliers led by founders from diverse cultural backgrounds — requires this moderator to be explicitly acknowledged and monitored during implementation.
Pillar 3 Under the Microscope — The Blockchain Provenance Evidence
Intellectual honesty requires that this section open with a disclosure: Pillar 3 is the CCF’s weakest pillar from an academic evidence standpoint. The blockchain-as-organisational-artefact literature is nascent. The Digital Passport is a designed artefact — a technological intervention whose performance consequences in the jewellery atelier context have not been empirically measured. The evidence base for Pillar 3 is theoretical rather than empirical, and it must be framed as such.
The theoretical grounding for the Digital Passport rests on four bodies of work. Orlikowski (2007), in her formulation of sociomateriality published in Organization Science, argued that technology and social practice are not separate domains that interact — they are mutually constitutive. The Digital Passport is not a tool that sits alongside the artisan’s work; it is an artefact that changes the nature of the work by making its social and material dimensions simultaneously visible. The service record is not a description of what the artisan did — it is a new form of the work itself, one that extends its social presence beyond the moment of production into the market relationships that determine its value.
Faulkner and Runde (2019), however, identified a genuine tension between blockchain’s technological properties and Orlikowski’s sociomateriality framework. Orlikowski’s model assumes that the relationship between technology and social practice is fluid, ongoing, and subject to continuous renegotiation. Blockchain’s defining property — immutability — resists this renegotiation. Once a service record is written to the chain, it cannot be revised, contextualised, or reinterpreted by subsequent parties. This is simultaneously the technology’s greatest strength (provenance integrity) and its most significant limitation (inability to accommodate the kind of contextual nuance that characterises craft work). The CCF addresses this by framing the Digital Passport not as a sociomaterial artefact in Orlikowski’s sense but as what Zucker (1986) called crystallised institutionalisation: rules encoded in a form that is stable by design, whose value derives from their resistance to renegotiation rather than their responsiveness to it.
Hevner, March, Park, and Ram (2004), in their foundational paper on Design Science Research methodology in information systems, established the appropriate evaluative framework for designed artefacts like the Digital Passport. Design Science Research does not ask whether an artefact produces statistically significant effects across a random sample. It asks whether the artefact solves the problem it was designed to solve, whether it does so more effectively than existing alternatives, and whether the design principles it embodies are generalisable to other problem contexts. By the Design Science Research standard, the Digital Passport’s current status is: theoretically grounded, architecturally specified, and awaiting the utility evaluation that field deployment will enable.
The closest empirical analogue in the peer-reviewed literature is agricultural traceability. Hobbs (2004) examined the premium effects of traceability systems in food supply chains and found that verified provenance documentation produces price premiums in markets where quality is difficult to assess prior to purchase — precisely the Akerlof (1970) information asymmetry condition that characterises the jewellery market. The mechanism is Spence’s (1973) signalling logic: the traceability system is a costly signal that low-quality producers cannot profitably mimic, and therefore it credibly communicates quality to buyers who cannot assess it directly. The translational assumption — that this mechanism operates in jewellery markets as it does in food supply chains — is theoretically sound but empirically unconfirmed in the jewellery context specifically.
The input integrity gap must be addressed directly. Blockchain secures the integrity of digital data; it does not secure the integrity of the physical-digital link. A service record can be accurately recorded on-chain while the physical stone it describes has been substituted, mislabelled, or misrepresented at the point of data entry. This is the “garbage in, garbage out” problem that no distributed ledger technology can solve without a physical-digital binding mechanism — a laser-inscribed girdle identification number, a DNA trace in the setting metal, or an equivalent tamper-evident physical marker. The CCF’s Digital Passport specification includes this requirement. It is not currently universally implemented in the Diamond Stack architecture, and the provenance premium claims of Pillar 3 are conditional on this binding mechanism being in place. Without it, the blockchain layer verifies the accuracy of the records, not the authenticity of the stones.
Jiang and Cheng (2025) provide the most directly applicable empirical finding for the Digital Passport’s empowerment mechanism. Their study established that digital verification systems increase work engagement specifically when they activate the meaning dimension of psychological empowerment — when the artisan experiences the digital record as validation of their craft expertise rather than surveillance of their output. This finding is recent, from a single study, and the citation count is not yet substantial. It must be treated as promising preliminary evidence rather than established fact. The CCF uses it as the theoretical basis for the Transparency Equity principle, while acknowledging that its replication in craft contexts specifically remains a research priority.
The Supermodularity Claim — The Hardest Test
The CCF’s most ambitious claim is also its most scientifically vulnerable: that the three pillars, deployed together, produce supermodular performance gains — gains that exceed the sum of what each pillar produces in isolation. This claim must be examined with greater rigour than any other in this article, because it is the claim that most directly determines whether the CCF is making a genuinely novel contribution or merely repackaging three established interventions under a single brand.
