Institutional incentives and faculty cooperation in scholarship of teaching and learning: a scenario-based study in medical education.

Institutional incentives and faculty cooperation in scholarship of teaching and learning: a scenario-based study in medical education - PMC Skip to main content An official website of the United States government Here's how you know Here's how you know Search Log in Primary site navigation Search Log in Search PMC Full-Text Archive Search in PMC Journal List User Guide As a library, NLM provides access to scientific literature. Inclusion in an NLM database does not imply endorsement of, or agreement with, the contents by NLM or the National Institutes of Health. Learn more: PMC Disclaimer | PMC Copyright Notice BMC Med Educ . 2026 Apr 22;26:901. doi: 10.1186/s12909-026-09267-8 Institutional incentives and faculty cooperation in scholarship of teaching and learning: a scenario-based study in medical education Luchuan Li 1, # , Tianfei Ran 2, # , Yongjie Qin 1 , Hongxiao Fan 1 , Chunji Huang 3, ✉ Author information Article notes Copyright and License information 1 Teaching Evaluation Center, Army Medical University, Chongqing, 400038 P.R. China 2 Department of Orthopedics, Xinqiao Hospital, Army Medical University, Chongqing, 400038 P.R. China 3 Army Medical University, Chongqing, 400038 P.R. China ✉ Corresponding author. # Contributed equally. Received 2026 Jan 28; Accepted 2026 Apr 16; Collection date 2026. © The Author(s) 2026 Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/ . PMC Copyright notice PMCID: PMC13234972  PMID: 42021238 Abstract Background Sustained collaboration between basic science and clinical faculty in the Scholarship of Teaching and Learning (SoTL) remains challenging, partly because institutional incentives are often misaligned with the conditions required for cooperation. Methods We conducted a scenario-based discrete-choice experiment with 240 medical faculty members from four universities in China. The scenarios varied expected synergy, collaboration cost, inequity risk, direct reciprocity, and reputational visibility. We estimated logit models and mapped the results onto a behavioral game-theoretic threshold framework. Results Increased expected synergy, direct reciprocity, and reputational visibility raised willingness to collaborate, whereas higher cost and inequity risk reduced cooperation. In the pooled model, direct reciprocity showed the largest positive association with cooperation. Group comparisons indicated that clinical faculty were more responsive to direct reciprocity, whereas basic science faculty were more responsive to reputational visibility. In the calibrated threshold analysis, supportive mechanisms lowered the minimum expected synergy required for incentive-compatible cooperation, with reputational visibility showing a larger threshold reduction among basic science faculty and direct reciprocity showing a larger threshold reduction among clinical faculty. Conclusions Cross-group SoTL collaboration may be strengthened by institutional designs that combine credible reciprocity structures, visible recognition of educational scholarship, and fairness safeguards, while taking into account the different mechanism sensitivities of basic science and clinical faculty. Supplementary Information The online version contains supplementary material available at 10.1186/s12909-026-09267-8. Keywords: Medical education, Scholarship of Teaching and Learning, Faculty collaboration, Institutional incentives, Discrete-choice experiment, Game-theoretic modeling Introduction The Scholarship of Teaching and Learning (SoTL) is fundamentally concerned with the systematic, evidence-informed study of teaching and learning processes, characterized by core elements such as critical reflection, public dissemination, and peer review [ 1 , 2 ]. It aims to improve educational practice and student outcomes, and prior scholarship has linked engagement in SoTL to improvements in teaching practice, course design, and students’ learning experiences [ 3 , 4 ]. Collaborative SoTL, particularly through interdisciplinary engagement, can further broaden the scope, rigor, and relevance of such inquiry [ 5 ]. In medical education, fostering collaboration between basic science and clinical faculty is especially critical, as integrating foundational knowledge with clinical experience is essential for developing coherent curricula and enhancing student learning [ 6 , 7 ]. General SoTL literature has long emphasized the importance of institutional support, targeted academic development, and mentoring or resource-related guidance in fostering educational scholarship [ 8 – 10 ]. More recent work specific to medical education has focused primarily on the development, evaluation, and structural support of educational scholarship—examining, for instance, the assessment of faculty SoTL competence [ 11 ], the role of workshops, mentoring, and peer writing groups [ 12 ], the function of faculty learning communities [ 13 ], and the contribution of SoTL to professional identity formation [ 14 ]. Collectively, this valuable body of research clarifies how SoTL can be strengthened and what institutions should provide to foster a collaborative environment. Related work across the health professions has also documented persistent barriers to scholarship, including time constraints, competing professional demands, and limited institutional recognition [ 15 , 16 ]. Nevertheless, this literature remains primarily evaluative, developmental, or programmatic in orientation. It addresses the question of what should be offered, but pays limited attention to how individual faculty weigh the decision to enter a specific cross-group collaborative project. More specifically, it remains unclear how instructors, when presented with a potential interdisciplinary SoTL opportunity, make nuanced trade-offs among anticipated educational gains, personal time costs, the risk of unfair recognition, uncertainty about reciprocal commitment from collaborators, and the possibility of visible institutional credit. The absence of insight into these micro-level socio-psychological decision-making mechanisms may leave institutional designs only partially aligned with the actual incentive structures governing faculty participation. To open this “decision black box,” a theoretical lens is needed that can account for both strategic interdependence and social preferences. Behavioral game theory provides such a framework [ 17 ]. Unlike traditional models based on pure material self-interest, behavioral game theory incorporates fairness concerns, reciprocity motives, and reputational incentives into decision making, thereby showing how cooperation depends on expectations about others’ behavior and on how institutional arrangements shape the perceived risks and rewards of participation [ 17 – 22 ]. In particular, inequity-aversion models suggest that anticipated unequal outcomes can discourage cooperation even when joint gains are possible [ 19 – 20 ]. Reciprocity-based models further indicate that individuals tend to reward perceived kindness and punish hostility, such that credible signals of cooperative intent from others can catalyze participation [ 18 , 21 ]. Reputational visibility (operationalized as an indirect reciprocity mechanism) and reputation-based models likewise show that, in repeated interactions or when institutions make behavior visible and memorable, building a good reputation can itself become an incentive for cooperation [ 22 ]. These mechanisms are directly relevant to cross-group SoTL collaboration, where participation depends not only on expected educational value but also on whether collaboration appears fair, mutually committed, and institutionally visible. Accordingly, we argue that decisions about cross-group SoTL collaboration are best understood as a behavioral game rather than as a purely static cost-benefit calculation. Such decisions are strategic choices made under uncertainty about collaborators’ behavior and about whether collaborative contributions will be institutionally recognized, and they are shaped by social preferences as well as instrumental considerations. To examine this proposition, the present study integrates a scenario-based discrete-choice experiment with a behavioral game-theoretic framework to empirically investigate how five core mechanisms derived from behavioral game theory—expected synergy gains, collaboration cost, inequity risk, direct reciprocity, and reputational visibility—affect and explain the willingness of basic science faculty and clinical faculty to engage in cross-group SoTL collaboration in medical education. In the context of Chinese medical education, basic science faculty and clinical faculty often work under different role expectations and evaluation pressures. Basic science faculty are more commonly embedded in departments