CSCS: Culture around Systemic Change Survey

• CSCS is a survey instrument that assesses both current and ideal departmental cultures, emphasizing collaboration, equity, and data-informed change.
• It employs a human-centered design with co-design sessions, interviews, and expert reviews to refine a robust 35-item, five-factor framework.
• Psychometric evaluation in physics departments confirmed a strong general factor with reliable variance explained, supporting its use in diagnosing systemic change.

Searching arXiv for the CSCS paper and closely related framework/theory papers to ground the article with current citations. The Culture around Systemic Change Survey (CSCS) is a survey instrument for assessing culture around systemic change, especially departmental culture related to educational reform in higher education and physics. In its empirically evaluated form, it was developed for faculty and staff to characterize both the “current” state of a department and the “ideal” department they aspire to, with explicit attention to whether change is collaborative, data-informed, sustainable, student-centered, and attentive to systemic injustices (Sachmpazidi et al., 13 Aug 2025). Its intellectual roots lie in a multi-level framework for transforming departmental culture to support educational innovation, which treats the academic department as the key unit of change within a university system comprising faculty, department, and administration (Corbo et al., 2014). A broader theoretical extension also frames CSCS as a way to measure how cultural systems process information, self-organize, and transition under stress, emphasizing coherence-seeking information processing, institutional filtering, path dependence, attractor states, and algorithmic mediation (Jansson, 1 Jan 2026).

1. Origins in systemic change research

The conceptual background for CSCS is the view that sustained improvement in undergraduate education requires systemic educational change rather than isolated reforms. In this framework, universities are complex, interdependent systems, and culture is central to durable change because it consists of shared beliefs, values, rituals, practices, and artifacts that evolve slowly. Systemic educational change is defined as coordinated, multi-level, second-order transformation that aligns structures, incentives, beliefs, rituals, and practices across faculty, departments, and administration to realize a shared vision of research-based, student-centered, equitable education (Corbo et al., 2014).

This orientation emerged partly from dissatisfaction with prior change efforts. The framework reports that most interventions targeted single levels of the university system, such as individual faculty teaching practices or administrative policy, without cross-level alignment. It also emphasizes that change activities were often driven by implicit or contradictory logics. In the synthesis cited there, Henderson et al. (2011) found that 85% of efforts fit one category and that dissemination and policy approaches alone were “clearly not effective” (Corbo et al., 2014). Within this line of argument, CSCS addresses a measurement problem: departments and institutions require instruments that can diagnose whether change is actually supported by the surrounding culture rather than by isolated practices alone.

In physics, this measurement problem became more acute as disciplinary organizations centered systemic change and equity in reform efforts. The empirical CSCS study states that physics programs are being asked to pursue second-order change—deep, cultural transformation that rethinks assumptions, power structures, and norms—rather than piecemeal fixes. It situates the instrument alongside APS-IDEA, EP3 DALI, and AIP TEAM-UP, all of which emphasize equity and justice as central to sustainable reform (Sachmpazidi et al., 13 Aug 2025).

2. Conceptual foundations and change logics

The foundational departmental-change framework organizes intervention around three university levels: faculty, department, and administration. At the faculty level, actors include individual instructors and self-organizing faculty teams; at the department level, actors include the department chair, executive or teaching committees, and whole-faculty governance; at the administration level, actors include deans, the provost, faculty senate, and central offices. The framework argues that durable change depends on coordination across these levels through shared core commitments, aligned incentives, sensemaking routines, data flows, and embedding structures that persist beyond a single project (Corbo et al., 2014).

The same framework draws on six perspectives on change: scientific management, evolutionary, social cognition, cultural, political, and institutional. These perspectives specify distinct leverage points and therefore imply distinct classes of survey constructs. Scientific management foregrounds rewards, resources, expectations, communication, and feedback. Evolutionary perspectives foreground adaptability, coherence, and responses to external pressures. Social cognition focuses on mental maps, sensemaking, and double-loop learning. Cultural perspectives prioritize shared vision, values, rituals, and narratives. Political perspectives highlight coalition-building, power mapping, agenda setting, and negotiation. Institutional perspectives attend to external legitimacy pressures such as accreditation, professional societies, and funding agencies (Corbo et al., 2014).