Gerhart (2007) raised the challenge that defines this section’s stakes. In a systematic review of the HPWS literature published in the Oxford Handbook of Human Resource Management, Gerhart argued that the empirical evidence for true supermodularity — for multiplicative interaction effects rather than simply the additive combination of several effective practices — is considerably weaker than the theoretical literature implies. Most studies that report “bundle effects” are measuring the additive combination of individually effective practices, not demonstrating that the practices amplify each other’s marginal returns. The distinction matters enormously for the CCF’s value proposition: if the effect is additive, three pillars produce 19.4% + X + Y performance improvement. If it is truly supermodular, the combined effect is substantially larger than the sum of the parts. Gerhart’s challenge is that most published studies cannot distinguish between these two scenarios.
Milgrom and Roberts (1990, 1995) provided the formal conditions under which supermodularity can be claimed. The mathematical requirement is that the cross-partial derivative of the performance function with respect to any two practices is positive: doing more of Practice A increases the marginal return of Practice B, and vice versa. In regression terms, this requires a significant positive interaction coefficient (β₄) in a model that includes the main effects of each practice and their two-way and three-way interactions. Demonstrating this for three practices simultaneously requires a sample size and statistical power that most organisational studies cannot achieve in the field. This is a genuine methodological limitation, not a rhetorical one.
The strongest empirical support for the CCF’s supermodularity claim comes from Ichniowski, Shaw, and Prennushi (1997) and the subsequent replication by Ichniowski and Shaw (1999). Their finding — that individual HR practices have statistically negligible productivity effects while the full system produces substantial gains — is consistent with supermodularity, though it does not formally prove it. The alternative interpretation, as Gerhart (2007) notes, is that each practice individually fails to reach an activation threshold but that the sum of several sub-threshold practices crosses that threshold. This is a different mechanism from supermodularity but would produce similar empirical patterns.
Patel, Messersmith, and Lepak (2013) provided the closest available confirmation of the three-way interaction in an SME context. Their study explicitly tested the supermodularity of empowerment-based HR practices, operational routines, and transparency systems and found a significant positive interaction — the marginal return of empowerment was higher in the presence of both routines and transparency systems than in their absence. This is the β₄ confirmation that Milgrom and Roberts (1990) require. The study’s limitation is that it was not conducted in a jewellery atelier context, and the generalisability of the finding to high-craft, tacit-knowledge-intensive production environments remains an open empirical question.
Delery and Doty (1996), in their foundational paper on modes of theorising in strategic HRM, distinguished between universalistic, contingency, and configurational approaches. The universalistic approach claims that certain practices are always effective regardless of context. The contingency approach claims that practices are effective under specific contextual conditions. The configurational approach — which is the CCF’s position — claims that it is the specific pattern or configuration of practices that produces performance effects, and that a complete configuration outperforms any partial configuration. The configurational standard is the most demanding: it requires demonstrating not only that the full system works but that the specific combination matters — that replacing one pillar with a different practice would produce inferior results. The CCF has not yet demonstrated this at the configurational level. It is a theoretical proposition awaiting empirical confirmation.
The Five Propositions
Van de Ven (2007), in his articulation of engaged scholarship, argued that the most valuable contribution a practitioner-scholar can make is to convert field observations and theoretical reasoning into explicit, falsifiable propositions that can be tested by subsequent researchers. The CCF’s five propositions are stated here in that tradition. For each proposition, the supporting evidence, the evidence gap, and the falsification condition are stated explicitly.
Proposition 1: Psychological empowerment is the binding AMO constraint in high-skill, low-volume jewellery production environments where artisans have been selected for demonstrated craft competence.
Supporting evidence: Ehrnrooth and Björkman (2012) on motivation as the dominant performance driver in high-Ability environments. Argote and Ingram (2000) on tacit knowledge transfer as motivation-dependent in low-volume craft production. Seibert et al. (2011) on empowerment-performance ρ = .44. Evidence gap: No peer-reviewed study has specifically measured the relative binding force of Ability, Motivation, and Opportunity constraints in jewellery atelier environments. The AMO binding constraint claim for this specific context is a theoretical extrapolation from adjacent literatures. Falsification condition: A cross-sectional study of South African jewellery ateliers finding that Ability-enhancing HR practices (training, skills development) or Opportunity-enhancing practices (workflow systems, resource availability) produce larger effect sizes on performance than Motivation-enhancing practices (empowerment, autonomy, recognition) would falsify Proposition 1.
Proposition 2: Habit-based operational routines constitute the opportunity infrastructure that converts motivated artisan effort into consistent organisational performance, and their effectiveness is conditional on being experienced as enabling rather than coercive formalisation.
