Behavior Analysis

Behavior Analysis

Behavior Analysis = BEHAVIOR + LOCATION. You are documenting a specific behavior in a specific place/context, not behavior in the abstract — and not yet mapping any particular target audience.

A Behavior Analysis describes what happens, where, when, how, and under what eligibility constraints, before any audience-specific judgement is layered on. It is the shareable, indexable foundation that downstream analyses — most importantly COM-B Analyses of specific target audiences — link to. One behavior analysis can be referenced by many COM-B analyses, each applying a different audience lens.

The framework is grounded in U.S. Army doctrine TM 3-53.11 Influence Process Activity: Target Audience Analysis, Chapter 2¹, and operationalised in the Behavior Analysis tool at researchtools.net².

Why “behavior + location” matters

Behaviors vary enormously by geography, regulation, infrastructure, and culture — even when the surface label is the same.

• Voting in U.S. presidential elections ≠ voting in Nigerian state elections — different eligibility, registration, frequency, security context, symbols.
• Solar panel adoption in California ≠ solar panel adoption in Lagos — different incentives, climate, grid reliability, financing.
• Even online behaviors (e.g. posting political opinions) vary by platform, language, jurisdiction, and platform-specific norms.

Treating “voting” as a single behavior produces analyses that are wrong everywhere because they’re trying to be true everywhere. Locking in location upstream of any audience analysis is what makes a behavior description usable for actual intervention design.

Architectural Position

Behavior Analysis sits upstream of all audience-specific work. The clean layering:

Layer	Framework	Primary question	Output
1. Behavior Analysis (this page)	Audience-agnostic behavior+location mapping	What happens, where, when, with what consequences and symbols?	A reusable, indexable, citation-ready behavior description
1.5. Operational Frame	Direction + desired audience behavior + baseline + measurement plan	Which way do we want to shift this behavior, in whom, and how will we know it changed?	Operator-anchored frame: direction enum + Michie/DOD measurement plan + audience candidates filtered by direction
2. COM-B Analysis	Capability, Opportunity, Motivation for a specific target audience against a specific behavior3	For this audience, what’s missing/present/competing in C, O, M, and which deficits are leverage points for THIS direction?	Direction-aware behavioural diagnosis (deficits per sub-component + direction-aware leverage interpretation)
3. BCW Intervention Design	Behaviour Change Wheel — 9 intervention functions, 7 policy categories, BCTs34	Given the diagnosis, what intervention will work, and how do we deliver it?	Designed intervention scored on APEASE

A single behavior analysis (e.g. “Cleaning hands using alcohol gel — UK NHS hospitals”) can spawn multiple COM-B analyses — one for nursing staff, one for senior doctors, one for visitors — each leading to a different intervention.

Don’t collapse the layers

The most common mis-step is mixing audience-specific judgements (“nurses are too busy to wash”) into the behavior analysis itself. The behavior analysis must stay objective: “the behavior is performed by all clinical staff during each shift, before/after patient contact”. The “too busy” claim belongs in a downstream COM-B analysis of Physical Opportunity for the nursing-staff audience.

Fast-path preview: the AI Behavior Analysis Tool

For early exploration the live platform provides a Behavior Analysis Tool — a single-screen AI assistant that produces a quick COM-B diagnosis from a free-text behavior description. It supports two contexts: Intelligence (adversary analysis) and Product/Research (user/stakeholder analysis).

The preview tool is not a substitute for a saved Behavior Analysis. It exists as a fast-path that displays a banner pointing users to the proper L1 → L2 flow, and offers a “Save as COM-B Analysis” button that hands the AI’s diagnosis off to the COM-B Analysis create form (with ?prefill=ai-tool). The receiver pre-populates the title, description, selected intervention functions, COM-B deficit levels, and per-component evidence notes from the AI output — but the form still requires the user to anchor the saved analysis to a Behavior Analysis (L1) before saving (linked_behavior_id is enforced both client- and server-side).