oriented toward foundational teaching and academic output, whereas clinical faculty typically balance clinical service, teaching, and research responsibilities [ 23 , 24 ]. In this setting, cross-group SoTL collaboration commonly seeks to bridge foundational and clinical knowledge through joint curriculum development, teaching innovation, or educational research. Against this background, the present study seeks to move research on SoTL collaboration beyond broad organizational-support narratives toward a micro-mechanism-based analysis of decision logic that can inform more precise, context-sensitive institutional design [ 23 – 25 ]. Methods Theoretical framework and model Why a behavioral game-theoretic framework? The choice of a behavioral game-theoretic framework is not merely terminological. Rather, it is theoretically appropriate because the mechanisms it highlights—conditional cooperation, inequity aversion, reciprocity, and reputation—closely match the decision psychology involved when medical educators consider ambiguous, time-consuming, and uncertain-return SoTL collaborations [ 16 – 20 ]. This framework is appropriate for three reasons. First, the value of collaboration is conditional. The ultimate educational or scholarly value of a project—its synergy gain—depends heavily on whether both parties contribute substantive effort. Cross-group SoTL collaboration therefore has the structure of a coordination or assurance problem, in which equilibrium depends on participants’ mutual expectations. Second, decisions are shaped by social preferences. Faculty members may be highly sensitive to fairness in authorship, workload allocation, and recognition; to the likelihood that a collaborator will reciprocate effort and goodwill; and to whether participation will generate visible reputational returns. In many cases, these considerations may matter as much as, or more than, immediate material benefits. Such motivations lie at the core of behavioral game theory [ 18 – 22 ]. Third, institutional environments reshape incentives. Institutional arrangements that reduce the perceived risk of being exploited—for example, through transparent rules, clearer authorship norms, or more explicit task allocation—can lower the threshold for collaboration. Likewise, arrangements that reward visible SoTL engagement, such as promotion criteria or formal teaching recognition, can increase the payoff to prosocial participation and thereby alter the structure of the game. Viewed in this way, cross-group SoTL collaboration is not simply a matter of static cost-benefit calculation. It is a socially conditioned strategic choice, made under uncertainty about others’ behavior and about how collaborative contributions will be valued within the institution. From theory to variables: defining five core mechanisms We operationalize key concepts from behavioral game theory into five mechanisms that are both empirically observable in medical education settings and potentially amenable to institutional intervention. Expected synergy gains refer to the additional educational or scholarly value that may be created through cross-group collaboration. In game-theoretic terms, this is the potential cooperative surplus that makes movement away from a low-cooperation equilibrium attractive. In the present study, it captures the added value that may arise when basic science and clinical perspectives are combined in teaching innovation or educational scholarship. Collaboration cost represents the participation constraint or opportunity cost associated with collaboration. Any collaborative decision must overcome this baseline threshold, including time investment, coordination burden, and the diversion of effort from other professional tasks [ 26 ]. In the model, this appears as the direct disutility attached to choosing the cooperative strategy. Inequity risk corresponds to fairness-preference theory [ 19 – 20 ]. When faculty expect that their contributions may be undervalued, that authorship may be distributed unfairly, or that workload may be unevenly assigned, the subjective utility of collaboration can decline sharply even when synergy gains are possible. This mechanism helps explain why vague or weakly specified collaborations often fail to attract participation. Direct reciprocity corresponds to reciprocity-based models [ 18 , 21 ]. A credible signal that the prospective collaborator will also contribute—for example, through explicit commitments, reciprocal resource exchange, or clear task allocation—can increase the expectation of mutual cooperation and thereby move the interaction toward a cooperative equilibrium. This mechanism is especially important in the formation of initial trust. Reputational visibility corresponds to indirect reciprocity and reputation-based models [ 22 ]. When institutions can record, recognize, and reward SoTL engagement, faculty may derive utility from being seen as contributors to educational improvement, even if the outcome of any single collaboration is uncertain. This mechanism provides a longer-term incentive for cooperation that is partially independent of the immediate behavior of a particular partner. Taken together, these five mechanisms translate the abstract logic of behavioral game theory into concrete and measurable determinants of collaborative choice. They also provide a bridge from theory to institutional design, because each can, in principle, be modified by policy, organizational norms, or incentive structures. Formal model We formalize the interaction as a static, asymmetric two-player game between a basic science faculty member (Group A) and a clinical faculty member (Group B). Each player chooses to Cooperate (C) or Not Cooperate (D). The model’s asymmetry is crucial, acknowledging that the two groups may value the same institutional incentive differently due to distinct roles, evaluation pressures, and opportunity costs. Let Group A choose and Group B choose . For each group total utility consists of a constant baseline payoff (R) plus an incremental payoff that depends on both players’ actions. Because R is common to all outcomes, it does not affect strategic comparisons; therefore, we focus on the incremental payoff component, specified as: 1 Here, and index the vignette’s Synergy and Cost levels (Low/Medium/High), and , , and are the binary attribute indicators. The parameters , , , , and are marginal utility weights (per level or per indicator) that translate the experimentally manipulated attribute levels into payoffs. The parameters and are the marginal utility weights (per level or per indicator) to be estimated. They translate the experimentally manipulated attributes into subjective payoffs, and their signs and magnitudes directly reflect the influence of the five core mechanisms. Crucially, each parameter maps directly onto a theoretical mechanism and its operationalization in our experiment: Synergy weight ( ): Captures the utility from Expected Synergy Gains . It corresponds to the cooperative surplus that motivates moving away from a non-cooperative equilibrium.Cost weight ( ): Represents the disutility from Collaboration Cost , the participation constraint or opportunity cost that must be overcome.Inequity aversion weight ( ): Reflects sensitivity to Inequity Risk , rooted in fairness-preference theory. It quantifies the aversion to being exploited in an unfair outcome.Direct reciprocity weight ( ): Captures the utility from a credible Direct Reciprocity signal, based on reciprocity models. It increases the expected payoff from mutual cooperation.Reputational visibility weight ( ): Represents the utility from Reputational Visibility , based on reputational visibility and reputation models. It provides a payoff for cooperating that is independent of the partner’s action. The purpose of this model is not to assume that faculty perform exact calculations, but to provide a structured, testable analytical framework. It allows us to quantify how the five mechanisms influence the relative attractiveness of cooperation, examine trade-offs between them, and simulate how institutional interventions (e.g., enhancing or reducing perceived ) could systematically lower the minimum synergy threshold required to trigger collaboration. This forms the conceptual bridge linking our discrete choice experiment to the subsequent threshold analysis. In plain language, the model assumes that a faculty member weighs the expected educational value of collaboration against the time and effort it requires, while also considering whether the collaboration seems fair, whether the partner’s commitment appears credible, and whether the institution will visibly recognize the contribution. The model is therefore intended as a structured representation of the decision logic underlying cooperation rather than as a literal claim that faculty compute exact numerical payoffs. Instrument