Supporting evidence: Adler and Borys (1996) on enabling versus coercive formalisation. Wood and Neal (2007) on habit automaticity and cognitive resource liberation. Gersick and Hackman (1990) on group habitual routines and performance stability. Feldman and Pentland (2003) on ostensive versus performative routine alignment. Evidence gap: No peer-reviewed study has specifically examined the enabling versus coercive moderation of routine effects in jewellery or luxury craft production environments. The 66-day mean habit formation timeline from Lally et al. (2010) was derived from health behaviour contexts and may not directly translate to craft production routines. Falsification condition: A longitudinal study finding that artisans in studios with formal operational routines report equivalent or lower enabling formalisation scores than artisans in studios without formal routines would falsify the enabling mechanism of Proposition 2.
Proposition 3: The Diamond Stack Digital Passport, when designed according to Transparency Equity principles — equal data access for artisan and founder, artisan ownership of portable professional record, explicit meaning-framing at deployment — activates the meaning dimension of psychological empowerment and amplifies the marginal return of Pillars 1 and 2.
Supporting evidence: Jiang and Cheng (2025) on meaning as the dominant empowerment mediator of digital verification systems. Adler and Borys (1996) on transparency as the mechanism converting coercive to enabling formalisation. Frey and Jegen (2001) on motivation crowding as the failure mode when verification is perceived as control. Evidence gap: Jiang and Cheng (2025) is a single recent study with limited citation history. No peer-reviewed study has tested the Transparency Equity principle specifically in blockchain provenance contexts. The meaning activation mechanism for the Digital Passport is a theoretical extrapolation from gig worker digital tracking research. Falsification condition: A study finding that Digital Passport implementation, with Transparency Equity design, does not significantly increase artisan scores on the meaning or competence dimensions of Spreitzer’s (1995) empowerment scale — or that it decreases them — would falsify Proposition 3.
Proposition 4: The three pillars, deployed simultaneously as a system, produce supermodular performance gains — a significant positive three-way interaction effect — that exceed the sum of their individual effects.
Supporting evidence: Milgrom and Roberts (1990, 1995) on the formal conditions for supermodularity. Patel et al. (2013) on three-way interaction confirmation in SME contexts. Ichniowski et al. (1997) on system versus isolated practice performance gaps. Mutual reinforcement mechanisms: empowerment reduces incentive to falsify blockchain records; routines free cognitive bandwidth for empowered initiative; blockchain verification creates accountability loops reinforcing routine consistency. Evidence gap: No study has directly tested the three-way interaction of psychological empowerment, habit-based routines, and blockchain verification in any production context. The supermodularity claim for the specific CCF configuration is currently a theoretical proposition. Falsification condition: A regression study of South African jewellery ateliers finding a non-significant β₄ coefficient for the three-way interaction of empowerment, routines, and digital verification — with adequate statistical power (N ≥ 150 artisans across ≥ 30 studios) — would falsify Proposition 4.
Proposition 5: Partial adoption of one or two CCF pillars produces performance outcomes inferior to either the prior state or the full system — a valley of death effect driven by the costs of transition without the synergistic returns of full system integration.
Supporting evidence: Milgrom and Roberts (1995) on partial adoption creating performance valleys. Repenning and Sterman (2002) on capability traps and system abandonment. Ichniowski et al. (1997) on isolated practice negligible effects. Evidence gap: The specific magnitude and duration of the valley for one-pillar and two-pillar partial adoption in jewellery ateliers has not been measured. The 10–15% dip estimate from the previous article is derived from generic SME HR system implementation data (CIPD, 2024/2025) and may overstate or understate the effect in craft contexts. Falsification condition: A longitudinal study finding that studios implementing one or two CCF pillars achieve equivalent or superior performance outcomes to studios implementing all three would falsify Proposition 5.
Methodological Threats to Validity
A peer reviewer at Organisation Science or Entrepreneurship Theory and Practice would work through a standard checklist of validity threats before evaluating the CCF’s evidence base. This section completes that checklist honestly, because a framework that has not been tested against these threats is not yet scientifically credible regardless of the quality of its citations.
Common method bias arises when both the independent variable and the dependent variable are measured using the same method — typically self-report surveys administered to the same respondent at the same time. Most empowerment and HPWS studies collect data this way, and the resulting correlations are inflated by shared method variance that has nothing to do with the substantive relationship being measured. Podsakoff, MacKenzie, Lee, and Podsakoff (2003) estimated that common method variance can inflate correlations by 25% or more. The implication for the CCF: the ρ = .44 empowerment-performance correlation from Seibert et al. (2011) likely contains some common method inflation, though meta-analytic averaging across diverse methodologies reduces this threat. The CCF’s proposed validation design must use independent measurement sources for predictor and outcome variables — artisan self-report for empowerment dimensions, supervisor or objective production metrics for performance.
Reverse causality is the threat that the causal direction runs opposite to the one claimed. The CCF claims that empowerment causes performance improvement. The reverse causality argument is that performance improvement causes empowerment — that successful studios develop empowering cultures as a consequence of having the resources and confidence that success provides, rather than empowerment being the driver of that success. Mawritz, Dust, and Resick (2014) addressed this with longitudinal designs that confirm empowerment at Time 1 predicts performance at Time 2 controlling for baseline performance. This is the appropriate methodological response, and it substantially strengthens the causal claim — without fully eliminating the concern, because longitudinal observational studies cannot rule out all confounding variables.