A concise mapping from these perspectives to CSCS-relevant constructs appears below.

Perspective	Primary leverage points	Example CSCS constructs
Scientific Management	Reward structures, resources, performance measures	Reward alignment, clarity of expectations, access to resources, feedback
Evolutionary	Adaptable structures, coherence, external monitoring	Curriculum coherence, flexibility, readiness to reconfigure structures
Social Cognition	Mental maps, data use, collaborative learning	Shared understanding of goals, sensemaking, professional development
Cultural	Vision, values, rituals, narratives	Shared vision, cultural support, alignment with student-centered and equity commitments
Political	Champions, committees, strategic alliances	Coalitions, influence pathways, conflict resolution, inclusivity in decision-making
Institutional	External networks, legitimacy, standards	Engagement with external networks, influence of accreditation or funders, sustainability beyond grants

This framework also specifies illustrative intervention mechanisms. Departmental Action Teams (DATs) are faculty-level, self-selected teams within a single department organized around a shared educational goal of departmental importance; they are meant to create lasting structures such as curriculum coordinator positions with course buyouts or service credit. The Visioning and Alignment Process (VAP) is a department-level align–act–adjust cycle involving a shared vision, surveys and interviews to elicit mental maps, action plans, and capacities for collaboration and communication. The Teaching Quality Framework (TQF) is an administration-level effort to define and reward teaching excellence in tenure and promotion and to redefine teaching award criteria (Corbo et al., 2014).

A broader theoretical extension reinterprets these culture questions in dynamic-systems terms. In that formulation, culture is not merely a set of traits but a system of interdependent beliefs, practices, and artefacts embedded in cognitive, social, and material structures. Individuals and groups are described as engaging in coherence-seeking information processing with dissonance reduction, while higher-order traits such as goals, skills, norms, and cognitive gadgets operate as “metafilters” that regulate subsequent selection. Epistemic niches, echo chambers, institutional filtering, path dependence, attractors, tension, and punctuated transformations are treated as dynamic properties of such systems (Jansson, 1 Jan 2026). This suggests a substantial broadening of the original departmental-culture conception into a general theory of cultural processing and reorganization.

3. Survey design and substantive dimensions

In the physics-focused instrument, CSCS was built through a Human-Centered Design process consisting of four virtual co-design sessions with 15 stakeholders, ten think-aloud interviews, and expert panel review by five survey experts. Stakeholders included APS and AIP initiative leaders, site-visit leaders, department chairs, APS committees, AIP Research representatives, faculty, and students. Items were revised to improve clarity, remove double-barreled wording, and align with faculty and staff contexts. The pilot instrument contained 50 items administered in Qualtrics on a 7-point Likert-type scale from Strongly disagree to Strongly agree, and each item was rated twice: once for the department’s current state and once for the ideal department (Sachmpazidi et al., 13 Aug 2025).

The original item pool targeted themes that closely track the systemic-change framework: centering students’ voices, partnering with students, advancing equity and inclusion, shared leadership, transparency and accountability of change teams, data-informed decision-making, context-dependence, sustainability, and changing hearts and minds (Sachmpazidi et al., 13 Aug 2025). These themes correspond closely to the earlier framework’s emphasis on research-based, student-centered teaching, equity and diversity, multi-level coordination, explicit change logics, and second-order transformation (Corbo et al., 2014).

Exploratory factor analysis of the current scale yielded a five-factor structure with 35 retained items. The factors were Open-Mindedness (OM), Student Involvement (SI), Collective Interpretation of Evidence (CE), Sustainability (S), and Disruption of Systemic Injustices (DI) (Sachmpazidi et al., 13 Aug 2025).

