Workplace learning in civil-military crisis exercises: constructive alignment, context support and defensive routines

Civil-military crisis exercises are not only instruments for emergency preparedness. They can also be understood as temporary workplace learning environments in which professionals from different organizations, mandates and knowledge traditions participate in situated work activities. Workplace learning research emphasizes that learning depends on the opportunities and support afforded by the workplace as well as on how individuals engage with those opportunities (Billett, 2001, 2004; Tynjälä, 2008). In civil-military settings, this means that participants’ opportunities to negotiate roles, test procedures, use legal and operational knowledge and reflect on coordination are shaped by both exercise design and the interorganizational context.

The need to understand these exercises as workplace learning environments has become more acute in Sweden. Geopolitical volatility, hybrid threats and cross-sectoral crises have increased the importance of collaboration between civilian authorities, municipalities, regions, private actors and military organizations. Sweden’s accession to NATO in 2024 has further intensified expectations that civil and military actors should be able to coordinate preparedness, evacuation, reception, accommodation and other crisis tasks across organizational boundaries (NATO, 2024; Government Offices of Sweden, 2024).

This national context matters for workplace learning because civil-military preparedness is not a stable technical routine but a changing professional practice. Sweden’s accession to NATO in 2024 and the renewed emphasis on total defense have altered expectations placed on civilian authorities, military units and regional actors (NATO, 2024; Government Offices of Sweden, 2024). Actors who previously trained within mainly national preparedness arrangements are increasingly expected to understand alliance coordination, wartime logistics, evacuation planning and the legal boundaries between civilian and military responsibilities. These shifts make exercises important not only as tests of readiness but as situated learning arenas in which participants interpret new roles, rehearse coordination across mandates and translate policy expectations into workable practice.

Exercises are widely used to develop these capabilities. They provide structured opportunities to rehearse decision-making, practice coordination and test interorganizational problem-solving under uncertainty (Boin and ‘t Hart, 2010; Comfort et al., 2010; Steigenberger, 2016). However, the mere performance of an exercise does not guarantee learning. Prior research has shown that collaboration exercises may become ritualized, repetitive and only weakly connected to real capability development when objectives are unclear, assessment is symbolic or critical feedback is avoided (Berlin and Carlström, 2008, 2015; Borell and Eriksson, 2013; Perry, 2004).

This problem is directly relevant to workplace learning because it concerns how learning opportunities are designed, afforded and taken up in a temporary work setting. Civil-military exercises involve multiple organizations, uneven experience, different chains of command and uncertain mandates; research on civil-military crisis management also documents the boundary and role tensions that can accompany such collaboration (Kalkman and Groenewegen, 2019). These characteristics make it unsafe to treat participation alone as evidence of learning. They instead call for attention to the learning affordances available to participants, their opportunities to ask questions and receive guidance and the conditions under which they can challenge assumptions (Billett, 2004; Edmondson, 1999, 2003).

This article, therefore, reframes civil-military crisis exercises as temporary, high-stakes workplace learning interventions. The analysis draws on constructive alignment (Biggs, 1996, 2012, 2014; Biggs and Tang, 2011), workplace learning research on affordances and participation (Billett, 2001, 2004; Tynjälä, 2008), Edmondson’s (1999, 2003) work on psychological safety and team learning and Argyris’s (1999) theory of defensive routines. Training-transfer research provides an additional boundary condition: even when learning processes are visible during an exercise, transfer requires generalization and maintenance in the work environment and cannot be assumed from exercise performance alone (Baldwin and Ford, 1988; Grossman and Salas, 2011).

From a workplace learning perspective, the quality of an exercise cannot be judged solely by scenario realism or the number of participating organizations. The relevant analytic question is whether the exercise affords opportunities to examine assumptions, practice unfamiliar tasks, obtain usable feedback and develop more elaborate joint understandings. Transfer to ordinary organizational practice is a separate question that requires evidence beyond the exercise itself (Baldwin and Ford, 1988; Grossman and Salas, 2011). The aim of the study is therefore to examine how constructive alignment, context support and defensive routines are associated with observable indicators of workplace learning in civil-military crisis exercises. The research question is: how are constructive alignment, context support and defensive routines associated with observable indicators of workplace learning in civil-military crisis exercises?