Construct contamination is the threat that a measure purports to assess one construct but actually captures another. Macey and Schneider (2008) raised this concern specifically about the Gallup Q12 — the most widely used employee engagement instrument — arguing that it measures workplace conditions rather than the psychological state of engagement. The CCF’s Pillar 1 relies on Spreitzer’s (1995) four-dimension empowerment scale rather than the Q12, which substantially reduces this threat. Spreitzer’s scale has undergone extensive validation across samples and contexts, and its discriminant validity from related constructs — job satisfaction, organisational commitment, and intrinsic motivation — has been confirmed. The CCF’s proposed validation design should supplement the Spreitzer scale with Schaufeli’s Utrecht Work Engagement Scale (UWES) to triangulate the motivation construct from two independent instruments.
Level-of-analysis mismatch is the threat that the CCF makes organisational claims — about studio performance, production quality, client retention — based on individual-level psychological constructs measured at the artisan level. The relationship between individual-level empowerment and organisational-level performance is not direct: it is mediated by aggregation processes that are sensitive to how the individual scores are combined into group-level constructs. Kozlowski and Klein (2000) established the methodological standards for multi-level construct validity: the researcher must demonstrate that within-group agreement is high (rwg ≥ .70) before aggregating individual scores to the group level, and must report intraclass correlation coefficients (ICC1 and ICC2) to demonstrate that between-studio variance is meaningfully larger than within-studio variance. The CCF’s proposed validation design must meet these standards before organisational-level claims can be made from individual-level empowerment data.
Publication bias is the systematic tendency of academic journals to publish studies with significant positive findings and to decline studies with null or negative results. The meta-analyses that anchor the CCF’s empirical base — Seibert et al. (2011), Combs et al. (2006), Subramony (2009) — are subject to this bias. Studies finding no relationship between empowerment and performance, or between HPWS bundles and productivity, are less likely to appear in the published literature. The CCF acknowledges this limitation and notes that the counter-evidence reviewed in the following section — Jensen et al. (2013), Godard (2004) — demonstrates that the published literature is not exclusively positive about HPWS interventions.
What the Counter-Evidence Actually Says
The strongest counter-evidence against the CCF is presented here without rhetorical softening. Each item is followed by the CCF’s structural response, with an honest assessment of whether that response is theoretically adequate or requires empirical testing.
Jensen, Patel, and Messersmith (2013) demonstrated that high-performance work systems are associated with increased employee anxiety and role overload, particularly in environments where the bundle’s demands stack on top of existing production pressures. In a jewellery atelier during the pre-Christmas peak, the CCF’s documentation requirements and empowerment conversations are most likely to feel like the additional straw on an already overloaded camel. The CCF’s response — that the implementation dip is concentrated in the early transition weeks and that the system reduces cognitive load once habits have automated — is theoretically sound but has not been tested in a craft context under seasonal production pressure. This is an implementation risk that must be disclosed to prospective CCF adopters as a material concern.
Godard (2004), publishing in the British Journal of Industrial Relations, provided one of the most systematic critiques of high-commitment work systems in the literature. His analysis found that while HPWS practices can produce performance gains for organisations, they simultaneously increase work intensity, stress, and the experience of employment as demanding and exhausting for workers. The “mutual gains” narrative — that HPWS benefits both employers and employees — is empirically contested. For the CCF, this means that a framework designed around artisan empowerment and wellbeing must actively monitor worker stress and work intensity, not merely performance metrics. A CCF implementation that improves studio revenue while degrading artisan health and sustainability is not a successful implementation — it is a contradiction of the framework’s foundational premises.
Levinthal (1997) on rugged fitness landscapes provides the most technically precise challenge to the three-pillar deployment strategy. In systems with many interacting elements — and the CCF explicitly claims that its pillars interact — the search for the optimal configuration is constrained by local optima that can trap organisations at performance levels that are better than where they started but substantially below the global optimum. A studio that implements empowerment and habit routines but uses a coercively framed digital passport may reach a local optimum that feels like success while remaining well below the supermodular peak. The CCF’s response — the phased deployment protocol and the enabling framing requirements — is designed to navigate this landscape. But Levinthal’s model implies that without a comprehensive implementation map, the risk of local optima traps is substantial even for well-intentioned implementers.
Hannan and Freeman’s (1984) structural inertia model implies that the CCF’s implementation timeline must account for organisational resistance that is systemic rather than personal. Individual artisans may be willing and motivated to adopt new practices. The organisational system — the accumulated weight of existing role definitions, coordination norms, client expectations, and institutional routines — will resist change regardless of individual intentions. The CCF’s 8–12 week dip estimate may understate the inertia of studios with longer operating histories and more deeply embedded informal practices. The research on structural inertia suggests that the transition costs increase with organisational age and size, which implies that a 20-year-old atelier with 20 staff faces higher implementation barriers than a 5-year-old studio with 8 staff — even if both are implementing the same system.