development and survey administration Scenario-based discrete-choice instrument We developed a self-designed, scenario-based discrete-choice instrument to elicit faculty preferences under structured SoTL collaboration trade-offs. The instrument was not designed as a reflective psychometric scale measuring a single latent trait. Instead, respondents were presented with a series of standardized collaboration scenarios in which key institutional conditions were experimentally varied, and they made a binary choice between cooperating and not cooperating. For this reason, conventional internal-consistency measures such as Cronbach’s alpha and reverse-scoring procedures are not directly applicable to the present instrument. The final questionnaire comprised 17 vignettes describing hypothetical cross-disciplinary SoTL collaboration opportunities between basic science faculty and clinical faculty. In each vignette, respondents were asked to assume that the prospective collaborator belonged to the other disciplinary group and to indicate whether they would participate in the collaboration. After each vignette, respondents could also provide an optional certainty rating (1–5) for their binary choice. Because this ancillary item was optional and was not part of the primary choice-modeling framework, it was not included in the main analyses. The five experimentally manipulated attributes were expected synergy gain (three levels), collaboration cost (three levels), inequity risk (two levels), direct reciprocity (two levels), and reputational visibility (two levels). To illustrate their operationalization, Box 1 presents one representative vignette. The full questionnaire is provided in the Supplementary Questionnaire. Box 1. Representative vignette used in the survey . Expected synergy gain (High) This cross-group SoTL collaboration is expected to substantially improve student learning outcomes and is very likely to generate a high-quality publishable output. Collaboration cost (Medium) You expect to invest a moderate level of additional effort (about 1 month). Inequity risk (Low; clear and traceable rules) Roles, authorship, and contribution-recording rules have been clarified at the start and can be tracked throughout. Direct reciprocity (High) The collaborator makes a clear and verifiable commitment to reciprocate and will take primary responsibility for data analysis and drafting the initial manuscript. Reputational visibility (High) Your contribution will be included in the school’s formal recording system and can be counted toward annual evaluation and education contribution portfolios. Choice task Would you participate in this collaboration? (C = cooperate; D = not cooperate) The 17 vignettes were selected from a full-factorial candidate set of 72 scenarios using a D-optimal fractional factorial design. This design was chosen to maximize information efficiency while keeping the respondent burden manageable. Design diagnostics indicated near-orthogonality, with low inter-attribute correlations and variance inflation factors below 1.2. Full design details are provided in the Supplementary Materials. In the primary estimation, the two three-level attributes, synergy and cost, were coded as ordinal linear variables representing low, medium, and high levels. The three institutional mechanisms with two levels were coded as binary indicators. To assess whether the equal-spacing assumption for the three-level attributes materially affected inference, we also estimated an alternative categorical specification for synergy and cost, with the corresponding robustness results reported in Table 2 . Table 2. Main-effects logistic regression results Panel A. Robustness check Effect β SE z p OR [95% CI] Synergy: Medium vs. Low 0.059 0.090 0.65 0.514 1.06 [0.89, 1.26] Synergy: High vs. Low 2.006 0.117 17.07 < 0.001 7.43 [5.90, 9.35] Cost: Medium vs. Low 0.243 0.113 2.15 0.031 1.28 [1.02, 1.59] Cost: High vs. Low -0.521 0.086 -6.04 < 0.001 0.59 [0.50, 0.70] Inequity risk (High vs. Low) -1.808 0.108 -16.69 < 0.001 0.16 [0.13, 0.20] Direct reciprocity (High vs. Low) 2.234 0.118 18.99 < 0.001 9.34 [7.41, 11.76] Reputational visibility (High vs. Low) 1.467 0.110 13.37 < 0.001 4.34 [3.50, 5.38] Constant -1.978 0.109 -18.13 < 0.001 0.14 [0.11, 0.17] Panel B. Primary specification Effect β SE z p OR [95% CI] Synergy level (per + 1 level) 0.927 0.054 17.07 < 0.001 2.53 [2.27, 2.81] Cost level (per + 1 level) -0.377 0.038 -10.01 < 0.001 0.69 [0.64, 0.74] Inequity risk (High vs. Low) -1.800 0.107 -16.89 < 0.001 0.17 [0.13, 0.20] Direct reciprocity (High vs. Low) 2.038 0.101 20.18 < 0.001 7.67 [6.30, 9.35] Reputational visibility (High vs. Low) 1.525 0.098 15.59 < 0.001 4.59 [3.79, 5.56] Constant -1.817 0.088 -20.74 < 0.001 0.16 [0.14, 0.19] Open in a new tab Note: Logistic regression with cluster-robust standard errors (clustered by participant). N = 4,080 observations; Clusters = 240 participants. Panel A log-likelihood = − 1890.43; McFadden pseudo-R² = 0.326. Panel B log-likelihood = − 1955.73; McFadden pseudo-R² = 0.303. In Panel B, Synergy and Cost are coded 0–2 (reported per + 1 level); other attributes are coded 0 = Low and 1 = High OR Odds ratio , CI Confidence interval The survey was administered online. All respondents completed the same 17 vignettes in a fixed sequence, and two dominance-check items were embedded to assess comprehension and attention. Because the number of tasks was limited and the pilot suggested an acceptable response burden, blocking was not used. Fixed-order presentation may still introduce fatigue or anchoring effects, and this possibility is acknowledged as a limitation. However, because the selected vignette set was near-orthogonal, item order was unlikely to be systematically aligned with particular attribute combinations. An information sheet appeared on the first page of the survey. Participation was voluntary, the survey was anonymous, no direct identifiers were collected, and proceeding to the questionnaire was taken as indication of informed consent. Pilot study and instrument refinement Before the main survey, we conducted an online pilot study to assess feasibility, clarity, and preliminary directional validity. Using purposive recruitment, 60 full-time faculty members with teaching responsibilities in medical education, including 30 basic science faculty and 30 clinical faculty, participated from one medical university in Chongqing. The pilot had four functions. First, it was used to assess whether the scenario descriptions and response tasks were understandable. Second, it was used to evaluate whether the number of tasks created an acceptable response burden. Third, it was used to confirm that the dominance-check vignettes functioned as intended. Fourth, it was used to examine whether the direction of the estimated attribute effects was broadly consistent with theoretical expectations. The pilot supported the practical feasibility of the instrument. Respondents were able to complete the task without excessive burden, the dominance checks behaved as expected, and the preliminary attribute directions were consistent with the intended interpretation of the instrument. Based on pilot feedback, the wording of the collaboration cost attribute was refined to improve perceptual discriminability across levels, resulting in the final three-level workload specification used in the main survey. Taken together, the pilot and refinement process provided face-validity, comprehension, and initial construct-supporting evidence for the finalized scenario-based choice instrument. Main survey and sampling The main survey used purposive sampling to recruit full-time medical faculty from four universities in three cities in China. This sampling strategy was adopted because the study aimed to identify institutionally relevant cooperation mechanisms in a theoretically specified faculty population rather than to estimate population prevalence in a representative national sample. Eligible participants were full-time faculty members who had active teaching responsibilities in medical education during the previous academic year and who could meaningfully interpret cross-group SoTL collaboration scenarios. Recruitment was coordinated through teaching affairs offices, faculty development centers, and departmental communication groups. Participation was voluntary, no monetary incentive was provided, and the survey was administered online during September–December 2024. A total of 240 faculty members entered the formal survey sample, including 100 basic science faculty and 140 clinical faculty. The group sizes were not fixed as equal quotas in advance and reflect the realized availability of eligible respondents reached through institutional recruitment channels during the survey period. Among recruited participants, all completed the questionnaire. Because invitation reach and nonresponse at the