The Behaviour Change Wheel — at a glance

The full BCW is a three-ring concentric diagram. The hub is the COM-B model with six sub-components. The middle ring holds the nine intervention functions that target deficits in C, O, or M. The outer ring holds the seven policy categories that deliver those intervention functions at scale.

Adapted from Michie, Atkins & West (2014). The Behaviour Change Wheel: A Guide to Designing Interventions. Silverback Publishing.

The interactive version of this wheel — where clicking a COM-B segment reveals which intervention functions apply, and clicking an intervention reveals which policy categories can deliver it — is built into the COM-B Analysis tool at researchtools.net. Source: src/components/frameworks/WheelSVG.tsx in the researchtoolspy repository.

Public API — query the canon directly

The same canonical reference data and recommendation logic are also exposed as three public endpoints. Useful for CLI tools, AI agents, MCP servers, and integrations that want canon-backed behaviour-change suggestions without re-implementing the BCW:

Endpoint	Purpose
GET https://researchtools.net/api/frameworks/comb-analysis/canon	Full reference: 6 COM-B sub-components, 9 functions, 7 policies, 16 BCT groupings, all matrices
POST https://researchtools.net/api/frameworks/comb-analysis/recommend	Body {deficits:{...}} → recommended intervention functions per BCW Guide Tables 2.1, 2.3
POST https://researchtools.net/api/frameworks/comb-analysis/recommend-bcts	Body {functions:[...]} → recommended BCTs per BCW Guide Table 3.3

Cached at the edge (canon: 1 hour). No auth required. Each response embeds canonical citations so downstream consumers can cite Michie et al. directly. Full curl examples + end-to-end pipeline (diagnose → functions → BCTs) in docs/COM_B_API.md.

The Eight Sections

The researchtools.net Behavior Analysis framework is built around eight sections, derived from TM 3-53.11 Chapter 2¹²:

1. Basic Information — title, description, location context, settings, temporal context, eligibility, complexity
2. Behavior Timeline — sequenced events with location, sub-steps, decision points and forks
3. Environmental Factors — physical infrastructure, resources, accessibility constraints
4. Social and Cultural Context — norms, influences, community dynamics
5. Consequences and Outcomes — immediate / long-term / generational; positive / negative / neutral / mixed; who is affected
6. Symbols and Signals — visual, auditory, social cues (with media support)
7. Observed Patterns — variations, common sequences, shortcuts, subgroup differences
8. Potential Target Audiences — who currently performs it, who could but doesn’t (flagged for downstream COM-B work — not analysed here)

Each section is documented in detail below, mirroring the live tool’s field structure.

1. Basic Information

The basic-info card answers the four foundational questions: what is the behavior, where does it occur, when, and who is eligible.

Behavior Identification

Field	Required	Description
Behavior title	✅	Specific, clear name. Example: “Voting in Presidential Elections”, “Installing Residential Solar Panels”
Behavior description	✅	Concise prose description. Avoid audience-specific framing.

Location & Geographic Context (REQUIRED)

This is the single most-emphasised field in the live tool. Behaviors that aren’t location-anchored cannot be reused, indexed, or compared across contexts.

Field	Description
Geographic scope	One of: local (city/neighborhood) · regional (state/province) · national (country-wide) · international (multiple countries) · global (worldwide — rare)
Specific locations	Free list of places — “California, USA”, “Lagos, Nigeria”, “European Union”. Indexed for search.
Location-specific notes	Regulations, cultural factors, infrastructure peculiarities

Behavior Settings

Where/how the behavior takes place. Multi-select:

in_person 🏢 · online 💻 · hybrid 🔄 · phone 📞 · mail 📬 · app 📱

Temporal Context

Field	Description
Frequency pattern	daily · weekly · monthly · quarterly · semi_annual · annual · biennial · seasonal · one_time · irregular · as_needed · custom
Custom frequency	When custom is chosen: every N minutes/hours/days/weeks/months/years
Time of day	Multi-select: morning 🌅 · afternoon ☀️ · evening 🌆 · night 🌙 · any_time 🕐
Typical duration	Free text — “5 minutes”, “1 hour”, “ongoing”
Timing notes	Seasonal variations, peak times, scheduling constraints