The article provides a comparative reanalysis of two exercise settings that have previously been reported separately [Hedlund and Alvinius, 2024, 2025a, 2025b]. It does not introduce new empirical data. The added value lies in applying a common workplace-learning framework across the two corpora and explicitly relating the comparison to workplace affordances, psychological safety, defensive routines and training transfer. The theoretical contribution is thus not a claim that the component concepts are new; it is the specification of a process model for temporary interorganizational learning settings and of the evidentiary boundary between observable learning-process indicators and demonstrated learning or transfer.

Constructive alignment was developed to explain how intended learning outcomes, learning activities and assessment can be designed to reinforce one another (Biggs, 1996, 2012, 2014; Biggs and Tang, 2011). In a workplace-learning application, alignment is treated here as a design property of the temporary learning environment rather than as proof that learning occurred. This distinction is important because workplace learning depends on both the opportunities afforded by the setting and participants’ engagement with them (Billett, 2001, 2004), while transfer beyond the training setting additionally depends on trainee, design and work-environment conditions (Baldwin and Ford, 1988; Grossman and Salas, 2011).

In civil-military crisis exercises, intended learning outcomes can specify the collaborative competencies participants are expected to develop. These may include the ability to establish a shared situational picture, identify legal mandates, coordinate evacuation tasks, clarify responsibilities across administrative levels or translate strategic decisions into operational plans. Teaching and learning activities are practical exercises that support these outcomes, such as scenario-based group work, tabletop simulations, decision-making drills and facilitated reflection. Assessment includes debriefings, feedback, self-assessment, scenario outputs and other mechanisms that allow participants to determine whether learning has occurred.

When these components are aligned, an exercise offers a coherent opportunity for intentional workplace learning: participants can connect expected capabilities to tasks and receive feedback on their work. When alignment is weak, the exercise may still be active and realistic while leaving the intended learning obscure. In this study, constructive alignment is therefore used diagnostically to examine the coherence of learning opportunities and feedback, not as an independent measure of learning outcomes or transfer.

Constructive alignment alone is insufficient in complex workplaces. Workplace-learning research shows that access to activities, guidance and participation is unevenly afforded by social and organizational settings (Billett, 2001, 2004; Tynjälä, 2008). In civil-military exercises, relational, procedural and structural conditions may therefore support or constrain engagement with an otherwise coherent design. Edmondson’s (1999, 2003) research further identifies context support, leader behavior and psychological safety as conditions associated with learning behavior in teams.

In high-stakes interorganizational exercises, context support may include legal expertise, clear command structures, sufficient time for reflection, familiarity among participants, realistic task distribution and access to facilitators who can translate abstract objectives into practical work. Such conditions are not peripheral. They are part of the learning infrastructure. Without them, participants may be willing to learn but unable to do so because they lack the legal, relational or procedural clarity needed to engage in the task.

This is particularly important in civil-military settings, where actors may use different terminology, follow different chains of command and operate under different mandates (Kalkman and Groenewegen, 2019; Steen-Tveit et al., 2024). Status differences and uncertainty can also affect whether participants speak up with questions or concerns, a mechanism documented in interdisciplinary action teams (Edmondson, 2003). Psychological safety is therefore treated here as a relational condition that may enable learning behaviors such as questioning, feedback-seeking and critical reflection; it is not inferred simply from positive interaction.

Even when exercises are designed with clear objectives and supported by relevant expertise, learning can be obstructed by defensive routines. Argyris (1999) defined defensive routines as tacit behaviors that protect individuals or organizations from embarrassment, conflict or exposure to failure. These routines are often unspoken and may be maintained because they preserve organizational legitimacy, avoid uncomfortable discussions or allow participants to present an exercise as successful even when little learning has occurred.