Factor	Concept
Open-Mindedness (OM)	Willingness to learn, engage differing perspectives, revise thinking, and share rationales for change
Student Involvement (SI)	Formal involvement of students in decision-making, implementation, and co-interpretation of data
Collective Interpretation of Evidence (CE)	Collaborative use of systematic evidence to identify problems, build consensus, and guide action
Sustainability (S)	Planning, monitoring, assessment, documentation, and risk management for sustaining change
Disruption of Systemic Injustices (DI)	Centering marginalized voices, addressing power differentials, and taking action toward justice

Representative item stems clarify the intended content. OM includes “open to revising their thinking,” “engage with differing perspectives,” and “share the reasoning behind the changes made.” SI includes “involve students in decision making,” “involve students in implementing changes,” and “Typically, systematic evidence … is interpreted in collaboration with student representatives.” CE includes “guides collective actions,” “leads to collective decision making,” and “is used to identify the nature of problems in our program(s).” S includes “build assessments into change plans,” “systematically monitor change efforts,” and “build plans for how to overcome potential challenges.” DI includes “ensure that marginalized people have an active role in departmental decision making,” “address power differentials,” and “take action to build a more just system” (Sachmpazidi et al., 13 Aug 2025).

The dynamic-systems extension proposes a much larger subscale architecture. It specifies 12 subscales: Coherence-Seeking, Dissonance Reduction Strategies, Metafilters—Goals, Metafilters—Norms, Metafilters—Skills, Epistemic Niche Orientation, Echo Chamber Insularity, Institutional Filtering, Path Dependence, Attractor Orientation, Tension/Readiness for Transformation, and Algorithmic Mediation. That proposal explicitly includes 7-point Likert items on recommender systems and LLM use, such as dependence on platform recommendations and use of LLMs to recombine information into drafts, briefs, or plans (Jansson, 1 Jan 2026). Because these elements are presented as a proposed operationalization, they are best understood as an extension rather than as the validated factor structure of the physics pilot.

4. Psychometric evaluation in physics departments

The reported psychometric evaluation was conducted on the current scale only, using responses from 111 participants across 33 physics departments. The sampling frame was the AIP list of U.S. physics programs offering a bachelor’s degree, with institutions in APS-IDEA, EP3 DALI, and TEAM-UP excluded, leaving 620 eligible institutions. Thirty-three institutions were randomly selected: 28 PhD-granting and 5 Bachelor’s-only; 20 public and 13 private. Of 129 initial respondents, 4 were removed for very low completion, 1 for careless responding, and 13 because missingness was concentrated on evidence items, yielding a final analytic sample of 111 with 6.0% missing data (Sachmpazidi et al., 13 Aug 2025).

The factorability diagnostics were strong. The overall KMO was 0.89, described as excellent, with

$$\mathrm{KMO} = \frac{\sum_{i\neq j} r_{ij}^2}{\sum_{i\neq j} r_{ij}^2 + \sum_{i\neq j} p_{ij}^2},$$

and Bartlett’s test of sphericity was reported as $\chi^2(1128)=5132.2, p<.001$ (Sachmpazidi et al., 13 Aug 2025). Little’s MCAR test was significant, missingness was judged MAR due to associations with position, and multiple imputation and full-information maximum likelihood produced consistent results. The final exploratory factor analysis used maximum likelihood with FIML in the umx R package, treated Likert items as continuous, used Pearson correlations, and applied oblique promax rotation (Sachmpazidi et al., 13 Aug 2025).

The measurement model was written as

$$\mathbf{x} = \Lambda \mathbf{f} + \boldsymbol{\epsilon},$$

with $\operatorname{Cov}(\mathbf{f}) = \mathbf{I}$ and $\operatorname{Cov}(\boldsymbol{\epsilon}) = \Psi$ (Sachmpazidi et al., 13 Aug 2025). After pruning items for low correlations, multicollinearity, weak loadings, cross-loadings, or low communalities, the final five-factor solution explained 69% of the total variance across 35 items. The variance contributions were OM 11%, SI 19%, CE 14%, S 11%, and DI 14% (Sachmpazidi et al., 13 Aug 2025).

