---
schema_version: "1.0.0"
id: "cognitive-resonance-model:en:chapter-13"
work_id: "urn:systemstheology:book:cognitive-resonance-model:chapter:chapter-13"
book_id: "cognitive-resonance-model"
chapter_id: "research-program"
chapter_slug: "chapter-13"
title: "Research Program"
book_title: "Cognitive Resonance Model"
language: "en"
source_language: "en"
translation_status: "source"
authors: ["Systems Theology"]
editorial_owner: "Systems Theology"
editors: []
review_status: "not_specified"
reviewers: []
content_version: "content-0d76206893ba"
content_hash_sha256: "0d76206893ba1dbb58c61664115a755cbc7109f6353a248ad041374ca078ef41"
published_at: "2026-07-15T21:14:45.000Z"
modified_at: "2026-07-15T23:50:19.254Z"
canonical_url: "https://systemstheology.com/library/cognitive-resonance-model/chapter-13/"
markdown_url: "https://systemstheology.com/research/books/cognitive-resonance-model/en/chapter-13.md"
license: "All rights reserved; research use subject to the Use Policy"
license_url: "https://systemstheology.com/use-policy/"
correction_url: "https://systemstheology.com/library/cognitive-resonance-model/chapter-13/#chapter-comments"
---
# Research Program
Researchers, methodologists, and tool builders. move from conceptual clarity to coding reliability, validation, proposed studies, ethics, and revision.
Operational models need disciplined testing. The goal is not to claim clinical authority, but to make the account clear enough to be tested, criticized, and improved.
## Operational Constructs
- Construct | Operational handle
- Incoming reality | Event, testimony, observation, diagnosis, result, repeated pattern, failure, or discovery.
- Factual model | What the person or group expected reality to do.
- Meaning frame | Identity, faith, values, mission, loyalty, hope, worldview, or assumed meaning.
- Prediction error | Degree of mismatch between event and factual model.
- Meaning gap | Degree of mismatch between event and meaning frame.
- Source trust | Perceived and assessed reliability of the channel carrying the signal.
- Capacity | Embodied, relational, clinical, communal, and temporal ability to process the pressure.
- Agency | The truthful response presently available, scaled to source trust, capacity, supports, and constraints.
- Resonance quality | Outcome fruit: reality-contact, humility, repair, confession, protection, wise action, communion, or persistent distortion.
## Coding Manual for Case Studies
The manual below is for research coding, supervised training, and institutional review. It is not a diagnostic instrument. Coders should code only what is present in the case material, mark uncertainty explicitly, and keep the five fields independent. A severe meaning gap does not automatically make the source trustworthy. Low capacity does not make a signal false. A strong action preference does not decide the prediction-error or meaning-gap score.
Coding unit. Code one pressure episode at a time: one incoming signal, one primary expected model, one primary meaning frame, one source channel, one capacity state, and one available action set. If a case contains several distinct signals, split it into separate episodes before assigning codes.
### Coding Order
Coders should work in this order before comparing notes:
- Write the incoming signal in one sentence.
- Identify the failed expectation, if one is present.
- Identify the threatened meaning frame, if one is present.
- Code prediction error and meaning gap separately.
- Code source trust from the channel evidence, not from agreement with the claim.
- Code capacity from safety, embodiment, support, coercion, and clinical risk.
- Code agency as the best available action class under current evidence and capacity.
- Add one edge-case note if the coding was difficult.
### Prediction-Error Anchors
Prediction error (PE) codes the mismatch between the incoming signal and the factual model. It is about expectation failure, not emotional pain by itself.
- Code | Anchor | Code when | Do not code from
- PE0 | No clear breach | No concrete expectation is named or implied, or the event fits the factual model. | Distress, dislike, disagreement, or vague unease.
- PE1 | Mild breach | A small expectation fails, but the model can absorb it with ordinary adjustment. | A large emotion attached to a small factual mismatch.
- PE2 | Material breach | The signal requires real updating, verification, explanation, or correction of the model. | Mere identity threat without a clear factual mismatch.
- PE3 | Severe breach | The signal overturns a central expectation, exposes serious risk, reveals a pattern, or invalidates the working model. | Catastrophic language unsupported by the case evidence.