recruitment stage were not systematically tracked, this completion rate should be interpreted only within the recruited sample rather than as a population response rate. The sample size was determined pragmatically for a multi-site scenario-based discrete-choice experiment, balancing information yield, estimation of main and interaction effects, and respondent burden; with 240 participants completing 17 vignettes, the study yielded 4,080 observations for clustered logit estimation. No formal a priori power calculation was used because the study aimed at mechanism identification across multiple experimentally varied attributes rather than inference on a single prespecified effect size. Data structure and quality assurance The analytic dataset was structured in long format, with one observation per respondent-vignette pair. The primary analytic dataset was defined by the binary cooperation choice for each respondent–vignette pair. In addition to disciplinary group, the survey also collected background information on academic rank, years of teaching experience, and prior SoTL collaboration experience; these variables are reported descriptively in Table 1 to characterize the sample rather than to serve as the primary explanatory variables in the main models. Because each participant evaluated multiple scenarios, the observations constitute repeated measures clustered within respondents. All regression models therefore used respondent-clustered robust standard errors.Data quality was assessed in three ways: response completeness, dominance-check performance, and consistency of the overall response pattern with the expected design logic. All respondents who entered the formal survey sample completed all vignette tasks, and the embedded dominance checks did not indicate systematic misunderstanding or widespread satisficing. These quality indicators support the usability of the final dataset, although they do not eliminate the possibility of residual fatigue or anchoring effects associated with fixed-order presentation. Table 1. Sample structure and design summary Panel A. Sample and data structure Item Overall Group A (Basic science) Group B (Clinical) Respondents (clusters) 240 100 140 Vignettes per respondent 17 17 17 Observations (respondent×vignette) 4,080 1,700 2,380 Mean cooperation rate (choice = 1) 0.448 0.4371 0.4571 Panel B. Sample characteristics (Overall , n = 240) Characteristic Category n % Academic rank Teaching assistant 7 2.92 Lecturer 40 16.67 Associate professor 108 45.00 Professor 85 35.42 Years of teaching 1–5 7 2.92 6–10 51 21.25 11–20 106 44.17 > 20 76 31.67 Prior SoTL collaboration Yes 105 43.75 No 135 56.25 Panel C. Attribute coding and empirical distribution in the vignette design Attribute Coding in estimation Levels Distribution in data (n , %) Synergy Slevel coded 0,1,2 (ordered) Low, Medium, High 0: 1200 (29.4%); 1: 1200 (29.4%); 2: 1,680 (41.2%) Cost Costlevel coded 0,1,2 (ordered) Low ≈ 1 week; Medium ≈ 1 month; High ≈ 1 semester 0:1200 (29.4%); 1:1,440 (35.3%); 2:1440 (35.3%) Inequity risk IneqRisk coded 0,1 0 = Low (rules clear and traceable); 1 = High (rules unclear unfairness) 0:2,160 (52.9%); 1:1,920 (47.1%) Direct reciprocity DR coded 0,1 0 = Low; 1 = High(credible commitment) 0:2,160 (52.9%); 1: 1,920 (47.1%) Reputational visibility IR coded 0,1 0 = Low (hard to record); 1 = High (recorded and credited in evaluation) 0:1,920 (47.1%); 1: 2,160 (52.9%) Open in a new tab Threshold analysis framework To translate the discrete-choice results into an interpretable institutional decision metric, we developed a cooperation-threshold framework based on expected utility comparisons. The threshold (S^*) represents the minimum expected synergy required for cooperation to become incentive-compatible under a given combination of institutional conditions. In substantive terms, a lower threshold means that cooperation becomes feasible under less favorable expected gains, whereas a higher threshold means that stronger expected synergy is required before cooperation becomes attractive. Expected net advantage of cooperation We model faculty cooperation decisions as comparisons of expected utilities. When a faculty member considers cooperating in a SoTL project, their expected net advantage depends on their belief about the counterpart’s behavior: 2 where denotes the baseline belief that the counterpart will cooperate, and , , , and denote utilities under mutual cooperation, unilateral cooperation, unilateral defection, and mutual defection, respectively. Payoff structure from vignette attributes We express utilities as functions of the vignette attributes. Under mutual cooperation, 3 If the respondent cooperates while the counterpart defects, 4 reflecting the absence of synergy and direct reciprocity, the inequity penalty when exploited, and reputational credit that still accrues with participation. For simplicity, we normalize defection outcomes to zero: Cooperation threshold derivation A faculty member chooses cooperation when the expected net advantage of cooperating over defecting is nonnegative ( ). Substituting Eqs. (3) and (4), we obtain: 5 Solving for the minimum synergy level that makes cooperation incentive-compatible (setting and solving for ) yields the cooperation threshold: 6 In Eq. (6), reputational visibility enters the numerator without belief-weighting because recognition accrues when the respondent chooses cooperation, irrespective of partner behavior. By contrast, synergy and direct reciprocity are realized only under mutual cooperation and therefore appear in belief-weighted terms. A negative threshold ( ) indicates that cooperation is incentive-compatible even for the lowest-synergy vignettes ( ); the design does not include negative synergy levels. When (the maximum synergy level in our vignette design), cooperation is not incentive-compatible under the given conditions. Sensitivity analysis design To assess institutional leverage, we conducted ceteris paribus threshold sensitivity analyses by varying one institutional mechanism from 0 (low) to 1 (high) while holding all remaining attributes at baseline ( = 1; non-varied mechanisms set to 0). Using group-specific calibrated parameters, we computed the resulting change in the cooperation threshold for each mechanism (Supplementary Table S4). Sensitivity analysis visualizations were generated using MATLAB R2022b, and reproducibility code for the underlying threshold calculations is provided in Supplementary Materials Sect. 4 . Mapping logit coefficients to game-theoretic payoffs The interaction logit model estimates how vignette attributes shift the log-odds of choosing cooperation. However, the threshold framework requires interpretable payoff weights. We therefore used a belief-based transformation to map the regression coefficients onto game-theoretic utility parameters. This mapping should be understood as a calibrated interpretive bridge rather than as a claim that logit coefficients are directly observed utility units. Baseline belief operationalization We operationalize baseline cooperation beliefs as the empirical cooperation rate observed in each group’s discrete-choice data: 7 where is the number of respondents in group , and indicates whether respondent cooperated in scenario . This operationalization treats baseline beliefs as group-level priors that reflect typical cooperation tendencies, thereby grounding the belief term in observed behavior rather than arbitrary assumptions. These beliefs provide a transparent anchor for the coefficient-to-payoff mapping and for the threshold calculations. Robustness to alternative belief specifications is reported in Supplementary Table S9. We make two simplifying assumptions. First, baseline beliefs are homogeneous within groups (all members of group share the same ). Second, the threshold derivation assumes an additive utility structure (attribute and mechanism effects enter linearly). Coefficient-to-payoff transformation The logit model estimates the change in the log-odds of choosing cooperation (relative to defection) associated with a one-unit increase in attribute X for group g; we denote this coefficient by . To express these coefficients on the payoff/utility scale used in the threshold model, we map to the corresponding utility weights by matching how each attribute enters the expected net advantage . Let : denote the baseline belief that the counterpart cooperates. Then the payoff/utility weights are obtained as: 8 Intuitively, attributes that materialize only under mutual cooperation (S and DR) are rescaled by , the inequity penalty applies only when the respondent cooperates but the counterpart defects (probability 1- ), and unconditional effects (IR and Cost) are not belief-weighted. Model validation strategy We evaluate whether the calibrated threshold model reproduces observed cooperation patterns in the vignette data. For each