Eligibility Requirements

Document any prerequisites for being able to perform the behavior. These are still audience-agnostic — they describe gating constraints that apply to whoever attempts the behavior:

Requirement type	Examples
Age	”Must be 18”, “Under 21 prohibited”
Legal	”US Citizen”, “Valid driver’s license”, “Permit required”
Skill	”Computer literacy”, “Reads English at grade 8 level”
Resource	”Internet access”, “Smartphone with camera”, “$X funding”
Other	Any other prerequisite

A boolean “This behavior has eligibility requirements” gates the section — many simple behaviors have none.

Behavior Complexity

Level	Definition
Single action	One simple step (press a button, make a call)
Simple sequence	2–5 straightforward steps in order
Complex process	Multiple steps with decisions and alternatives
Ongoing practice	Continuous or repeated behavior over time

Complexity drives the level of timeline detail expected in section 2.

2. Behavior Timeline

A sequenced model of when, where, and how long the behavior plays out. The live tool supports interactive timelines with three node types²:

Node	What it captures
Event	A discrete step in the sequence. Has label, optional time (HH:MM or relative like T+30min), description, location, and a flag for is this a decision point?
Sub-step	Granular sub-actions inside an event — label, description, duration
Fork	An alternative path or parallel possibility. Has a condition (“If X happens”, “Alternative path”) and its own nested event sequence

Events can also be nested — an event in one behavior can link to an entire other behavior analysis (linked_behavior_id)², so a complex behavior like “applying for asylum” can decompose into sub-behaviors that have their own analyses.

Worked example — voting: Event 1: Register to vote (location: county clerk’s office, sub-steps: gather ID, fill form, submit). Event 2: Receive sample ballot (T+2 weeks, by mail). Event 3: Decision point — vote early, vote on election day, or vote by mail (three forks, each with its own event sequence).

3. Environmental Factors

Document the physical and environmental context. Prompt questions from the live tool²:

• What physical infrastructure exists? (buildings, roads, facilities)
• What resources are available? (equipment, materials, spaces)
• What are the accessibility considerations?
• What physical constraints or enablers exist?
• What environmental conditions affect this? (weather, climate, terrain)

These factors are physical opportunity in COM-B vocabulary — but at this layer they’re documented descriptively, without yet judging whether any particular audience has access to them.

4. Social and Cultural Context

Cultural norms, social influences, and community dynamics. Prompts²:

• What are the cultural norms around this behavior?
• What social influences exist? (family, peers, community)
• Are there community leaders or influencers?
• What group dynamics or social pressures exist?
• How do people talk about or communicate about this behavior?

This maps to social opportunity in COM-B — again, captured descriptively.

5. Consequences and Outcomes

What happens after the behavior is performed. Each consequence is captured with three dimensions²:

Dimension	Values	Why it matters
Timeframe	immediate · long_term · generational	A behavior with positive immediate consequences but negative generational consequences (e.g. groundwater pumping for irrigation) requires different intervention thinking than a uniformly-bad behavior
Valence	positive · negative · neutral · mixed	Mixed-valence consequences are common and important to flag
Who is affected	Free text	The who is part of the consequence, not the audience analysis — captures who experiences the consequence (which may be third parties)

Prompts²:

• What are the immediate consequences?
• What are the long-term outcomes?
• What rewards exist? (intrinsic and extrinsic)
• What costs or penalties exist?
• What unintended consequences occur?