In civil-military crisis exercises, defensive routines may take several forms: problems may be kept out of plenary discussion, general praise may substitute for performance-focused debriefing or familiar exercise patterns may be repeated without testing whether capability has changed. Research on debriefing shows that structured reflection can support performance improvement, whereas military research on debriefing also documents tensions between knowledge sharing and knowledge hiding when failure is difficult to discuss (Fanning and Gaba, 2007; Tannenbaum and Cerasoli, 2013; Firing et al., 2020). Defensive routines are therefore analytically relevant to what becomes discussable after action, not merely to participants’ intentions.

In workplace learning, defensive routines matter because they can restrict the reflective and feedback processes through which professionals identify errors, reconsider assumptions and adapt practice (Argyris, 1999). Psychological safety is relevant at this point because speaking up, asking questions and reporting concerns are learning behaviors that can be inhibited by interpersonal risk (Edmondson, 1999, 2003). The framework therefore treats defensive routines as an inhibitory mechanism that can weaken the use of otherwise available learning opportunities.

The integrated model specifies a process rather than treating the three constructs as parallel explanations. First, constructive alignment describes formal learning architecture: intended outcomes are linked to role-relevant activities and feedback. Second, context support – including expertise, role clarity, time, participant familiarity and facilitation – conditions whether participants can use that architecture. Psychological safety is located within this contextual layer because it concerns whether participants can take interpersonal risks required for questioning, feedback and critical reflection (Edmondson, 1999, 2003). Third, these conditions are expected to be reflected in learning behaviors such as clarification, feedback-seeking, reflection and coordination. Defensive routines can interrupt this process by suppressing critique or converting evaluation into symbolic affirmation. The empirical endpoints available in this study are therefore observable indicators of progression, such as increasingly elaborated task outputs and role clarification. They are not independent measures of cognitive learning, behavioral competence or transfer. Transfer to ordinary work is treated as a downstream outcome requiring separate evidence (Baldwin and Ford, 1988; Grossman and Salas, 2011).

The study adopts a comparative qualitative reanalysis of two contrasting Swedish exercise settings. The comparison is theory-elaborating rather than experimental or quasi-experimental: it asks whether patterns in the two corpora are consistent with the integrated workplace-learning model, not whether the design of one setting caused superior learning. Comparative case analysis is suitable for examining context-dependent mechanisms and generating analytic propositions, provided that differences between cases are treated explicitly (George and Bennett, 2005; Merriam and Tisdell, 2016).

The two cases were selected because earlier studies documented contrasting exercise architectures and learning-related observations (Hedlund and Alvinius, 2024, 2025a, 2025b). They are not matched cases. Case A is one five-day exercise with a stable, immersive sequence across organizational levels, whereas Case B is a composite of seven exercises conducted over 14 exercise-days in different regions and with heterogeneous participant groups. The cases therefore differ simultaneously in duration, immersion, group stability, setting, participant composition and unit of analysis. These differences are central design limitations and preclude attributing cross-case differences to constructive alignment or context support alone.

No new empirical data was collected for this article. Earlier publications analyzed the exercise settings separately using constructive alignment and, for the seven-exercise corpus, team learning/context support (Hedlund and Alvinius, 2024, 2025a, 2025b). The present reanalysis applies the same deductive categories to both corpora – constructive alignment, context support/psychological safety, defensive routines and observable learning-process indicators – and interprets them through a workplace-learning and transfer lens. Analytic consistency is therefore procedural rather than based on identical data volumes or matched participants. Author-identifying references to the earlier studies are anonymized as Author(s) for double-blind review.

Case A was a five-day civil-military tabletop exercise conducted at a Swedish airbase in June 2024, involving approximately 140 participants from around 40 organizations at strategic, regional/operational and local levels. The exercise was organized as a three-stage handover: strategic-level work on day one, regional/operational work on days two and three and local/municipal work on days four and five. Participants were selected from the same geographical and administrative context and generally had prior working relationships. Legal expertise was continuously available. The exercise used two explicit learning objectives concerning evacuation and accommodation under wartime conditions and an enhanced constructive-alignment design (Hedlund and Alvinius 2025a).

The substantive scenario in Case A concerned evacuation and accommodation under wartime conditions. The setting required participants to connect preparedness responsibilities, legal interpretation and interorganizational coordination across decision levels. The prior study reports that no formal task assessment was conducted; participant reflections during group work and final evaluation, together with observed behavior and task outputs, were used as indicators of perceived learning, engagement and practical relevance (Hedlund and Alvinius 2025a). This distinction is retained in the present reanalysis.