Reliability evidence was strong at the whole-instrument level. Cronbach’s alpha was approximately 0.97, omega-total approximately 0.98, and omega-hierarchical approximately 0.88. Subscale alpha values were approximately 0.89–0.95 and subscale omega-total approximately 0.90–0.95. However, omega-hierarchical for subscales ranged only approximately 0.18–0.40, with OM highest at approximately 0.40 and SI and CE approximately 0.18. The study interprets this pattern as evidence that subscale scores draw heavily on variance from a general factor, so the total score is currently a stronger indicator of “culture around systemic change” than stand-alone subscale scores (Sachmpazidi et al., 13 Aug 2025).

Construct validity evidence included convergent and discriminant validity. Average Variance Extracted values were above 0.50 for all five constructs: OM approximately 0.57, SI approximately 0.65, CE approximately 0.70, S approximately 0.64, and DI approximately 0.65. Squared inter-construct correlations were lower than both constructs’ AVEs for each pair, supporting discriminant validity (Sachmpazidi et al., 13 Aug 2025). At the same time, inter-factor correlations were moderate to strong—for example, SI–CE approximately 0.72, SI–S approximately 0.74, and CE–S approximately 0.73—indicating a coherent general culture-around-systemic-change factor (Sachmpazidi et al., 13 Aug 2025).

The pilot also reported descriptive differences among factors on the 1–7 current scale: OM approximately 4.97, CE approximately 4.85, DI approximately 4.54, SI approximately 4.45, and S approximately 4.44. OM and CE were significantly higher than SI, S, and DI, with medium effects, suggesting stronger receptivity to learning and evidence use than to student partnership, sustainability infrastructure, and explicit justice work (Sachmpazidi et al., 13 Aug 2025). A common misreading would be to treat these gaps as fixed deficits; the study instead recommends using them to identify relative strengths and opportunities.

5. Scoring models, dynamic analysis, and formal extensions

For the validated physics instrument, scoring can proceed by computing the mean or sum of items keyed to each factor, with reverse-keying for negatively valenced items such as “hold stubbornly to their own opinion.” Current–ideal gaps are defined as

$\Delta = \text{Ideal} - \text{Current},$

and can be aggregated by factor to identify priorities. The study also gives a regression-based factor score estimator,

$$\hat{\mathbf{f}} = (\Lambda' \Sigma^{-1} \Lambda)^{-1} \Lambda' \Sigma^{-1} \mathbf{x},$$

and recommends Cohen’s $d$ for gap magnitude (Sachmpazidi et al., 13 Aug 2025).

The broader dynamic-systems extension proposes a different scoring framework. For subscale $k$ with $n_k$ items scored $\chi^2(1128)=5132.2, p<.001$0, the subscale mean is

$\chi^2(1128)=5132.2, p<.001$1

and an overall index is

$\chi^2(1128)=5132.2, p<.001$2

with equal weights for diagnostic use or SEM-derived weights for predictive use (Jansson, 1 Jan 2026). That extension also specifies Cronbach’s alpha, composite reliability, Average Variance Extracted, and the CFA measurement model

$\chi^2(1128)=5132.2, p<.001$3

while recommending multilevel CFA or SEM when individuals are nested in teams or communities (Jansson, 1 Jan 2026).

Its dynamic analysis proposals extend well beyond conventional departmental climate measurement. Longitudinal administration at multiple waves, such as quarterly, is proposed to observe path dependence and attractors. Markov modeling is suggested for transitions among discrete cultural states, with stationarity defined by

$\chi^2(1128)=5132.2, p<.001$4

Tension is proposed to be measured as entropy of state transitions,

$\chi^2(1128)=5132.2, p<.001$5

where higher $\chi^2(1128)=5132.2, p<.001$6 indicates unstable, high-strain dynamics and lower $\chi^2(1128)=5132.2, p<.001$7 indicates settled attractors (Jansson, 1 Jan 2026). If relational data are collected, the proposal further recommends network metrics such as modularity $\chi^2(1128)=5132.2, p<.001$8 and clustering coefficient $\chi^2(1128)=5132.2, p<.001$9 to quantify echo chambers and local reinforcement (Jansson, 1 Jan 2026).