6. Symbols and Signals

Visual, auditory, or social cues associated with the behavior. Each symbol²:

Field	Description
Name	What is this symbol? — “Red baseball cap”, “Victory gesture”
Symbol type	visual · auditory · social · other
Description	What does it represent / signify?
Context	When / where is it used?
Media	Image upload (URL or base64) for visual symbols; audio upload for auditory symbols

Prompts²:

• What symbols are associated with this behavior?
• What signals indicate someone is about to do this or has done it?
• Are there visual cues? (clothing, logos, badges)
• Are there auditory cues? (sounds, music, verbal phrases)
• What social status or identity markers exist?

7. Observed Patterns

Variations and patterns in how different people perform the behavior — still without naming or analysing specific audiences. Prompts²:

• What variations exist in how people perform this?
• What are common sequences or typical paths?
• What shortcuts or workarounds do people use?
• How does performance vary by subgroup or demographic? (Subgroups identified here flow into section 8 as candidates for downstream COM-B analyses.)
• What adaptations do people make to constraints?

8. Potential Target Audiences

The handoff section. The live tool flags this explicitly: “Identify audience segments for COM-B Analysis”². Prompts²:

• Who currently performs this behavior?
• Who could perform it but doesn’t?
• What are the key demographic segments?
• What psychographic differences exist between groups?
• Who influences whether others perform this?

This is a list of candidates, not an analysis. Each candidate audience becomes the subject of a separate COM-B Analysis³ that links back to this behavior analysis. That downstream analysis is where capability/opportunity/motivation deficits get diagnosed and BCW interventions get designed.

The §8 list is split by direction-leverage: which audiences are leverage points if you wanted to increase the behavior, which if you wanted to decrease it. L1 itself does not pick a direction (that’s the Operational Frame’s job below) — it surfaces both lists. An audience can appear in both if its behavior distribution within that group is mixed.

Operational Frame — Between L1 and L2

The Operational Frame sits between L1 (Behavior Analysis) and L2 (COM-B Diagnosis). It captures the operator’s intent — direction, desired audience behavior, baseline rate, measurement plan, operational so-what — that L1 deliberately does not contain.

The frame produces these fields per (behavior, objective) pair:

Field	Description
direction	increase · decrease · shift · introduce — parsed from the operator’s free-text objective via verb keying
desired_behavior	{title, description, delta_from_reference, substitute_for?} — what the audience does if the objective succeeds. May differ from the L1 reference behavior (opposite for ↓, alternative for shift, novel for introduce, same for ↑/maintain)
baseline	Michie behavior-frequency framework — current_rate_estimate, measurement_method, confidence (high/medium/low/unknown)
measurement	Three Michie layers + DOD framing: behavior_indicators (countable signals), com_b_shift_indicators (capability/opportunity/motivation pre/post), outcome_indicators (downstream effects), plus moe_candidates (DOD MOE — observable change) and mop_candidates (DOD MOP — output quantities)
operational_so_what	Causal chain from PSYOP behavior change to maneuver CDR’s intent — must name a tactical/operational outcome (rejecting non-statements like “improve security”)
filtered_audience_candidates	The §8 candidates filtered by direction, each with a com_b_hypothesis previewing the L2 work
warnings	Non-actionable objective flags, ROE concerns, language-mismatch hints, shift-without-substitute warnings

Why the frame is its own layer

Without the frame, the §6 Symbols section silently drifts toward “behavior is happening” cues (razor wire installed, citadel doors built) when the operational target is exactly the audiences where the behavior is not happening — which is where intervention has leverage. And L2 without an objective is the compliance form-filling that the PSYOP branch is rightly mocked for: a TAAW that gets filed and never read because it has no operational so-what tying back to the maneuver CDR’s intent.

The frame fixes this by forcing the operator to name (a) which way they want to shift the behavior, (b) what the audience does today vs what they need to do, (c) how anyone would know if it changed, and (d) what tactical outcome that change unlocks. The result feeds L2 with a context that makes the COM-B diagnosis directly actionable.