Case B comprises seven civil-military tabletop exercises conducted between April 2022 and October 2023. Together they involved 661 participants, between 7 and 40 organizations per exercise, and 14 exercise-days in total (Hedlund and Alvinius, 2024). Participants were primarily personnel with preparedness responsibilities from military regions, county administrations, health care and police regions, with municipal and fire-service participation in several exercises. The earlier study describes substantial heterogeneity in crisis-management experience; municipal preparedness coordinators were often relatively new to their roles, while regional and professional participants brought different forms of sector expertise that did not necessarily include interorganizational crisis-management experience (Hedlund and Alvinius, 2024).

The comparison is consequently a structured contrast between two nonequivalent empirical settings. It is used to examine whether the same analytical dimensions appear in different configurations, not to rank a single “successful” intervention against a homogeneous control condition. Any difference in observable learning indicators can plausibly reflect the combined influence of design, duration, immersion, participant stability, prior familiarity, setting and evidentiary depth. Table 1 summarizes both the substantive comparison and these comparability limits.

The empirical corpora are not identical. Case A relied on participant observation across the five-day exercise, with detailed field notes guided by constructive alignment and context-support concepts; participant reflections during group work and the final evaluation were also documented (Hedlund and Alvinius 2025a). Case B combined documents distributed by exercise leaders in advance with participant observation in different groups. The observer followed activities for an entire day at each of the seven exercises and produced detailed daily field notes using a constructive-alignment checklist (Hedlund and Alvinius, 2024). Informal participant comments were recorded as field material where they arose, but they were not conducted as a standardized interview sample.

The available source records do not provide standardized totals for observation hours, number of informal conversations or document volume that are directly comparable across the two cases. Nor do they provide individual-level distributions of prior exercise experience for all participants. Rather than reconstructing such counts retrospectively, the present study reports the empirical coverage that can be traced to the original studies and treats unequal evidentiary depth as a limitation. Table 1 therefore distinguishes reported corpus characteristics from information that was not systematically recorded.

Across both cases, the reanalysis focused on the same observable categories: references to learning objectives during tasks; clarity of roles, mandates and legal frameworks; availability of expertise and facilitation; opportunities for feedback and reflection; expressions of uncertainty, questioning or resistance; defensive patterns around critique; and changes in the elaboration of task outputs. These categories were used as indicators of the learning process and exercise conditions, not as measures of individual competence.

The material was not treated as evidence of individual competence, cognitive gain or transfer. Observations and exercise products were analyzed as indicators of how the temporary learning environment afforded or constrained collective learning behaviors and progression within the exercise. This distinction is important because increasingly detailed outputs may reflect facilitation, accumulated scenario information or group composition as well as learning. Claims are therefore limited to observed process indicators unless independent outcome evidence is available.

Selected data excerpts were translated from Swedish to English and checked by the authors. The present manuscript distinguishes observations reported in the original empirical studies from the cross-case interpretations made in this reanalysis. No new participant data was collected for this article.

The reanalysis followed a deductive thematic approach (Proudfoot, 2023). A common coding frame was applied to both corpora. Constructive alignment captured the relationship among intended learning outcomes, activities and feedback/assessment; context support captured expertise, role clarity, participant familiarity, time and facilitation; psychological safety captured observed opportunities for questions, disagreement and admission of uncertainty; and defensive routines captured avoidance of critique, symbolic success narratives and reluctance to examine performance. A final category recorded observable indicators of progression without coding them as verified learning outcomes.

The cross-case analysis proceeded in two stages. First, each case was summarized independently using the common categories. Second, the summaries were compared to identify convergent and divergent patterns and to consider plausible alternative explanations arising from structural case differences. The analytic question was whether the observed pattern was consistent with the proposed process model, not whether one component produced a causal effect. This common frame improved procedural consistency across the reanalysis while leaving the unequal corpus sizes and case structures visible.