### Meaning-Gap Anchors
Meaning gap (MG) codes the mismatch between the incoming signal and the meaning frame: identity, faith, loyalty, mission, duty, hope, justice, calling, safety, or love. It is about meaning-pressure, not factual surprise by itself.
- Code | Anchor | Code when | Do not code from
- MG0 | No clear breach | The event has no evident pressure on identity, faith, loyalty, mission, hope, or moral meaning. | Technical correction or ordinary uncertainty.
- MG1 | Mild breach | The meaning frame is unsettled but still intact; the person can name tension without major rupture. | Momentary embarrassment or annoyance alone.
- MG2 | Material breach | The meaning frame needs reinterpretation, lament, repentance, counsel, repair, or a deeper frame. | A factual update that does not touch the meaning frame.
- MG3 | Severe breach | The meaning frame may be shattered, false, idolatrous, abusive, despairing, or too small to hold reality. | Strong doctrine, ideology, or group loyalty unless the case shows meaning-level pressure.
### Source-Trust Coding
Source trust (ST) codes the channel carrying the signal. Coders should look for access, competence, method, corroboration, incentives, records, accountability, and exposure to correction.
- Code | Anchor | Typical indicators
- ST-high | Strong channel trust | Firsthand access or reliable record; competent method; independent corroboration; clear chain of custody; incentives named; open to correction.
- ST-mixed | Partial or contested channel trust | Some access or corroboration, but gaps, conflicts, incomplete records, memory limits, role pressure, or unresolved incentives remain.
- ST-low | Weak channel trust | Hearsay, anonymous claim without support, manipulated context, no method, obvious incentive distortion, AI output without verification, or protected rumor.
- ST-unknown | Insufficient channel data | The case does not give enough information to assess the source responsibly.
### Capacity Coding
Capacity (CAP) codes the receiver's present ability to process the pressure truthfully. It concerns timing, support, safety, and care. It does not decide whether the signal is true.
- Code | Anchor | Typical indicators
- CAP-green | Ordinary reflection is available | Basic safety, enough sleep and attention, no immediate coercion, manageable emotion, and enough support for ordinary processing.
- CAP-yellow | Support or slower timing is needed | High stakes, fatigue, trauma activation, relational pressure, grief, shame, institutional pressure, or limited support.
- CAP-red | Interpretation must yield to safety or qualified help | Active danger, abuse, coercion, self-harm risk, psychosis, medical emergency, severe dissociation, credible threat, or active cover-up.
### Agency Coding
Agency (AG) codes the truthful action available now. It should be scaled to source trust and capacity. Code available action, not merely the action the person prefers.
- Code | Use when the next truthful action is | Common confusion
- AG-adjust | Update an ordinary plan, expectation, schedule, or practice without a major repair process. | Treating ordinary adjustment as if it required deep crisis interpretation.
- AG-verify | Check source, records, testimony, measurement, or context. | Endless checking to avoid action.
- AG-wait | Delay interpretation until capacity, evidence, or timing improves. | Passive avoidance.
- AG-document | Preserve records, dates, messages, observations, or decisions. | Weaponized documentation without care for truth.
- AG-protect | Establish safety, boundaries, separation, reporting, or safeguarding. | Calling protection unforgiveness.
- AG-refer | Involve clinician, supervisor, safeguarding officer, elder board, legal counsel, mediator, or qualified helper. | Dumping responsibility without handoff.
- AG-practice | Repeat a concrete skill, habit, discipline, or repair behavior under feedback. | Consuming more information instead of training.
- AG-negotiate | Seek a truthful agreement about duty, boundary, timing, role, resource, or shared expectation. | Using compromise language to avoid truth, protection, or responsibility.
- AG-lament | Grieve, pray, name loss, and refuse forced meaning. | Staying in grief to avoid repair.
- AG-confess / repent | Own wrongdoing and turn toward repair. | Shame collapse without concrete change.
- AG-repair | Apologize, restore, compensate, correct, reconcile where safe, or change practice. | Performative repair that protects image.
- AG-correct / disclose | Publicly correct falsehood, policy, teaching, record, or institutional communication. | Over-disclosure before verification.