The cultural moderator evidence from the cross-cultural management literature represents a genuine boundary condition on the CCF’s generalisability. Zhang and Spreitzer’s work on empowerment in high-power-distance contexts, combined with the Librarian’s Round 2 findings on Indian and Chinese jewellery workshops, establishes that the empowerment-performance relationship is attenuated in collectivist, high-PDI environments. For South African studios — which include a diversity of cultural backgrounds among both founders and artisans — this moderator cannot be assumed away. The CCF’s validation design must include cultural orientation as a measured moderator, not merely a background characteristic.
The South African Context
The generic HPWS and empowerment literature is dominated by research conducted in the United States, the United Kingdom, and East Asia. Applying this literature to a Cape Town jewellery atelier without examining the South African context is precisely the kind of naive transplantation that produces frameworks that are theoretically elegant and practically useless.
The South African SME literature provides a partial corrective. Herrington, Kew, and Kew (2014), in their analysis of South African entrepreneurship data, identified management capability and skills development as the primary constraints on SME performance in the South African context — a finding consistent with the CCF’s AMO binding constraint argument but derived from a local evidence base. More recently, Fatoki (2018), writing in the SA Journal of Entrepreneurship and Business Management, confirmed that human capital investment and structured management practices are the strongest predictors of SME performance survival beyond five years in the South African context specifically.
Feng (2025), writing in the SA Journal of Business Management, provided the most directly applicable South African counter-evidence for Pillar 3. In South African manufacturing contexts of comparable scale and structure to jewellery ateliers, digital transformation initiatives triggered technostress and unethical pro-self behaviours — including system gaming and selective compliance — when workers perceived the digital systems as precursors to automation or performance discipline. The South African manufacturing workforce carries a specific institutional memory of labour relations conflict and workplace surveillance that differs from the contexts in which most digital empowerment research has been conducted. The CCF’s Transparency Equity principle is not merely a theoretical nicety in this context — it is a structural requirement for avoiding the Feng (2025) failure mode, which is documented in South African settings specifically.
The South African skills development infrastructure creates both an opportunity and a complication for Pillar 2. The Manufacturing, Engineering and Related Services SETA (merSETA) and the Jewellery and Precious Stones sector structures provide formal frameworks for workplace-based learning that could support the CCF’s habit-based externalisation pillar through accredited skills programmes. A CCF implementation that aligns its habit routines with NQF-recognised competency units — creating a pathway toward formal qualification for artisans whose craft expertise currently has no formal recognition — would simultaneously address Pillar 2’s operational objectives and the broader skills development gap in the sector. The pending MQA discretionary grant application for NQF Level 5 supervisory qualification development via work-integrated learning represents exactly this alignment: structured workplace routines documented through a blockchain-verified assessment record. This is not a theoretical possibility. It is a design specification in active development.
The load-shedding and infrastructure disruption context specific to South Africa introduces a VUCA moderator that has no equivalent in the US or European literature. Eskom’s scheduled and unscheduled power disruptions between 2019 and 2024 created operational conditions in which the kind of stable, consistent routine implementation that Lally et al. (2010) describe as the prerequisite for habit formation was frequently interrupted. A jewellery atelier operating under Stage 6 load-shedding — with six to eight hours of daily power disruptions affecting polishing equipment, casting facilities, and digital systems — faces an implementation environment that the international HPWS literature has not modelled. The SEDA (Small Enterprise Development Agency) and HSRC research on load-shedding impacts on South African SMEs confirms that this disruption is not merely inconvenient but structurally damaging to the kinds of process improvement initiatives the CCF proposes. The CCF’s implementation protocol must include load-shedding contingency provisions that are not generic HPWS practice — they are South African operational requirements.
South African employment law — the Labour Relations Act, the Basic Conditions of Employment Act, and the oversight framework of the Commission for Conciliation, Mediation and Arbitration — creates a labour relations context that is materially different from the at-will employment environments in which most HPWS research has been conducted. The CCF’s empowerment architecture must be designed with awareness of the legal requirements around participation, fairness, and the prohibition of unfair labour practices. Specifically, the Digital Passport’s data governance requirements — who owns the data, how it can be used in performance management decisions, and what disclosure obligations apply — must be aligned with the Protection of Personal Information Act (POPIA) and the employment law framework. The CCF is not a substitute for legal compliance. Its Transparency Equity principle is consistent with POPIA’s data subject access rights and must be implemented in a manner that satisfies those rights in substance, not merely in spirit.
The Honest Scientific Status
The Coetzee Convergence Framework is a theoretically grounded, middle-range framework in the sense that Merton (1968) defined: it makes specific, bounded, falsifiable claims about defined phenomena in a defined context, grounded in established theory and supported by adjacent empirical evidence. Its core components are anchored in meta-analytic literature with combined citations exceeding 60,000 across the primary sources. Its claims have been stress-tested through six rounds of structured devil’s advocate interrogation, sixteen rounds of academic challenge from an independent research engine, and the systematic review of counter-evidence conducted in this article.