Both empirical settings originate in earlier studies by the author team, and the analytical concepts used here overlap with concepts used in those publications (Hedlund and Alvinius, 2024, 2025a, 2025b). In Case A, participant observation was conducted during an exercise whose design was explicitly informed by an enhanced constructive alignment model (Hedlund and Alvinius 2025a). This proximity creates a plausible risk of confirmatory interpretation. The present reanalysis addresses that risk by stating the nonequivalence of the cases, applying the same deductive categories to both corpora, separating observed indicators from stronger learning-outcome claims and explicitly considering structural alternative explanations. These safeguards reduce but cannot eliminate positionality effects.

Both empirical settings originate in earlier studies by the author team, and the analytical concepts used here overlap with concepts used in those publications (Hedlund and Alvinius, 2024, 2025a, 2025b). In both Case A and Case B, the first author participated solely as a researcher and observer and had no role in the operational design or delivery of the exercises. Although the design of the exercise in Case A was informed by the enhanced constructive-alignment model discussed in Hedlund and Alvinius 2025a, the first author was not involved in designing or implementing the exercise itself. The first author conducted the observations in both cases, while both authors participated in the subsequent analysis of the empirical material. The second author, who had not participated in either exercise or in the data collection, contributed a more distanced perspective during the analysis, providing an additional safeguard against confirmatory interpretation. Nevertheless, the first author’s proximity to the empirical settings and the overlap with concepts developed in earlier work create a plausible risk of confirmatory interpretation. The present reanalysis addresses that risk by acknowledging the nonequivalence of the cases, applying the same deductive categories to both corpora, separating observed indicators from stronger learning-outcome claims, explicitly considering alternative structural explanations and involving a second author who was not present during data collection. These safeguards reduce but cannot eliminate positionality effects.

The study was submitted to the Swedish Ethics Review Authority for approval. According to Decision Ref. No. 2022-02944-01, the project was classified as not involving personal data, thereby exempting it from detailed review. Key ethical considerations included informing participants about the study’s purpose and their rights and ensuring the confidentiality and anonymity of their identities and organizational affiliations. Participants were assured that their personal information would remain confidential in any published findings or reports.

The comparative case architecture, empirical basis and principal comparability limits are summarized in Table 1.

The comparative reanalysis identified differences in how the two exercise settings afforded and displayed workplace-learning processes. Because the cases are structurally nonequivalent, the findings are presented as within-case patterns and cross-case contrasts rather than as causal effects. Four themes are reported: constructive alignment, context support, defensive routines and observable indicators of learning progression.

Case A demonstrated a high degree of pedagogical intentionality and coherence. Two clearly formulated learning outcomes were communicated at the start of the exercise: to plan the evacuation of civilians from a designated conflict zone and to coordinate the reception and accommodation of evacuees. These outcomes were embedded in a structured scenario and linked to a phased progression across strategic, operational and local levels. Each level was responsible for initiating, refining or implementing decisions, thereby reinforcing its role in the broader coordination chain.

Participants in Case A frequently referred to the learning objectives when clarifying task responsibilities. Activities were sequenced across strategic, operational and local levels and group outputs and facilitator-led reflection created opportunities for feedback. Because no formal pre-/posttest or independent task assessment was used, these features are interpreted as evidence of a coherent learning process and feedback architecture, not as proof that specified learning outcomes were attained.

The structured design also helped participants work across professional boundaries. Rather than discussing collaboration in general terms, groups had to translate broad preparedness responsibilities into sequential tasks and shared outputs. This made differences in terminology, mandate and organizational habits visible during the exercise itself. For workplace learning, this visibility was important because it allowed participants to identify where coordination problems arose and adjust their understanding as the task unfolded.

Case B showed a different pattern. Where objectives existed, they were vague and declarative, for example, to develop a shared situational picture, without being translated into measurable learning outcomes or concrete tasks. Learning activities were generic and weakly differentiated. Participants with varying levels of experience were assigned to groups and asked to complete tasks with limited guidance. One participant described the experience, noting that the group was given tasks but no clear direction and therefore defaulted to familiar routines. Another participant observed that it was unclear what learning was supposed to occur.