- AG-escalate | Move to emergency, formal, disciplinary, protective, or legal process. | Escalating to punish when lower warranted action is enough.
### Examples and Edge Cases
- Case sketch | Likely profile | Coding note
- Weather forecast fails and plans change. | PE1--2 / MG0 / ST-high or mixed / CAP-green / AG-adjust + AG-verify | Factual model pressure without major meaning pressure.
- Medical diagnosis contradicts a belief that faithfulness guarantees health. | PE2 / MG3 / ST-high if medically supported / CAP-yellow / AG-refer + AG-lament + AG-adjust | Same signal creates factual and meaning pressure.
- Anonymous accusation against a leader with no records yet. | PE0--1 / MG2 / ST-low or unknown / CAP-yellow / AG-document + AG-verify + AG-protect if risk exists | Do not let low source trust erase protection duties when possible risk is serious.
- Multiple independent abuse reports with records. | PE3 / MG3 / ST-high / CAP-red or yellow / AG-protect + AG-refer + AG-escalate + AG-correct/disclose when verified | Institutional meaning pressure must not override evidence or safety.
- AI summary sounds confident but gives no sources. | PE0--1 / MG0--1 / ST-low / CAP-green / AG-verify | Fluency is not source trust.
- Person feels intense shame after a minor correction. | PE1 / MG2--3 / ST-high or mixed / CAP-yellow / AG-lament + AG-refer + AG-repair if needed | Emotional intensity may indicate meaning gap, not severe factual breach.
- Group stays calm after repeated near misses. | PE2--3 / MG1--2 / ST-high / CAP-green or yellow / AG-correct/disclose + AG-adjust | Calm can be false resonance when repeated signals are normalized.
Edge-case rules. Code unknown rather than guessing. Split compound cases into episodes. Record the strongest warranted code, not the loudest emotion. Escalate capacity to red when safety requires it, even if prediction error or meaning gap is still uncertain. In institutional cases, code the vulnerable person's capacity separately from the institution's capacity to process the signal.
### Interrater Reliability Goals
Reliability is not agreement for its own sake. It shows whether coders can apply the model without turning it into private interpretation. Use training vignettes, independent coding, blinded comparison where possible, adjudication notes, and revision of anchors before outcome claims are made.
For nominal fields such as source-trust category, capacity category, and agency class, report percent agreement and a chance-corrected statistic such as Cohen's kappa for two coders or Fleiss' kappa / Krippendorff's alpha for more coders. For ordered PE and MG levels, report weighted kappa or ordinal alpha. Krippendorff's common rule of thumb treats \(α ≥ 0.80\) as a strong target and \(α ≥ 0.667\) as the lowest range for tentative conclusions; CRM studies should aim for \(0.80\) or higher before making strong claims, and should revise the manual if core fields repeatedly fall below \(0.70\). [^interrater-reliability-goals-1]
Minimum reporting should include:
- coder training procedure and number of practice cases;
- whether coders saw full cases, excerpts, or structured vignettes;
- reliability for PE, MG, ST, CAP, and AG separately;
- common disagreement patterns and how they changed the manual;
- whether final codes came from independent ratings or adjudicated consensus;
- examples of low-agreement edge cases retained for future testing.
[^interrater-reliability-goals-1]: Klaus Krippendorff, Content Analysis: An Introduction to Its Methodology, 4th ed. (SAGE, 2018); Jacob Cohen, A coefficient of agreement for nominal scales, Educational and Psychological Measurement 20, no. 1 (1960): 37--46.
## Validation Ladder
The model becomes stronger through staged validation, not by being defended as finished. If it is used only as a reflective map, the standard is clarity and usefulness. If it becomes a questionnaire, training protocol, pastoral workflow, clinical-adjacent aid, or digital tool, the standard rises. Complex-intervention guidance gives the right sequence: develop the theory, test feasibility, evaluate effects, study context, and revise implementation rather than jumping straight from an elegant idea to confident deployment. [^validation-ladder-1]
- Stage | Question | Evidence needed
- Conceptual clarity | Do users understand the difference between prediction error, meaning gap, source trust, capacity, and agency? | Interviews, cognitive interviewing, inter-rater discussion, failure-case review.