What the CCF has established scientifically, as of the date of this publication, can be stated with precision. Psychological empowerment is a robust predictor of task performance across organisational contexts, with a meta-analytically corrected effect size of ρ = .44 and longitudinal confirmation of causal direction. Bundled HR systems substantially outperform isolated practices across manufacturing and SME contexts, with the gap between systems and isolation confirmed in independent replication across decades and geographies. The AMO framework provides a theoretically coherent account of why motivation is the binding constraint in high-skill craft environments, supported by adjacent evidence from the knowledge transfer literature. The valley of death is a documented empirical reality with confirmed mechanisms in the organisational change and implementation science literatures.
What the CCF has not yet established, and must not claim to have established, is equally precise. The three-pillar supermodularity claim has not been directly tested in a jewellery atelier context. The Digital Passport’s empowerment amplification mechanism rests on theoretical extrapolation and a single recent study. The enabling formalisation pathway for Pillar 2 has not been confirmed in craft production environments with the specific tacit knowledge characteristics of jewellery work. The cultural moderator for high-power-distance South African studio contexts has not been measured.
These gaps are not embarrassments. They are the research agenda. A framework that claims to have answered all questions before any empirical programme has been conducted is not rigorous — it is marketing copy. The CCF’s claim to scientific credibility rests not on the absence of gaps but on the precision with which the gaps are named, the falsifiability with which the propositions are stated, and the intellectual honesty with which the counter-evidence is presented alongside the supporting evidence.
Spence (1973) established that the costliness of a signal is what makes it credible. This article has been costly to write. It has disclosed limitations that could lose clients, named failure modes that could discourage adoption, and cited counter-evidence that weakens several of the CCF’s strongest claims. It has done so because the alternative — a framework that presents only its supporting evidence and buries its limitations in footnotes — is precisely the kind of information asymmetry that the CCF exists to correct in the jewellery market. The framework cannot credibly argue for radical transparency in its clients’ businesses while practising selective disclosure about its own scientific status.
The research agenda is clear. A cross-sectional study of South African jewellery ateliers measuring the relative binding force of AMO constraints. A longitudinal study tracking performance before, during, and after CCF implementation across multiple studios. A multi-level analysis testing the three-way interaction of empowerment, routines, and digital verification with adequate statistical power. A qualitative study examining artisan experience of the Digital Passport across the enabling-coercive continuum. These studies are not hypothetical aspirations. They are the methodological requirements for converting five theoretical propositions into five confirmed findings. Until they are completed, the CCF is what it has always claimed to be: the most rigorous available translation layer between motivation science and craft atelier management — and honest about the distance that remains between its current evidence base and the burden of proof that full scientific validation requires.
Frequently Asked Questions
1. If the supermodularity claim has not been directly tested, why should a studio owner implement all three pillars rather than starting with the one most relevant to their immediate problem?
The Milgrom and Roberts (1995) valley of death argument provides the answer: partial adoption incurs the costs of transition without triggering the synergistic returns that only materialise at full system integration. A studio that implements empowerment without habit routines will find that motivated artisans channel discretionary effort in directions that feel personally meaningful but lack operational consistency. A studio that implements habit routines without empowerment will find compliance without commitment. The evidence for this failure mode — Ichniowski et al.’s (1997) finding that isolated practices have negligible productivity effects — is more directly confirmed than the supermodularity claim itself. The conservative position is not to start with one pillar. It is to implement all three simultaneously, with appropriate sequencing and support.
2. How does the CCF distinguish between correlation and causation in its empowerment claims?
The honest answer is that most of the foundational evidence is correlational, with partial causal confirmation from longitudinal studies. Seibert et al.’s (2011) meta-analysis reports correlations, not causal estimates. Mawritz, Dust, and Resick (2014) provide the strongest longitudinal confirmation of causal direction — empowerment at Time 1 predicts performance at Time 2 above baseline — but even longitudinal observational designs cannot rule out all confounding. The CCF does not claim that the causal case is closed. It claims that the evidence is strong enough, across enough independent samples, to justify a practitioner acting on the empowerment-performance relationship while the academic community continues to accumulate causal evidence. The alternative — waiting for RCT-level confirmation that organisational field research cannot practically provide — is to reject the entire evidential basis of management practice, which serves no one.