The same extension specifies SEM-ready hypotheses. Coherence-Seeking and Metafilters—Skills are hypothesized to increase Learning Quality, which in turn enhances Change Readiness when tension is high. Dissonance Reduction is hypothesized to increase Echo Chamber Insularity, reducing Learning Quality and slowing Adaptation Speed. Algorithmic Mediation is treated as a mixed mechanism: it may increase insularity via collaborative filtering, but may also increase Learning Quality when used to expand perspectives (Jansson, 1 Jan 2026). These hypotheses are explicitly proposed rather than empirically established in the physics validation study.

6. Use, governance, ethics, and limitations

The principal use of CSCS is diagnostic and developmental rather than punitive. In the physics study, it is presented as a tool to diagnose culture, prioritize interventions, and track progress. Recommended actions are aligned to the five factors: learning communities and structured reflection for OM; standing, paid student advisory boards and formal student roles for SI; routine departmental data reviews and shared dashboards for CE; written change plans, monitoring routines, and documentation repositories for S; and equity audits, psychologically safe mechanisms for voice, and explicit attention to power differentials for DI (Sachmpazidi et al., 13 Aug 2025).

The broader departmental-change framework implies that CSCS results can be embedded into governance routines at three levels. At the faculty level, results can inform feedback loops, professional development, and reward alignment such as service credit and buyouts for DAT participation. At the department level, results can be integrated into faculty meetings and retreats, used to update action plans, and displayed in dashboards tracking progress on core commitments. At the administration level, aggregated results can inform campus priorities, resource allocation, TQF development, and teaching award criteria (Corbo et al., 2014). This suggests that CSCS is most consistent with the underlying theory when used as part of an align–act–adjust cycle rather than as a one-time climate snapshot.

Ethical guidance is a recurring theme across the sources. Respondent protection requires anonymity, especially in small units; results may need to be aggregated or suppressed where $$\mathbf{x} = \Lambda \mathbf{f} + \boldsymbol{\epsilon},$$0 is small. Survey fatigue should be mitigated by limiting survey length, rotating modules, and scheduling appropriately. Data use should be explicitly non-punitive, with results shared in context and linked to action planning rather than sanctions. Equity considerations include capturing differential impacts, monitoring response bias, involving diverse stakeholders in instrument development, and avoiding stigmatization of groups when interpreting properties such as insularity [(Corbo et al., 2014); (Jansson, 1 Jan 2026)].

Several limitations are clearly identified. The psychometric evaluation was conducted only on the current scale; ideal ratings were highly skewed toward agreement, producing ceiling effects that precluded factor analysis at that stage. The pilot sample was modest, with limited responses per department. The subscale $$\mathbf{x} = \Lambda \mathbf{f} + \boldsymbol{\epsilon},$$1 values indicate that the general factor currently dominates subscale-specific variance, so factor scores should be interpreted cautiously in isolation. Future work is identified as confirmatory factor analysis, measurement invariance testing, item refinement, larger and more diverse samples, and broader administration including departments engaged in formal change initiatives (Sachmpazidi et al., 13 Aug 2025).

Taken together, the available work positions CSCS as a family of closely related measurement efforts organized around a shared concern: whether organizational culture supports second-order, equitable, evidence-based change. In its validated physics-department form, it is a human-centered, dual-frame instrument with a strong general factor and a five-factor structure. In its broader theoretical extensions, it also becomes a systems diagnostic for coherence-seeking, institutional filtering, path dependence, and transformation dynamics. The convergence across these formulations is the claim that durable change depends not only on what organizations do, but on the cultural processes through which they interpret evidence, distribute voice, sustain coordination, and redefine what counts as coherent action (Sachmpazidi et al., 13 Aug 2025).