- Coding reliability | Can trained users code the same case similarly without forcing agreement? | Case vignettes, independent coding, agreement metrics, adjudication notes.
- Construct validity | Do the constructs relate to adjacent measures without collapsing into them? | Factor analysis, convergent and discriminant validity, test-retest patterns.
- Usefulness | Does CRM improve clarity, source handling, help-seeking, truthful action, or repair? | Pilot studies, comparison conditions, follow-up outcomes.
- Safety | Does it reduce false closure without increasing rumination, shame, coercion, or delayed care? | Adverse-event tracking, referral audits, red-pressure review, participant feedback.
- Implementation | Can it be used responsibly across cultures, institutions, and digital tools? | Translation studies, process evaluation, stakeholder review, context adaptation.
[^validation-ladder-1]: Peter Craig et al., A new framework for developing and evaluating complex interventions: update of Medical Research Council guidance, BMJ 374 (2021): n2061, https://www.bmj.com/content/374/bmj.n2061.
## Instrument, Translation, and Release Gates
The word validated should never be used as a single all-purpose stamp. The Standards for Educational and Psychological Testing requires evidence for the interpretation and use being claimed, including validity, reliability or precision, fairness, administration, scoring, and the consequences of use. COSMIN likewise places content validity first: the proposed content must be relevant, comprehensive, and comprehensible for the construct, target population, and context before internal structure or reliability can make the instrument meaningful. [^instrument-translation-and-release-gates-1]
For CRM this produces a release sequence.
- Proposed use | Minimum gate before release | Permitted label before the gate
- Personal reflection map | Clear instructions, user comprehension, red-pressure stop rules, privacy expectations, and adverse-use feedback. | Proposed field guide for voluntary low-risk reflection.
- Facilitator or pastoral training | Defined competence, supervised practice, case-based reliability, power and consent rules, referral pathways, and safeguarding review. | Training draft; not certification or authorization to practice outside one's role.
- Research case coding | Stable coding unit, trained independent coders, field-specific agreement, uncertainty rules, and transparent adjudication. | Proposed coding manual.
- Questionnaire or score | Target construct and population, content validity, cognitive interviews, structural validity, reliability and measurement error, hypotheses testing, cross-cultural validity, fairness, and score-interpretation evidence. | Candidate items; not the CRM Inventory as an established measure.
- Intervention or decision support | Feasibility, comparison condition, benefit and harm outcomes, qualified governance, implementation evidence, and a stopping rule. | Experimental protocol; not treatment, diagnosis, or clinical decision rule.
- AI or digital implementation | All relevant gates above plus privacy, audit logs, hallucination testing, crisis routing, human review, bias testing, dependency monitoring, and defined accountability. | Prototype reflection aid.
Translation is adaptation, not word replacement. A meaning-frame question that sounds private and individual in one language may carry household, honor, caste, communal, religious, or political meanings in another. International Test Commission guidance therefore requires attention to cultural context, construct overlap, item development, administration, scoring, interpretation, equivalence, and documentation. [^instrument-translation-and-release-gates-2] CRM translation work should use bilingual and cultural experts, target-user cognitive interviews, independent review, pilot data, and explicit tests of whether PE, MG, ST, CAP, and AG still distinguish what the local setting needs them to distinguish. Back-translation alone is not enough.
No score without a use claim. Before reporting any total or subscale score, state who will use it, with whom, for what decision, under what conditions, and with what consequence if it is wrong. If those fields cannot be stated, keep the output as ordinary-language reflection rather than measurement.
[^instrument-translation-and-release-gates-1]: American Educational Research Association, American Psychological Association, and National Council on Measurement in Education, Standards for Educational and Psychological Testing (2014), open-access files at https://www.testingstandards.net/; COSMIN, Content validity and study-design resources, https://www.cosmin.nl/.
[^instrument-translation-and-release-gates-2]: International Test Commission, The ITC Guidelines for Translating and Adapting Tests, 2nd ed. (2017), https://www.intestcom.org/files/guideline_test_adaptation_2ed.pdf.
## Testable Predictions
The predictions below turn the prose into claims that can fail. That matters. A model of truthful pressure should be willing to face pressure itself.