group and vignette , we compute the implied cooperation threshold using the calibrated parameters and the closed-form threshold expression. Given the vignette’s synergy level , the model predicts cooperation when and defection otherwise. We summarize predictive performance in two complementary ways: (i) individual-level prediction accuracy (the share of respondent–vignette choices correctly classified), and (ii) the Spearman correlation between model-predicted and empirically observed cooperation rates across the 17 scenarios within each group. This threshold-based check provides a direct behavioral validation of the coefficient-to-payoff mapping beyond conventional in-sample fit statistics. Ethics All procedures performed in this study were in accordance with the 1964 Helsinki Declaration and its later amendments or comparable ethical standards. This study has been approved by the Institutional Review Board of Army Medical University (Approval No. AMU20240522). According to the “Ethical Review Measures for Life Sciences and Medical Research Involving Humans” (National Health Commission, Ministry of Education, Ministry of Science and Technology, and National Administration of Traditional Chinese Medicine of the People’s Republic of China, No. 4 [2023]), this study meets the exemption criteria specified in Article 32 of the regulation. Therefore, the following ethical review committees granted exemption: Chongqing Medical University Ethical Review Committee (Exemption approval date: June 10, 2024), Naval Medical University Ethical Review Committee (Exemption approval date: June 21, 2024), and Air Force Medical University Ethical Review Committee (Exemption approval date: June 28, 2024). The specific exemption criteria were as follows: the study did not cause harm to participants and did not involve commercial interests; and the data used in this study were fully anonymized and did not contain any personal identifiers or sensitive information. The experiment was explained to all participants, allowing for questions or comments, and informed consent was obtained from all participants. Data and identity were kept confidential and used only for the purpose of this study. All procedures were performed in accordance with relevant guidelines and regulations. Results Sample characteristics and validity checks The analytic sample comprised 240 full-time faculty members from four medical universities in Chongqing, Xi’an, and Shanghai, including 100 basic science faculty (Group A) and 140 clinical faculty (Group B). Each respondent completed 17 vignette tasks, yielding 4,080 respondent-vignette observations. Data-quality checks, including dominance checks, supported the usability of the dataset (Sect. 2.3 ). Because the survey was administered through institutional channels and used a fixed vignette order, the data should be interpreted as structured stated-preference evidence rather than as a fully naturalistic measure of real-world collaboration behavior. Table 1 summarizes the sample structure and respondent characteristics. The mean empirical cooperation rate was 0.4371 among basic science faculty and 0.4571 among clinical faculty (Table 1 ). These group-specific cooperation rates were subsequently used to calibrate the baseline belief term in the threshold analysis. In addition to disciplinary group, the survey collected background information on academic rank, teaching experience, and prior SoTL collaboration experience. These variables are reported descriptively to characterize the sample rather than to imply population representativeness or to function as the primary explanatory variables in the main models. Importantly, the prior SoTL collaboration item captures respondents’ general prior collaboration experience, rather than dyad-specific prior working history with the prospective collaborator described in a vignette. The full vignette design matrix and attribute correlation checks are provided in Supplementary Tables S1–S2. Logit estimates of cooperation decisions We first estimated a pooled main-effects logit model to characterize the average association between vignette attributes and the probability of choosing cooperation. We then estimated an interaction model to examine whether attribute sensitivity differed between basic science faculty and clinical faculty. Finally, we fit group-stratified models as a robustness check. All models used respondent-clustered robust standard errors because each participant evaluated multiple vignettes. Main-effects logit: average drivers of cooperation In the primary ordinal-linear specification, all five vignette attributes were significantly associated with the odds of choosing cooperation (Table 2 , Panel B). Higher expected synergy was associated with higher odds of cooperation, whereas higher collaboration cost was associated with lower odds of cooperation. Perceived inequity risk showed a strong negative association with cooperation choice, indicating marked aversion to scenarios in which respondents might contribute without receiving fair reciprocal effort or credit. Both reciprocity mechanisms were positively associated with cooperation, with direct reciprocity showing the largest average association and reputational visibility showing the second largest.The robustness specification using categorical coding for synergy and cost yielded a substantively similar pattern (Table 2 , Panel A). In particular, the high-cost condition remained clearly deterrent, whereas the medium-cost contrast was comparatively small, suggesting that the negative cost effect was concentrated at the higher end of the workload spectrum. Because the ordered specification captured the overall monotonic gradient more parsimoniously and remained consistent with the threshold framework, it was retained as the primary model.These coefficients should be interpreted as structured associations in stated-preference scenarios rather than as direct estimates of real-world policy effects; reassuringly, the alternative categorical specification did not materially change the substantive incentive pattern. Heterogeneity: interaction logit The interaction model indicated clear group-differentiated response patterns (Table 3 ). Relative to basic science faculty, clinical faculty were significantly more responsive to direct reciprocity (DR × groupB: β = 1.348, p < 0.001) and significantly less responsive to reputational visibility (IR × groupB: β = -1.722, p < 0.001). Clinical faculty also showed stronger sensitivity to inequity risk (IneqRisk × groupB: β = -0.460, p = 0.039), whereas cost appeared to be a slightly weaker deterrent in this group (Costlevel × groupB: β = 0.182, p = 0.029). By contrast, the interaction for synergy was not statistically significant (Slevel × groupB: β = -0.126, p = 0.309), suggesting that both groups responded in broadly similar ways to expected collaboration gains. Table 3. Interaction logit estimates Variable β SE(cluster) z p OR 1.081 0.106 10.22 < 0.001 2.95 -0.486 0.068 -7.16 < 0.001 0.62 -1.638 0.174 -9.40 < 0.001 0.19 DR 1.326 0.113 11.70 < 0.001 3.77 IR 2.570 0.167 15.41 < 0.001 13.07 groupB 0.587 0.207 2.84 0.004 1.80 × groupB -0.126 0.123 -1.02 0.309 0.88 × groupB 0.182 0.083 2.18 0.029 1.20 × groupB -0.460 0.223 -2.06 0.039 0.63 DR × groupB 1.348 0.191 7.05 < 0.001 3.85 IR × groupB -1.722 0.194 -8.86 < 0.001 0.18 Constant -2.301 0.180 -12.82 < 0.001 0.10 Open in a new tab Model fit: N = 4,080; clusters = 240; Log-likelihood=-1825.84; McFadden Pseudo Taken together, these interaction estimates suggest that the two faculty groups differ less in whether supportive mechanisms matter than in which mechanisms matter most. Under the structured scenarios used here, clinical faculty were particularly responsive to credible mutual commitment, whereas basic science faculty were more responsive to institutionally visible recognition. Group-stratified models The group-stratified models closely mirrored the interaction-model results. In both groups, higher synergy was positively associated with cooperation and higher cost was negatively associated with cooperation. Inequity risk reduced the likelihood of choosing cooperation in both groups. However, the relative salience of the two reciprocity-related mechanisms differed by faculty role: reputational visibility was substantially stronger among basic science faculty, whereas direct reciprocity was substantially stronger among clinical faculty.These stratified results support the substantive interpretation of the interaction model and reinforce the conclusion that institutional supports are likely to have different leverage across faculty groups rather than a uniform effect across the entire sample. Calibrated behavioral parameters from interaction model Parameter estimation Using the belief-based mapping in Sect. 2.5 and the group-specific empirical cooperation rates reported in Table 1 , we derived the