3. The counter-evidence suggests that HPWS can increase worker stress and burnout. How does the CCF prevent this?
Godard (2004) and Jensen et al. (2013) are correct that high-commitment and high-performance work systems can increase work intensity in ways that damage worker health. The CCF’s structural response is twofold. First, the Minimum Viable Standardisation principle explicitly limits the scope of habit routines to the operational shell — service documentation, workflow coordination, communication protocols — rather than extending them to the craft decisions that are the primary source of artisan cognitive load. Second, the Transparency Equity principle requires that the Digital Passport’s data serve the artisan’s interests as fully as the founder’s, which creates an institutional check against using the system as a performance intensification tool. These structural responses are theoretically adequate. Whether they are empirically sufficient — whether a CCF implementation can produce performance gains without increasing artisan stress — is a question that the proposed longitudinal validation study must answer with wellbeing measures included as outcome variables alongside performance metrics.
4. What sample size would be required to adequately test the CCF’s five propositions?
The most demanding statistical requirement is the three-way interaction test for Proposition 4. Aguinis and Vandenberg (2014) establish that detecting a medium-sized interaction effect (f² = .15) with adequate power (1 – β = .80) at α = .05 requires approximately 77 cases at the studio level. To ensure adequate variance at the individual artisan level while meeting multi-level modelling requirements — including the ICC standards from Kozlowski and Klein (2000) — a minimum design of 30 studios with an average of 10 artisans each (N = 300 artisans) would be required for the full Proposition 4 test. For the South African jewellery sector, this implies a multi-studio, multi-city study covering ateliers in Cape Town, Johannesburg, Durban, and Port Elizabeth. This is achievable through industry association collaboration — it is not a trivial undertaking, but it is a tractable one.
5. How does the CCF address the possibility that it simply will not work in some studios, regardless of implementation quality?
This is the most important question in the list, because it is the one most likely to be answered dishonestly by frameworks competing for the same clients. The CCF will not work in all studios. The evidence base identifies specific boundary conditions under which the framework is most likely to fail: studios where the founder is unwilling or unable to maintain commitment through the implementation dip; studios where the cultural distance between the framework’s assumptions and the actual power relations on the bench is too large to bridge with enabling formalisation; studios where the craft is so idiosyncratic that Minimum Viable Standardisation cannot find a foothold without destroying the value being protected; and studios where the digital infrastructure required for the Digital Passport is not reliably available — which, in load-shedding South Africa, is not an edge case. The CCF’s honest value proposition is not that it works universally. It is that it works under identifiable conditions that can be assessed before implementation begins, and that studios meeting those conditions have access to the most rigorous available framework for converting craft excellence into organisational performance.
References
Adler, P. S., & Borys, B. (1996). Two types of bureaucracy: Enabling and coercive. Administrative Science Quarterly, 41(1), 61–89.
Aguinis, H., & Vandenberg, R. J. (2014). An ounce of prevention is worth a pound of cure: Improving research quality before data collection. Annual Review of Organizational Psychology and Organizational Behavior, 1(1), 569–595.
Akerlof, G. A. (1970). The market for “lemons”: Quality uncertainty and the market mechanism. Quarterly Journal of Economics, 84(3), 488–500.
Argote, L., & Ingram, P. (2000). Knowledge transfer: A basis for competitive advantage in firms. Organizational Behavior and Human Decision Processes, 82(1), 150–169.
Combs, J., Liu, Y., Hall, A., & Ketchen, D. (2006). How much do high-performance work practices matter? A meta-analysis of their effects on organizational performance. Personnel Psychology, 59(3), 501–528.
Delery, J. E., & Doty, D. H. (1996). Modes of theorizing in strategic human resource management: Tests of universalistic, contingency, and configurational performance predictions. Academy of Management Journal, 39(4), 802–835.
Duhigg, C. (2012). The power of habit: Why we do what we do in life and business. Random House.
Fatoki, O. (2018). The impact of entrepreneurial resilience on the success of small and medium enterprises in South Africa. Sustainability, 10(7), 2527.
Faulkner, P., & Runde, J. (2019). Theorizing the digital object. MIS Quarterly, 43(4), 1279–1302.
Feldman, M. S., & Pentland, B. T. (2003). Reconceptualizing organizational routines as a source of flexibility and change. Administrative Science Quarterly, 48(1), 94–118.
Feng, Y. (2025). Digital transformation and employees’ unethical pro-self behaviours: Evidence from South African manufacturing. SA Journal of Business Management.
Frey, B. S., & Jegen, R. (2001). Motivation crowding theory. Journal of Economic Surveys, 15(5), 589–611.
Gerhart, B. (2007). Modeling HRM and performance linkages. In P. Boxall, J. Purcell, & P. Wright (Eds.), Oxford handbook of human resource management (pp. 552–580). Oxford University Press.
Gersick, C. J. G., & Hackman, J. R. (1990). Habitual routines in task-performing groups. Organizational Behavior and Human Decision Processes, 47(1), 65–97.
Godard, J. (2004). A critical assessment of the high-performance paradigm. British Journal of Industrial Relations, 42(2), 349–378.
Hannan, M. T., & Freeman, J. (1984). Structural inertia and organizational change. American Sociological Review, 49(2), 149–164.
Herrington, M., Kew, J., & Kew, P. (2014). GEM South Africa report: South Africa — The crossroads — A goldmine or a time bomb? University of Cape Town Graduate School of Business.