These are pattern-level predictions, not scripts for how every person must respond. A high meaning gap may look like anger in one person, silence in another, and compulsive explanation in a third. The test is whether the breach profile helps predict better than a flat description such as stress, confusion, or crisis.
- High prediction error with low meaning gap should mainly produce curiosity, technical correction, or model revision.
- High meaning gap with low prediction error should mainly produce grief, identity threat, moral struggle, spiritual reinterpretation, or narrative repair.
- High prediction error and high meaning gap should predict panic, defensiveness, denial, polarization, collapse, repentance, or deep transformation depending on source trust, capacity, and support.
- Low source trust should reduce fact uptake even when evidence quality is high.
- Low capacity should increase avoidance, rumination, premature closure, spiritual bypassing, dissociation, or overcorrection.
- Naming the breach type should improve response quality because it keeps people from treating a fact problem as a meaning problem or a meaning problem as a fact problem.
- Breach-profile notation should increase shared clarity in mentoring, pastoral, clinical-adjacent, and institutional conversations, provided users understand that it is a field notation rather than a validated score.
- In misinformation-style tasks, source-handling prompts should reduce premature sharing, rumor uptake, and confidence in unverified claims.
- False-resonance explanations should reduce distress or conflict in the short term but predict weaker reality-contact, delayed repair, or repeated breach at follow-up.
- Red-pressure training should increase appropriate referral, documentation, protection, and use of formal process in abuse, self-harm, coercion, medical-risk, and institutional-cover-up vignettes.
- Communities with stronger correction channels should show less institutional false resonance after failure.
## Proposed Study Portfolio
The studies below are designed to make the model useful, testable, and revisable. They move from coding reliability to practical effect, then to pastoral and institutional use. Each study should be preregistered where possible, report null findings, track adverse effects, and preserve the distinction between immediate relief and truthful repair.
### Study 1: Can Coders Distinguish Prediction Error from Meaning Gap?
Question. Can trained coders reliably distinguish factual model pressure from meaning-frame pressure across varied cases?
Design. Vignette and case-excerpt coding study. Use 40--60 cases drawn from diagnosis, grief, betrayal, misinformation, scientific failure, institutional abuse, workplace failure, and faith crisis. Randomize case order. Coders work independently after brief training, then adjudicate disagreements.
Sample. Pilot with 12--20 coders; main study with 40--80 coders from mixed backgrounds such as psychology, ministry, education, organizational leadership, and lay mentoring.
Primary outcomes. Weighted kappa or ordinal alpha for PE and MG separately; percent agreement for whether a case is fact-dominant, meaning-dominant, mixed, or unclear.
Hypothesis. Coders should reach acceptable reliability after training, with stronger agreement on clear PE0/PE3 and MG0/MG3 cases than on mixed middle cases.
Failure pattern to learn from. If coders repeatedly treat emotional intensity as MG3, or treat every factual surprise as PE3, the anchors need revision before outcome studies proceed.
### Study 2: Does CRM Journaling Outperform Ordinary Journaling?
Question. Does structured CRM journaling produce better clarity and action than ordinary journaling or expressive writing?
Design. Randomized trial with three arms: ordinary journaling, expressive writing, and CRM mapping. Participants write about a meaningful but non-emergency pressure for 15--20 minutes on three days, then complete follow-up at one week and one month.
Sample. Pilot \(n=90\); main study \(n=240\)--\(360\), screened to exclude acute safety crises that require care before reflection.
Primary outcomes. Breach-type clarity, rumination or brooding, action quality rated from blinded summaries, help-seeking where appropriate, and whether participants completed one bounded truthful next step.
Hypothesis. CRM journaling should outperform ordinary journaling on breach clarity, source checking, capacity awareness, and action quality. It may not outperform expressive writing on immediate emotional relief, and that distinction matters.
Revision trigger. If CRM increases rumination, shame, premature certainty, or delayed help-seeking, the field guide and capacity rules must be revised before broader use.
### Study 3: Does Source-Trust Prompting Reduce Misinformation Sharing?
Question. Does a short source-trust prompt reduce willingness to share unverified claims without making users cynical about reliable claims?