calibrated behavioral parameters shown in Table 4 . For the four most policy-relevant parameters, the calibrated direct reciprocity weight was 3.034 (95% CI: 2.513, 3.554) for basic science faculty and 5.850 (95% CI: 5.147, 6.553) for clinical faculty, whereas the calibrated reputational visibility weight was 2.570 (95% CI: 2.242, 2.898) for basic science faculty and 0.848 (95% CI: 0.652, 1.044) for clinical faculty. These interval estimates are consistent with the stronger reciprocity sensitivity observed among clinical faculty and the stronger reputational-visibility sensitivity observed among basic science faculty in the interaction model. Table 4. Calibrated Behavioral Parameters by Faculty Group Parameter Basic Science (Group A) Clinical (Group B) Interpretation 2.473 2.090 Synergy weight (utils per level under mutual cooperation) 0.486 0.304 Cost weight (utils per level) 3.034 5.850 Direct reciprocity weight 2.570 0.848 Reputational visibility weight 2.911 3.866 Inequity aversion penalty (when exploited) 0.437 0.457 Baseline cooperation belief Open in a new tab Note: All parameters derived using the transformation formulas in Sect. 2.5.2 . Standard errors computed via delta method; 95% confidence intervals in Supplementary Table S3 Model validation The calibrated threshold model reproduced the observed choice patterns reasonably well, with prediction accuracies of 78.3% for basic science faculty and 76.9% for clinical faculty and strong scenario-level Spearman correlations across the 17 scenarios. These values substantially exceeded the naïve majority-class benchmark of 56.3% and 54.3%, respectively. Behavioral contrasts between faculty groups The calibrated parameters suggest three substantive contrasts between the two faculty groups. First, clinical faculty placed greater relative weight on direct reciprocity, indicating that credible mutual commitment had stronger leverage in this group under the calibrated model. Second, basic science faculty placed greater relative weight on reputational visibility, indicating that institutionally visible recognition was relatively more salient for this group. Third, both groups showed substantial inequity aversion, with somewhat stronger calibrated sensitivity among clinical faculty.These contrasts should be interpreted as differences in model-implied mechanism salience rather than as fixed psychological traits. Substantively, they suggest that a uniform institutional strategy may leave important faculty-specific barriers unaddressed. Cooperation thresholds and institutional mechanism sensitivity Using the calibrated parameters in Table 4 , we quantified how institutional mechanisms shift the minimum expected synergy required for cooperation to become incentive-compatible. In all threshold calculations, collaboration cost was fixed at the medium level. We then varied one mechanism at a time from low to high while holding the remaining mechanisms at their baseline settings. Figure 1 displays these ceteris paribus threshold shifts for reputational visibility, direct reciprocity, and inequity risk, together with a summary panel.Because synergy was coded on a 0–2 ordinal scale corresponding to low, medium, and high levels, the interpretation of the threshold is straightforward. A negative threshold indicates that cooperation would already be incentive-compatible at the lowest synergy level represented in the vignette design. By contrast, a threshold above 2 indicates that even the highest synergy level represented in the design would be insufficient to make cooperation incentive-compatible under those conditions.Under the baseline scenario, the threshold was 0.450 for basic science faculty and 0.318 for clinical faculty, indicating that both groups required positive expected synergy in the absence of additional institutional supports. When reputational visibility was activated, the threshold declined in both groups, with a larger reduction for basic science faculty (to -1.928) than for clinical faculty (to -0.569). When direct reciprocity was activated, the threshold also declined in both groups, but the reduction was larger for clinical faculty (to -2.481) than for basic science faculty (to -0.777). By contrast, high inequity risk shifted the threshold upward in both groups, reaching 1.965 for basic science faculty and 2.515 for clinical faculty, the latter exceeding the highest synergy level represented in the vignette design.These threshold shifts are consistent with the interaction-logit results. Under the calibrated model, reputational visibility functions as a stronger threshold-reducing mechanism for basic science faculty, whereas direct reciprocity functions as a stronger threshold-reducing mechanism for clinical faculty. Inequity risk acts in the opposite direction, raising the cooperation threshold in both groups and, for clinical faculty, pushing cooperation beyond the feasible range of the present vignette design.The mechanism-specific interpretation follows directly from the structure of the threshold equation. Direct reciprocity reduces the threshold through a belief-weighted mutual-cooperation channel, which helps explain its strong leverage among clinical faculty. Reputational visibility reduces the threshold through an unconditional cooperative-return channel, which helps explain its stronger leverage among basic science faculty. Inequity risk raises the threshold because it operates as an exploitation penalty attached to unilateral cooperation. In practical terms, supportive mechanisms lower the minimum gains needed to justify cooperation, whereas inequity concerns increase them. Fig. 1. Open in a new tab Cooperation thresholds (S*) across institutional mechanisms and faculty groups. Group A = basic science faculty; Group B = clinical faculty. In all panels, thresholds are computed under baseline conditions (Costlevel = 1), with non-varied mechanisms held at 0. Negative S^*values indicate cooperation is incentive-compatible even at the lowest synergy level (Slevel = 0), whereas values > 2 exceed the vignette feasibility boundary. A Sensitivity of S^* to perceived inequity risk (IneqRisk): low (IneqRisk = 0) vs high (IneqRisk = 1). Group A: 0.450 → 1.965; Group B: 0.318 → 2.515 (exceeds the feasibility bound at S⁎ = 2). B Sensitivity of S^* to reputational visibility (IR): low (IR = 0) vs high (IR = 1). Group A: 0.450 → −1.928; Group B: 0.318 → −0.570. C Sensitivity of S^* to direct reciprocity (DR): low (DR = 0) vs high (DR = 1). Group A: 0.450 → −0.777; Group B: 0.318 → −2.481. D Summary of S^* under the baseline setting and under each mechanism activated to 1 (IR, DR, or IneqRisk), with other mechanisms held at baseline (Costlevel = 1). Labels report S^* and, in parentheses, the absolute difference from baseline (ΔS^*) and the percent change within each group. Sample: 240 participants (100 basic science; 140 clinical), 4,080 respondent–vignette observations Discussion Principal findings This study used a scenario-based discrete-choice experiment and a calibrated threshold framework to examine how institutional mechanisms may shape the feasibility of cross-group SoTL collaboration between basic science faculty and clinical faculty. Three main findings emerged. First, institutional supports appeared to have group-differentiated leverage rather than uniform influence. Under the calibrated model, reputational visibility reduced the cooperation threshold more strongly for basic science faculty, whereas direct reciprocity reduced the threshold more strongly for clinical faculty. This pattern suggests that collaboration barriers are not identical across faculty roles and that the same institutional lever may not have the same practical salience for different groups. Second, perceived inequity risk emerged as a strong deterrent in both groups. Under high inequity risk, the threshold rose substantially for both faculty groups and exceeded the highest synergy level represented in the vignette design for clinical faculty. This result does not establish that inequity concerns will always block real-world collaboration, but it does suggest that perceptions of unfair workload or credit allocation may materially undermine the cooperation conditions modeled here. Third, the threshold framework provides a more policy-relevant interpretation of the discrete-choice results than odds ratios alone. Instead of indicating only whether an attribute is positively or negatively associated with cooperation, the threshold results show how much expected collaborative value is needed before cooperation becomes incentive-compatible under different institutional conditions. In that sense, the analysis helps translate preference structures into conditional design implications. These findings should therefore be interpreted as evidence about mechanism salience under structured stated-preference conditions, not as direct causal estimates of policy