Case B had no systematic assessment mechanism beyond broad plenary sessions. These sessions often confirmed completion of the exercise rather than examining performance against explicit learning criteria. Observations of confusion, generic outputs and surface-level engagement were therefore coded as indicators of weak alignment and limited opportunities for feedback, rather than as direct measures of an absence of learning.

In Case A, stronger context support co-occurred with more sustained engagement in the exercise tasks. Legal experts were available in real time, helping participants clarify applicable rules and jurisdictional boundaries; participants also worked within a relatively coherent geographical and administrative network. These observations are consistent with the proposition that access to expertise and role clarity can improve the usability of workplace learning opportunities (Billett, 2001, 2004), but the case design does not isolate their independent effects.

Group composition also supported learning. Participants were assigned based on geographic and institutional alignment, so several had prior experience working with one another. Leadership roles followed existing command structures, improving vertical coordination and mirroring real-world decision chains. Roles such as facilitator, note-taker and presenter were rotated among several groups, encouraging participation and distributing responsibility for the learning process.

Case B lacked several of these enabling conditions. Groups were assembled with insufficient regard for participants’ roles, experience or prior collaboration. Some participants reported that they did not know who others were or what organizational mandate they represented. Legal experts were absent, leaving participants uncertain about decision-making authority and jurisdictional limits. The same generic exercise model was used across regions despite differences in organizational capacity and prior experience. This one-size-fits-all design limited the exercise’s ability to support workplace learning for participants with different needs.

In Case B, heterogeneous prior experience meant that participants did not enter with a common starting point. The earlier study describes municipal preparedness coordinators who were often new to their roles alongside regional and professional participants with different sector expertise (Hedlund and Alvinius, 2024). A largely uniform exercise format therefore afforded uneven opportunities to participate productively. This pattern is consistent with workplace-learning research emphasizing that affordances and engagement interact, but it cannot be separated from the other structural differences between the cases (Billett, 2004; Tynjälä, 2008).

Both cases included resistance to structured exercise methods, but the resistance was handled differently. In Case A, some municipal participants initially questioned the structured decision-making model and worried that it imposed unnecessary rigidity on local practice. Facilitators did not dismiss this resistance. Instead, they used it as an opportunity for clarification and reflection. They explained the purpose of the model, adjusted terminology to fit local practice and emphasized that structure could support rather than override local autonomy. Over time, participants came to see the method as a common framework for collaboration.

In Case A, participants were observed to express discomfort with the structured method, while facilitators responded with clarification and adaptation rather than sanctions. These observations are consistent with conditions that can support psychological safety and speaking up (Edmondson, 1999, 2003), but psychological safety was not measured directly. The findings are therefore limited to observed opportunities for disagreement and reflection.

In Case B, defensive routines were more pervasive and less openly addressed. Participants expressed frustration during breaks and informal conversations but rarely raised critical issues in plenary settings. Some sessions ended with scripted success declarations despite visible confusion or dissatisfaction. A participant noted that the same discussions had been repeated for years without progression and that actors continued to improvise without knowing what was legally permitted. Some senior participants privately acknowledged that the exercises were maintained partly to preserve institutional legitimacy rather than to generate demonstrable learning.

The Case B observations are consistent with the risk that critique becomes decoupled from formal evaluation when concerns are voiced informally but not carried into plenary reflection. This interpretation aligns with Argyris’s (1999) account of defensive routines and with military debriefing research showing that learning can be constrained when knowledge about failure is not openly shared (Firing et al., 2020). The data do not establish participants’ motives, but they do document a recurring separation between informal criticism and formal success narratives.

Case A showed several observable indicators consistent with learning progression during the exercise. Groups produced increasingly detailed plans that incorporated legal constraints, role clarification and interorganizational coordination, and participants described the structured process as clarifying responsibilities. These observations do not establish “cognitive learning” in an outcome-measurement sense: the study did not use independent pre-/postmeasures, retention tests or behavioral assessment. The more defensible inference is that task outputs and reflections became more elaborated as the exercise progressed.