Design. Online randomized experiment using a simulated feed or family-chat interface. Compare no prompt, generic accuracy prompt, and CRM source-trust prompt. Include true, false, misleading, and uncertain claims across health, finance, disaster, crime, politics, and local safety.
Sample. Pilot \(n=300\); main study \(n=800\)--\(1 , 200\), balanced by age, education, political identity, religious identity where relevant, and digital-literacy level.
Primary outcomes. Sharing intention, actual click-to-check behavior, source-quality discrimination, confidence calibration, delayed recall, and whether emotionally urgent claims are slowed for checking.
Hypothesis. The source-trust prompt should reduce sharing of weak-channel claims and improve calibration while preserving willingness to act on high-trust warnings.
Revision trigger. If the prompt merely suppresses all sharing, increases partisan dismissal, or reduces response to reliable warnings, the source-trust language is too blunt.
### Study 4: Does Capacity-First Triage Reduce Harmful Forced Meaning?
Question. When people are exhausted, grieving, threatened, or trauma-activated, does capacity-first triage reduce harmful attempts to force meaning too early?
Design. Training experiment with mentors, peer supporters, pastoral workers, coaches, and helping professionals. Participants respond to standardized vignettes before and after training. Compare ordinary advice training with CRM capacity-first training.
Sample. Pilot \(n=80\); main study \(n=200\)--\(300\), with expert review panels rating responses.
Primary outcomes. Correct identification of red/yellow capacity, referral or protection decisions, reduction in forced meaning, reduction in shame-inducing counsel, and improved sequencing of safety, evidence, lament, and interpretation.
Hypothesis. Capacity-first training should improve the order of response: safety and support first, interpretation later.
Safety rule. No live crisis counseling is conducted inside the study unless the study has clinical governance, referral pathways, and acute-risk protocols.
### Study 5: Does CRM Improve After-Action Reviews in Teams?
Question. Does adding PE, MG, ST, CAP, AG, and false resonance checks improve team after-action reviews after failure or near misses?
Design. Cluster randomized study with student teams, healthcare simulations, ministry teams, product teams, or nonprofit teams. Compare standard after-action review with CRM-enhanced review after a simulated or real project failure.
Sample. Pilot with 12--20 teams; main study with 40--80 teams, ideally across more than one setting.
Primary outcomes. Signal preservation, root-cause quality, psychological safety, blame reduction without accountability loss, quality of corrective actions, follow-through at 30--60 days, and whether mission or reputation language blocked correction.
Hypothesis. CRM-enhanced review should improve learning when teams face not only what failed but also the meaning frame the team wanted to protect.
Revision trigger. If the model slows teams without improving action quality, shorten the team protocol and reserve full coding for high-stakes cases.
### Study 6: Does Bounded Review Reduce Rumination?
Question. Does turning pressure into a named breach profile, one action sentence, and a scheduled review reduce rumination better than open-ended reflection?
Design. Micro-longitudinal study. Participants report an active interpersonal, academic, work, or faith pressure. Randomize to open reflection, problem-solving worksheet, or CRM bounded review. Use daily check-ins for 7--14 days and follow-up at one month.
Sample. Pilot \(n=90\); main study \(n=240\)--\(400\), excluding acute danger and cases requiring immediate professional care.
Primary outcomes. Brooding, worry, perceived closure, avoidance, quality of next action, sleep or stress self-report, and whether the review date prevents both endless analysis and premature closure.
Hypothesis. CRM should reduce unbounded rumination by making the pressure reviewable: this is the breach, this is the next action, this is when new evidence or fruit will be checked.
Failure pattern to learn from. If profiles become obsessive self-monitoring, the model needs lighter language and stronger stopping rules for low-risk cases.
### Study 7: Does CRM Help Pastors and Mentors Avoid Spiritual Bypassing?
Question. Does CRM training help pastors, mentors, and spiritual leaders avoid spiritual answers that bypass evidence, protection, lament, confession, or qualified care?
Design. Randomized training study using pastoral vignettes: grief, abuse disclosure, deconstruction, marital coercion, leader failure, medical diagnosis, scrupulosity, and congregational conflict. Compare ordinary pastoral empathy training with CRM discernment training.