effects in real institutional settings. Conditional implications for institutional design The present findings support three conditional implications for institutional design in similar settings. Reputational visibility may be particularly salient for basic science faculty Under the calibrated model, institutionally visible recognition produced the largest threshold reduction for basic science faculty. One interpretation is that faculty working in more research-oriented academic environments may be especially responsive to signals that educational scholarship is visible, auditable, and valued in evaluation processes. In practical terms, this could include explicit recognition of SoTL contributions in promotion criteria, formal documentation of educational scholarship activities, and high-visibility awards or reporting mechanisms tied to teaching-related scholarly work [ 27 ]. Credible reciprocity structures may be particularly salient for clinical faculty For clinical faculty, direct reciprocity produced the largest threshold reduction. This pattern is consistent with the idea that when time is scarce and clinical obligations are substantial, cooperation may be more attractive when mutual commitment is credible and role expectations are clearly defined. Practical responses may include explicit collaboration agreements, shared deliverables, paired leadership structures, and protected-time arrangements that make reciprocal contribution more believable and enforceable. Fairness safeguards may function as enabling conditions Across both groups, inequity risk raised the cooperation threshold and offset the benefits of supportive mechanisms. This does not mean that fairness safeguards are universally sufficient or always necessary in a strict causal sense. Rather, the calibrated model suggests that transparent workload allocation, explicit authorship or credit agreements, and evaluation systems that appropriately recognize different forms of contribution may serve as enabling conditions for collaboration in this context [ 28 , 29 ]. Theoretical contributions This study makes three contributions to the literature on SoTL collaboration. First, it introduces a threshold-based interpretive framework that converts stated-preference estimates into an explicit cooperation-feasibility condition. Rather than showing only whether an institutional factor increases or decreases willingness to collaborate, this framework indicates the minimum expected synergy required for cooperation under a given institutional configuration. Under the calibrated baseline setting, for example, cooperation becomes incentive-compatible at a lower synergy threshold for clinical faculty than for basic science faculty, illustrating how the same collaborative opportunity may be evaluated differently across faculty roles. Second, the study refines the literature on institutional support in SoTL by disaggregating support into separable behavioral mechanisms, including expected gains, cost, reciprocity, fairness, and reputational visibility. In this sense, institutional support is not treated as a broad background condition, but as a set of distinct incentive components that shape collaborative choice through different pathways. Third, the study extends SoTL scholarship conceptually by framing cross-group educational collaboration as a problem of strategic interdependence embedded in role-specific incentive environments. This perspective links micro-level decision logic to meso-level institutional design and helps explain why faculty in different roles may respond differently to the same collaborative conditions. Limitations and future directions This study has several limitations. First, the data are based on stated preferences in hypothetical scenarios rather than observed behavior under real institutional conditions. The findings should therefore be interpreted as evidence about mechanism salience and modeled cooperation feasibility, not as direct estimates of behavioral effects following policy implementation. Second, the study was conducted in four Chinese medical universities and used purposive recruitment through institutional channels. Although this design was appropriate for mechanism identification in a clearly defined faculty population, it limits representativeness and external generalizability. The sample may overrepresent faculty who were more engaged with educational work or more willing to respond to institutionally coordinated recruitment. Third, several simplifying modeling assumptions were required. Baseline beliefs were calibrated at the group level and treated as homogeneous within groups. The threshold framework assumed additive contributions of the institutional mechanisms and therefore did not estimate possible interaction effects such as complementarity between reciprocity and recognition. In addition, the binary coding of institutional mechanisms supports discrete low-versus-high comparisons but does not allow dose-response interpretation. Fourth, the instrument used a fixed vignette order. Although the design was near-orthogonal and the dominance checks supported response quality, future studies would be strengthened by randomized vignette presentation. The inequity-risk manipulation also represents a simplified proxy for multidimensional fairness perceptions and should not be interpreted as exhausting all relevant forms of perceived unfairness in faculty collaboration. Fifth, some potentially relevant determinants were not incorporated into the main model. These include prior collaboration history, self-efficacy regarding SoTL work, and topic-specific fit between prospective collaborators. Such factors may influence baseline trust, perceived feasibility, and willingness to participate, and they warrant explicit examination in future research. Sixth, although the instrument was theory-driven and refined through pilot testing, we did not conduct formal expert-panel content validation or a priori power calculation. These choices were consistent with a scenario-based discrete-choice design rather than a reflective scale, but they still limit how fully validation and sample-size justification can be documented. Future work should therefore pursue at least three directions: replication in other institutional and cultural settings, field-based validation of whether these stated-preference patterns predict observable collaboration behavior, and expanded designs that estimate mechanism interactions and incorporate additional individual-level psychological and relational variables. Group A = basic science faculty; Group B = clinical faculty. In all panels, thresholds are computed under baseline conditions (Costlevel = 1), with non-varied mechanisms held at 0. Negative values indicate cooperation is incentive-compatible even at the lowest synergy level (Slevel = 0), whereas values > 2 exceed the vignette feasibility boundary. (A) Sensitivity of to perceived inequity risk (IneqRisk): low (IneqRisk = 0) vs. high (IneqRisk = 1). Group A: 0.450 → 1.965; Group B: 0.318 → 2.515 (exceeds the feasibility bound at S⁎ = 2). (B) Sensitivity of to reputational visibility (IR): low (IR = 0) vs. high (IR = 1). Group A: 0.450 → −1.928; Group B: 0.318 → −0.570. (C)Sensitivity of to direct reciprocity (DR): low (DR = 0) vs. high (DR = 1). Group A: 0.450 → −0.777; Group B: 0.318 → −2.481. (D) Summary of under the baseline setting and under each mechanism activated to 1 (IR, DR, or IneqRisk), with other mechanisms held at baseline (Costlevel = 1). Labels report and, in parentheses, the absolute difference from baseline (Δ ) and the percent change within each group. Sample: 240 participants (100 basic science; 140 clinical), 4,080 respondent–vignette observations. Supplementary Information Supplementary Material 1. (305KB, pdf) Supplementary Material 2. (14.2KB, pdf) Supplementary Material 3. (265.1KB, pdf) Acknowledgements The authors thank all the experts who participated in the study. Authors’ contributions L.C.L. and T.F.R. performed the research and wrote the main manuscript. Y.J.Q. and H.X.F. revised parts of the manuscript. C.J.H. designed the study. Funding This research was supported by the Project of Chongqing Graduate Education Teaching Reform Research (No. yjg243141) and the General Project of Chongqing Education Science “14th Five-Year Plan” 2024 Annual Program (No. K24YG5010043). The article processing charge was provided by Research Project on Innovation in the Education Goverance System (Project No. JYKY-C2025013). Data availability The datasets analyzed during the current study are available from the corresponding author on reasonable request. Reproducibility code for the threshold calculations is provided in Supplementary Materials Sect.  4. Declarations Ethics approval and consent to participate All procedures performed in this study were in accordance with the 1964 Helsinki Declaration and its later amendments or comparable ethical standards. This study was approved by the Institutional Review Board of Army Medical University (Approval No. AMU20240522). According to the Ethical Review Measures for Life Sciences and Medical Research Involving Humans (National Health Commission, Ministry of Education, Ministry of Science and Technology, and National Administration of Traditional Chinese Medicine of the People’s Republic of China, No. 4 [2023]), this study met the exemption criteria specified in Article 32. Accordingly, exemption was also granted by the Ethical Review Committee of Chongqing Medical University (June 10, 2024), the Ethical Review Committee of Naval Medical University (June 21, 2024), and the Ethical Review Committee of Air Force Medical University (June 28, 2024). The exemption criteria were that the study posed no harm to participants, involved no commercial interests, and used fully anonymized data without personal identifiers or sensitive information. The study was explained to all participants, opportunities were provided for questions and comments, and informed consent was obtained from all participants. Data and identities were kept confidential and used only for the purposes of this study. All procedures were performed in accordance with relevant guidelines and regulations. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Luchuan Li and Tianfei Ran contributed equally to this work. References 1. Boyer EL. Scholarship reconsidered: priorities of the professoriate. Princeton (NJ): Carnegie Foundation for the Advancement of Teaching; 1990. [ Google Scholar ] 2. Hutchings P, Shulman LS. The scholarship of teaching: new elaborations, new developments. Change. 1999;31(5):10–5. [ Google Scholar ] 3. Huber MT, Hutchings P. The advancement of learning: building the teaching commons. San Francisco (CA): Jossey-Bass; 2005. [ Google Scholar ] 4. Brew A, Ginns P. The relationship between engagement in the scholarship of teaching and learning and students’course experiences. Assess Eval High Educ. 2008;33(5):535–45. 10.1080/02602930701698959. [ Google Scholar ] 5. Walpole SC, Mortimer F. Evaluation of a collaborative project to develop sustainable healthcare education in eight UK medical schools. Public Health. 2017;146:93–9. 10.1016/j.puhe.2017.01.021. [ DOI ] [ PubMed ] [ Google Scholar ] 6. Bhattacharjee R, Reynolds A, Zhan L, Knittig L, Nagaraj R, Zhao Y. Developing a framework for promoting interest and engagement of scholarship of teaching and learning for medical students. Med Educ Online. 2024;29(1):2336332. 10.1080/10872981.2024.2336332. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 7. Ahmed Y, Taha MH, Khayal S. Integrating research and teaching in medical education: challenges, strategies, and implications for healthcare. J Adv Med Educ Prof. 2024;12(1):1–7. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 8. Cofrancesco J, Barone MA, Serwint JR, Goldstein M, Westman M, Lipsett PA. Development and implementation of a school-wide institute for excellence in education to enable educational scholarship by medical school faculty. Teach Learn Med. 2018;30(1):103–11. 10.1080/10401334.2017.1339097. [ DOI ] [ PubMed ] [ Google Scholar ] 9. Mendia-Urrutia A, Rekalde-Rodríguez I, Cruz-Iglesias E. Keys to constructing an academic development support strategy based on the Scholarship of Teaching and Learning (SoTL). Innov Educ Teach Int. 2025. 10.1080/14703297.2025.2530117. [ Google Scholar ] 10. Cook DA. Getting started in medical education scholarship. Keio J Med. 2010;59(3):96–103. 10.2302/kjm.59.96. [ DOI ] [ PubMed ] [ Google Scholar ] 11. Xu J, Xu N, Zhang F, Wang H, Li H, Yuan L. Research on the indicator system of scholarship of teaching and learning ability for medical teachers in the context of new medical science—based on the grounded theory method. BMC Med Educ. 2025;25:976. 10.1186/s12909-025-07612-x. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 12. Cameron MW, Crowther LN, Huang GC, et al. Faculty development and infrastructure to support educational scholarship: a scoping review of author development programs in academic medicine. Acad Med. 2023;98(1):112–22. 10.1097/ACM.0000000000004896. [ DOI ] [ PubMed ] [ Google Scholar ] 13. Kochhar K, Longtin K, Wilson S, Ho M, Brokaw J, Wallach P. Implementation of Faculty Learning Communities to Support Medical Education Scholarship in a Regional Campus System. J Reg Med Campuses. 2023;6(2). 10.24926/jrmc.v6i2.4886. 14. Amorelli D, Parker JM. SoTL as a catalyst for the development of medical educator scholar identity. New Dir Teach Learn. 2024;17817–24. 10.1002/tl.20585. 15. Persaud H. Barriers to scholarship among health profession faculty at a public institution of higher learning. Med Sci Educ. 2024;34(2):413–20. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 16. Smesny AL, Williams JS, Brazeau GA, Weber RJ, Matthews HW, Das SK. Barriers to scholarship in dentistry, medicine, nursing, and pharmacy practice faculty. Am J Pharm Educ. 2007;71(5):91. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 17. Camerer CF. Behavioral game theory: experiments in strategic interaction. Princeton (NJ): Princeton University Press; 2003. [ Google Scholar ] 18. Rabin M. Incorporating fairness into game theory and economics. Am Econ Rev. 1993;83(5):1281–302. [ Google Scholar ] 19. Fehr E, Schmidt KM. A theory of fairness, competition, and cooperation. Q J Econ. 1999;114(3):817–68. 10.1162/003355399556151. [ Google Scholar ] 20. Bolton GE, Ockenfels A. ERC: a theory of equity, reciprocity, and competition. Am Econ Rev. 2000;90(1):166–93. 10.1257/aer.90.1.166. [ Google Scholar ] 21. Fehr E, Gächter S. Fairness and retaliation: the economics of reciprocity. J Econ Perspect. 2000;14(3):159–81. 10.1257/jep.14.3.159. [ Google Scholar ] 22. Nowak MA, Sigmund K. Evolution of indirect reciprocity. Nature. 2005;437(7063):1291–8. 10.1038/nature04131. [ DOI ] [ PubMed ] [ Google Scholar ] 23. Li H, Zhou X. Medical education reform in China. Postgrad Med J. 2018;94(1118):673–4. 10.1136/postgradmedj-2018-136145. [ DOI ] [ PubMed ] [ Google Scholar ] 24. Wang W. Medical education in China: progress in the past 70 years and a vision for the future. BMC Med Educ. 2021;21:453. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 25. Guo Y, Yin Y, Zuo Z, Liu W, Elsey C. International medical school faculty development: the results of a needs assessment survey among medical educators in China. Adv Health Sci Educ Theory Pract. 2009;14(1):91–106. [ DOI ] [ PubMed ] [ Google Scholar ] 26. Link AN, Swann CA, Bozeman B. A time allocation study of university faculty. Econ Educ Rev. 2008;27(4):363–74. [ Google Scholar ] 27. Creel A, Paul C, Robyn Bockrath M et al. Promotion Criteria for Medical Educators: Are We Climbing a Ladder with Invisible Rungs? Academic Pediatrics, 2024, 24(4):5. 10.1016/j.acap.2024.01.002 [ DOI ] [ PubMed ] 28. Warsame R, Asiedu GB, Engstrom J, Pruthi S, Shanafelt TD, West CP. Academic promotions in medicine: an appraisal of fairness. Mayo Clin Proc. 2024;99(3):417–26. [ DOI ] [ PubMed ] [ Google Scholar ] 29. Rey-Biel P. Inequity aversion and team incentives. Scand J Econ. 2008;110(2):297–320. 10.1111/j.1467-9442.2008.00540.x. [ Google Scholar ] Associated Data This section collects any data citations, data availability statements, or supplementary materials included in this article. Supplementary Materials Supplementary Material 1. (305KB, pdf) Supplementary Material 2. (14.2KB, pdf) Supplementary Material 3. (265.1KB, pdf) Data Availability Statement The datasets analyzed during the current study are available from the corresponding author on reasonable request. Reproducibility code for the threshold calculations is provided in Supplementary Materials Sect.  4. Articles from BMC Medical Education are provided here courtesy of BMC ACTIONS View on publisher site PDF (2.0 MB) Cite Collections Permalink RESOURCES Similar articles Cited by other articles Links to NCBI Databases Follow NCBI NCBI on X (formerly known as Twitter) NCBI on Facebook NCBI on LinkedIn NCBI on GitHub NCBI RSS feed Connect with NLM NLM on X (formerly known as Twitter) NLM on Facebook NLM on YouTube National Library of Medicine 8600 Rockville Pike Bethesda, MD 20894 Web Policies FOIA HHS Vulnerability Disclosure Help Accessibility Careers NLM NIH HHS USA.gov Back to Top

Related documents

• Navigating Conflicts of Interest in Academia#87400
• Innovation in Medical Education and Clinical Practice : An Exploration of Today’s Revolutionary Concepts, Tools and Applications / edited by Summer S. Chahin, Roger W. Apple, Cheryl A. Dickson, Berrin Ergun-Longmire#1153961
• Getting Started as a Medical Teacher in Times of Change#1017412
• Teaching humanities in UK medical schools: towards community-building and coherence.#625826
• A dimensão afetiva e a docência: investigando possíveis relações com a prática de professores em exercício#376278