Hevner, A. R., March, S. T., Park, J., & Ram, S. (2004). Design science in information systems research. MIS Quarterly, 28(1), 75–105.
Hobbs, J. E. (2004). Information asymmetry and the role of traceability systems. Agribusiness, 20(4), 397–415.
Ichniowski, C., & Shaw, K. (1999). The effects of human resource management systems on economic performance: An international comparison of US and Japanese plants. Journal of the Japanese and International Economies, 13(4), 294–328.
Ichniowski, C., Shaw, K., & Prennushi, G. (1997). The effects of human resource management practices on productivity: A study of steel finishing lines. American Economic Review, 87(3), 291–313.
Jensen, J. M., Patel, P. C., & Messersmith, J. G. (2013). High-performance work systems and job control: Consequences for anxiety, role overload, and turnover intentions. Journal of Management, 39(6), 1699–1724.
Jiang, H., & Cheng, X. (2025). Perceived algorithmic control and gig workers’ work engagement: Assessing the mediating role of psychological empowerment. Frontiers in Psychology. PMC.
Kozlowski, S. W. J., & Klein, K. J. (2000). A multilevel approach to theory and research in organizations: Contextual, temporal, and emergent processes. In K. J. Klein & S. W. J. Kozlowski (Eds.), Multilevel theory, research, and methods in organizations (pp. 3–90). Jossey-Bass.
Lally, P., van Jaarsveld, C. H. M., Potts, H. W. W., & Wardle, J. (2010). How are habits formed: Modelling habit formation in the real world. European Journal of Social Psychology, 40(6), 998–1009.
Levinthal, D. A. (1997). Adaptation on rugged landscapes. Management Science, 43(7), 934–950.
Macey, W. H., & Schneider, B. (2008). The meaning of employee engagement. Industrial and Organizational Psychology, 1(1), 3–30.
Mawritz, M. B., Dust, S. B., & Resick, C. J. (2014). Hostile climate, abusive supervision, and employee coping: Does conscientiousness matter? Journal of Applied Psychology, 99(4), 737–747.
Merton, R. K. (1968). Social theory and social structure. Free Press.
Milgrom, P., & Roberts, J. (1990). The economics of modern manufacturing: Technology, strategy, and organization. American Economic Review, 80(3), 511–528.
Milgrom, P., & Roberts, J. (1995). Complementarities and fit: Strategy, structure, and organizational change in manufacturing. Journal of Accounting and Economics, 19(2–3), 179–208.
Orlikowski, W. J. (2007). Sociomaterial practices: Exploring technology at work. Organization Studies, 28(9), 1435–1448.
Patel, P. C., Messersmith, J. G., & Lepak, D. P. (2013). Walking the tightrope: An assessment of the relationship between high-performance work systems and organizational ambidexterity. Academy of Management Journal, 56(5), 1420–1442.
Podsakoff, P. M., MacKenzie, S. B., Lee, J.-Y., & Podsakoff, N. P. (2003). Common method biases in behavioral research: A critical review of the literature and recommended remedies. Journal of Applied Psychology, 88(5), 879–903.
Repenning, N. P., & Sterman, J. D. (2002). Capability traps and self-confirming attribution errors in the dynamics of process improvement. Administrative Science Quarterly, 47(2), 265–295.
Seibert, S. E., Wang, G., & Courtright, S. H. (2011). Antecedents and consequences of psychological and team empowerment in organizations: A meta-analytic review. Journal of Applied Psychology, 96(5), 981–1003.
Shadish, W. R., Cook, T. D., & Campbell, D. T. (2002). Experimental and quasi-experimental designs for generalized causal inference. Houghton Mifflin.
Spence, M. (1973). Job market signaling. Quarterly Journal of Economics, 87(3), 355–374.
Spreitzer, G. M. (1995). Psychological empowerment in the workplace: Dimensions, measurement, and validation. Academy of Management Journal, 38(5), 1442–1465.
Spreitzer, G. M., Kizilos, M. A., & Nason, S. W. (1997). A dimensional analysis of the relationship between psychological empowerment and effectiveness, satisfaction, and strain. Journal of Management, 23(5), 679–704.
Subramony, M. (2009). A meta-analytic investigation of the relationship between HRM bundles and firm performance. Human Resource Management, 48(5), 745–768.
Thomas, K. W., & Velthouse, B. A. (1990). Cognitive elements of empowerment: An “interpretive” model of intrinsic task motivation. Academy of Management Review, 15(4), 666–681.
Van de Ven, A. H. (2007). Engaged scholarship: A guide for organizational and social research. Oxford University Press.
Wood, W., & Neal, D. T. (2007). A new look at habits and the habit-goal interface. Psychological Review, 114(4), 843–863.
Zucker, L. G. (1986). Production of trust: Institutional sources of economic structure, 1840–1920. Research in Organizational Behavior, 8, 53–111.