Sample. Pilot \(n=60\); main study \(n=180\)--\(300\) pastors, elders, mentors, ministry workers, and seminary students.
Primary outcomes. Expert-rated response quality, correct referral or safeguarding, reduction in forced forgiveness, reduction in leader-protective framing, proper use of lament, and ability to separate God's character from failed human channels.
Hypothesis. CRM-trained participants should be more likely to protect vulnerable people, preserve evidence, name grief, avoid premature meaning, and choose a concrete faithful action.
Revision trigger. If users turn CRM into a new spiritual script or substitute it for pastoral wisdom, the training must emphasize listening, humility, referral, and case-specific judgment.
## Study Path
Good development starts with careful description and moves only gradually toward tools or interventions.
- Construct refinement. Use qualitative interviews and case studies across illness, grief, betrayal, moral failure, faith crisis, church conflict, institutional failure, scientific belief revision, misinformation pressure, and family rupture. Code prediction error, meaning gap, source trust, capacity, agency, false resonance, and repair path.
- CRM Inventory. Develop a scale with subscales for prediction error, meaning gap, source trust, capacity, agency, and resonance quality. Test reliability, factor structure, convergent and discriminant validity, test-retest stability, and measurement invariance using current scale-development and COSMIN-style standards. [^study-path-1]
- Existing-measure alignment. Compare CRM variables with meaning in life, posttraumatic cognitions, religious or spiritual struggle, moral injury, cognitive flexibility, intolerance of uncertainty, rumination, depression, anxiety, PTSD symptoms, and help-seeking. Candidate measures include the Meaning in Life Questionnaire, Posttraumatic Cognitions Inventory, Religious and Spiritual Struggles Scale or RSS-5, psychological flexibility/inflexibility measures such as CompACT or MPFI, intolerance-of-uncertainty measures, rumination measures, and routine outcome monitoring instruments. [^study-path-2]
- Experimental disconfirmation. Present expectation-violating information under varied source-trust and identity-threat conditions. Test whether naming the breach as factual, meaning-level, or mixed improves accuracy, humility, emotional regulation, and action.
- Longitudinal crisis studies. Track people facing diagnosis, bereavement, job loss, betrayal, public failure, or faith crisis. Test whether early CRM mapping predicts later avoidance, rumination, resilience, repentance, repair, help-seeking, and integration.
- Intervention pilots. Compare CRM journaling or pastoral discernment against expressive writing, ordinary reflection, or care as usual. Outcomes should include clarity, reduced rumination, truthful action, source checking, help-seeking, and repair without bypassing.
- Institutional after-action studies. Apply CRM coding to churches, schools, companies, nonprofits, and public agencies after failure. Test whether shared meaning, reputation, or mission language blocked prediction error from becoming correction.
- Cross-cultural and translation studies. Test whether the distinctions work across languages, family structures, religious settings, explanatory models of illness and suffering, institutional forms, and individualist or collectivist contexts. Behavioral-science research has repeatedly warned against making broad claims about human psychology from narrow WEIRD samples. [^study-path-3] Cross-cultural work should therefore ask whether prediction error, meaning gap, source trust, capacity, and agency are recognized similarly or differently across cultures. Begin with emic interviews and local terms rather than assuming the English constructs have travelled. A translated inventory must demonstrate measurement invariance or another defensible form of cross-cultural equivalence before mean scores or thresholds are compared; linguistic fluency alone is not construct validity. Use a cultural-formulation interview when clinical or spiritual explanations differ, while retaining a non-negotiable safety floor against coercion, abuse, retaliation, and silencing vulnerable people. [^study-path-4]
- Open-science development. Preregister hypotheses, measures, inclusion criteria, and analysis plans where possible. [^study-path-5] Null results and boundary findings should be welcomed because they will show where this account works, where it needs revision, and where another account is better.
[^study-path-1]: COSMIN reliability and measurement-error standards, https://bmcmedresmethodol.biomedcentral.com/articles/10.1186/s12874-020-01179-5; Godfred O. Boateng et al., Best Practices for Developing and Validating Scales for Health, Social, and Behavioral Research, https://pubmed.ncbi.nlm.nih.gov/32025957/.