Case B displayed fewer comparable indicators of progression. Output was often vague or repetitive, and participants questioned the value of recurring activities. The corpus also contained limited evidence that lessons were systematically documented and carried into subsequent exercises. However, transfer into ordinary work was not directly assessed in either case, and the greater turnover and multiexercise structure of Case B make continuity a contextual difference rather than a controlled explanatory variable.

Taken together, the cases display contrasting configurations of exercise design, context support, feedback and defensive routines. However, the comparison cannot determine whether those features, rather than duration, immersion, group stability, participant composition or other structural differences, account for the observed contrasts. The findings below should therefore be read as patterned associations within two nonequivalent settings.

The comparison supports treating civil-military crisis exercises as temporary workplace learning environments rather than assuming that preparedness activity is inherently developmental. This interpretation converges with workplace-learning research showing that learning opportunities depend on how work settings afford participation and guidance (Billett, 2001, 2004; Tynjälä, 2008) and with emergency-exercise research demonstrating that exercise structure and format matter for perceived learning and usefulness (Berlin and Carlström, 2015; Borell and Eriksson, 2013; Roud et al., 2021). At the same time, the present cases do not permit the stronger claim that alignment caused the observed contrast.

The theoretical relevance of constructive alignment is therefore narrower and more specific than a claim that an aligned exercise automatically produces learning. Alignment describes the coherence of the learning opportunity: what participants are expected to develop, what they do and how their work is discussed. Workplace-learning research adds that the opportunity must also be accessible and taken up, while transfer research distinguishes immediate training performance from generalization and maintenance in the workplace (Billett, 2004; Baldwin and Ford, 1988; Grossman and Salas, 2011). For exercise planners, this means that scenario design and learning design should be treated as related but distinct tasks.

The cross-case pattern also highlights context support as part of the learning infrastructure. Case A combined legal expertise, role clarity, prior relationships and structured facilitation; Case B contained greater heterogeneity and weaker access to comparable support. This pattern is consistent with Billett’s (2001, 2004) account of workplace affordances and with Edmondson’s (1999, 2003) evidence that context support, leader behavior and psychological safety are associated with learning behavior. It also echoes findings from cross-organizational crisis exercises that emphasize the importance of relationships and coordination conditions (Steen-Tveit et al., 2024).

For workplace-learning research, the important point is that participants enter temporary interorganizational settings with different histories, authority, expertise and opportunities to contribute. The exercise is therefore not a neutral container into which a pedagogical design is inserted. The observed contrast suggests that formal alignment and workplace affordances should be analyzed together, while the nonequivalent case structures caution against treating context support as a single variable with an isolated effect.

This distinction also changes what should count as evaluation evidence. Participation counts and satisfaction can document reach or acceptability, but they do not establish learning or transfer. Debriefing research indicates that structured reflection and feedback can improve subsequent performance, particularly when the debrief is aligned with the task and facilitated effectively (Fanning and Gaba, 2007; Tannenbaum and Cerasoli, 2013). Training-transfer research further requires evidence that learning is generalized and maintained in the work setting (Baldwin and Ford, 1988; Grossman and Salas, 2011). Future civil-military exercise evaluation should therefore separate process indicators, immediate task performance and longer-term transfer.

Defensive routines add a social explanation for why available feedback opportunities may not be used. In Case B, informal criticism was not always reproduced in plenary evaluation, a pattern consistent with Argyris’s (1999) account of organizational defenses. Military debriefing research similarly shows that organizational learning depends on whether people share or hide knowledge about problematic performance (Firing et al., 2020). These literatures support interpreting the observed formal/informal discrepancy as a potential learning barrier, while the present data does not establish why individual participants chose silence or critique.

Case A provides a contrasting observational pattern: disagreement about the structured method was surfaced and discussed. This is compatible with Edmondson’s (1999, 2003) account of speaking up under psychologically safer conditions and with debriefing research emphasizing facilitated reflection (Fanning and Gaba, 2007; Tannenbaum and Cerasoli, 2013). The inference is process-oriented: facilitation appeared to keep disagreement available for collective examination. The study did not directly measure psychological safety or demonstrate that this interaction produced retained learning.