[^study-path-2]: For PTCI see Foa et al., The Posttraumatic Cognitions Inventory: Development and validation, https://www.psy.ox.ac.uk/publication/311258. For RSS-5 see Currier et al., Development and Evaluation of the Religious and Spiritual Struggles Scale-5, https://link.springer.com/article/10.1007/s10862-024-10182-9. For current ACT mechanism-measurement recommendations see Macri and Rogge, https://www.sciencedirect.com/science/article/abs/pii/S0272735824000539.
[^study-path-3]: Joseph Henrich, Steven J. Heine, and Ara Norenzayan, The weirdest people in the world?, Behavioral and Brain Sciences 33, no. 2--3 (2010): 61--83, https://pubmed.ncbi.nlm.nih.gov/20550733/.
[^study-path-4]: American Psychiatric Association, Cultural Formulation Interview, https://www.psychiatry.org/getmedia/5cc5329d-3bd4-4c6a-bae1-dfd0d6496f44/APA-DSM5TR-CulturalFormulationInterview.pdf; Kyunghee Han, Stephen M. Colarelli, and Nathan C. Weed, Methodological and Statistical Advances in the Consideration of Cultural Diversity in Assessment, Psychological Assessment 31, no. 12 (2019): 1481--1496, DOI: 10.1037/pas0000731.
[^study-path-5]: See the Association for Psychological Science introduction to preregistration, https://www.psychologicalscience.org/publications/psychological_science/preregistration.
## Research Ethics
CRM studies would often touch grief, trauma, faith, family rupture, institutional betrayal, shame, and health uncertainty. Ethics therefore belongs inside the development process, not beside it as an administrative afterthought. Human-subjects research should follow ordinary research-ethics principles: respect for persons, beneficence, and justice. In practice, that means informed consent, risk-benefit assessment, fair participant selection, confidentiality, referral pathways, and careful handling of vulnerable participants. [^research-ethics-1]
Minimum ethics requirements for CRM research and tool development are:
- Do not recruit people in acute crisis unless the study is designed and supervised for acute-risk work.
- Separate research participation from pastoral, clinical, employment, school, or institutional pressure whenever possible.
- Make clear that CRM is not a diagnosis, substitute clinician, legal process, or proof that a source is true.
- Track harms as well as benefits: rumination, shame, premature forgiveness, delayed reporting, spiritual bypassing, false accusation, or coercive institutional closure.
- Build referral and safeguarding pathways into any study involving abuse, self-harm risk, coercion, or institutional misconduct.
[^research-ethics-1]: The Belmont Report, Office for Human Research Protections, HHS, https://www.hhs.gov/ohrp/regulations-and-policy/belmont-report/read-the-belmont-report/index.html.
## Minimum Viable Pilot
A first practical test should be small, clear, and hard on the model. Recruit participants facing a non-emergency but meaningful pressure: academic failure, job loss, family conflict, faith doubt, diagnosis adjustment, or institutional disappointment. Randomize them to ordinary reflection, expressive writing, or CRM mapping. Compare immediate clarity, rumination, source-checking, capacity awareness, action quality, help-seeking, and follow-up integration after one week and one month. Expressive-writing research is the right comparison because it already tests structured disclosure around stressful experiences rather than generic journaling. [^minimum-viable-pilot-1]
The pilot should not ask whether CRM makes people feel better fastest. It should ask whether it helps them name the right breach, avoid false closure, seek appropriate support, and take a more truthful action.
[^minimum-viable-pilot-1]: Joanne Frattaroli, Experimental Disclosure and Its Moderators: A Meta-Analysis, Psychological Bulletin 132, no. 6 (2006): 823--865, https://doi.org/10.1037/0033-2909.132.6.823.
## Example Inventory Items
These are not validated items. They show the kind of measurement work validation would require.
- This event violated what I expected would happen.
- I can name the specific expectation that failed.
- This event threatens the meaning frame I use to understand my life.
- I know which source or channel I am trusting.
- I have enough support and stability to process this honestly.
- I can name one truthful action available to me now.
- My current explanation helps me face reality rather than avoid it.
- My current explanation protects someone or something from needed correction.