The theoretical contribution is a process specification for temporary interorganizational workplace learning rather than a new construct. Constructive alignment specifies the formal learning architecture; workplace affordances and context support specify whether participants have the resources and opportunities to engage; psychological safety concerns whether interpersonal risk permits questioning and critique; and defensive routines describe how reflection and feedback can be inhibited. Observable exercise outputs sit downstream of these processes, whereas transfer to ordinary work is a further outcome that the present study does not measure. This synthesis clarifies both the complementarity and the boundaries of the component theories.

The practical implications follow from this process model but should be treated as design propositions rather than proven causal prescriptions. Exercise planners can define specific workplace capabilities, connect them to role-relevant activities, plan feedback against explicit criteria, provide legal and other expert support and create structured opportunities for questions and critique. They should also document what evidence will count as immediate progression and what separate follow-up evidence will be needed to assess transfer. These principles are consistent with research on workplace affordances, training transfer and structured debriefing (Billett, 2004; Grossman and Salas, 2011; Tannenbaum and Cerasoli, 2013).

The framework may be relevant beyond Sweden because simulations and tabletop exercises are widely used in emergency and high-risk professional settings, but its transferability should be tested rather than assumed. Existing research shows that exercise format can be associated with different perceived learning and usefulness outcomes (Roud et al., 2021), while cross-organizational crisis research underscores the importance of context-specific coordination conditions (Steen-Tveit et al., 2024). The present study adds a workplace-learning vocabulary for examining such variation, not evidence that one exercise architecture will work uniformly across countries or sectors.

The study has several central limitations. First, the comparison is structurally uneven: Case A is one five-day immersive exercise, whereas Case B is a composite of seven exercises conducted over 14 exercise-days in multiple settings. Duration, immersion, group stability, geography, participant heterogeneity and unit of analysis therefore vary together with pedagogical design. Second, the empirical corpora are unequal and do not contain standardized, directly comparable counts of observation hours, informal conversations or document volume. Third, participant experience is described at group level in the source studies rather than measured with a common individual-level instrument. These limitations prevent causal attribution of the observed cross-case differences.

Fourth, the evidence supports claims about observed learning processes and progression indicators, not verified cognitive learning, retained competence or workplace transfer. No independent pre-/postmeasures, behavioral performance tests or longitudinal transfer measures were used. Fifth, the authors’ prior involvement in studying these exercises, and the theory-informed design of Case A, create a risk of confirmatory interpretation. The common deductive frame, explicit alternative explanations and moderated outcome claims increase transparency but do not remove this risk.

Future research should test the proposed process model with more comparable cases and prospective designs. Matched exercise formats, systematic participant profiles and standardized observation protocols would help separate pedagogical design from duration, composition and setting. Pre-/postknowledge or performance measures could test immediate learning, while longitudinal observation could assess whether capabilities generalize and are maintained in ordinary work (Baldwin and Ford, 1988; Grossman and Salas, 2011). Research should also examine facilitation and debriefing practices directly, including whether they support speaking up and reduce knowledge hiding (Tannenbaum and Cerasoli, 2013; Firing et al., 2020).

The comparison suggests that civil-military crisis exercises differ in the extent to which they afford observable workplace-learning processes. In these cases, clearer alignment, stronger context support and opportunities for critical dialogue co-occurred with more elaborated task outputs, whereas weaker feedback structures and defensive routines co-occurred with fewer indicators of progression. Because the settings were not equivalent, these relationships should not be read as causal estimates.

For workplace-learning research, high-stakes interorganizational exercises are useful settings for examining how formal learning architecture interacts with workplace affordances and interpersonal conditions. The study’s main proposition is that design, context support and defensive routines should be analyzed as a process that shapes opportunities for learning behavior. Demonstrating learning outcomes and transfer requires additional evidence beyond observation and exercise products. This distinction offers a more cautious basis for designing and evaluating civil-military exercises as workplace-learning interventions.

During preparation of this revision draft, OpenAI GPT-5.6 Sol was used as an editorial aid to identify passages requiring clarification, reorganize author-provided material and suggest language revisions. The tool did not collect or alter research data. The authors are responsible for independently verifying the accuracy of all substantive revisions and references and for ensuring that the final submitted manuscript complies with the publisher’s current generative-AI policy.