Direction-aware leverage in L2

L2’s COM-B deficits map stays direction-agnostic — it describes the audience’s COM-B state objectively. But the intervention implication flips by direction (per BCW Guide Box 2.6):

• ↑ direction: a major_barrier is high priority (gap to close — strengthen via training/persuasion/etc); adequate is low priority (already there, no leverage)
• ↓ direction: an adequate is high priority (entrenched — target via restriction/environmental_restructuring/coercion); major_barrier is low priority (already partially blocked)
• shift: composite — high priority where original is enabled OR where substitute is blocked
• introduce: like ↑, but emphasizes education/modelling/training

L2’s direction_aware_leverage field surfaces this so the operator doesn’t read the deficit map upside-down.

Signal bot pipeline

The community Signal bot exposes the L1 → Frame → L2 pipeline directly:

!bcw <behaviour>                              → L1 Behavior Analysis (audience-agnostic)
!bcw <behaviour> | <objective>                → Operational Frame
!bcw <behaviour> | <objective> | <audience>   → L2 COM-B for that audience
!bcw psy:maj po:def                           → deterministic deficit map (skip AI)

Examples:

!bcw midterm voting in america
!bcw midterm voting in america | increase first-time voter turnout
!bcw fighting off pirates | decrease resistance during USCG VBSS | small fishing vessel crews

The pipe | separates segments; the operator’s objective sits between behavior and audience. Direction is parsed from the objective via verb keying.

How a Behavior Analysis Becomes Useful

The intent of the framework is that a single, well-documented behavior analysis becomes a reusable asset:

• Indexed by location, scope, settings, frequency, complexity, category, and tags — so others searching for “residential solar adoption in California” find it².
• Public-submission-ready — the data model includes is_public, tags, category, upvotes, and usage_count (how many COM-B analyses link to it)².
• Linked-to by COM-B analyses for specific audiences — multiplying the value of the original effort.
• Citable in research, intelligence products, and policy documents.

This is fundamentally different from the “single-analyst-writes-a-one-off-document” pattern: the behavior analysis is designed to be shared infrastructure for downstream behaviour-change work.

Audience-Agnostic Discipline — A Checklist

When filling in any section, run each line through this filter:

• Could this statement be true regardless of who performs the behavior?
• Is this describing the environment, timing, eligibility gate, consequence, symbol, or observed pattern — rather than a particular group’s relationship to those things?
• Have I avoided phrases like “the user is too busy”, “young people don’t care about”, “customers struggle with”? (These are COM-B claims, not behavior-analysis facts.)
• Have I anchored every claim to the location specified in section 1?

If a piece of information is audience-specific (e.g. “medical residents in their first year report fatigue as a barrier”), defer it. Note the audience as a candidate in section 8, then create a COM-B Analysis for that audience that contains the claim.

When (and How) to Move to COM-B

After the behavior analysis is complete, identify the priority audience(s) from section 8. For each, open a separate COM-B Analysis³ with a linked_behavior_id pointing to this analysis². The COM-B analysis then asks, for that audience only:

• Physical capability — do they have the skills, strength, stamina?
• Psychological capability — knowledge, comprehension, mental skills?
• Physical opportunity — environmental access, time, resources?
• Social opportunity — cultural and interpersonal support?
• Reflective motivation — beliefs, intentions, goals, identity?
• Automatic motivation — emotions, impulses, habits?

The deficits identified there feed Steps 5–8 of the BCW Guide³ — selecting intervention functions, policy categories, behaviour change techniques (from BCTTv1’s 93 BCTs⁴), and mode of delivery, scored against APEASE.

The live COM-B Analysis tool at researchtools.net walks the analyst through all four steps:

1. APEASE Evaluation — score each candidate intervention on Affordability, Practicability, Effectiveness/cost-effectiveness, Acceptability, Side-effects/safety, and Equity (BCW Guide Table 1, pp. 18–20).
2. Behaviour Change Techniques — pick from the full BCTTv1 (93 BCTs in 16 groupings). The tool surfaces the most-frequently-used BCTs for the selected intervention functions first (BCW Guide Table 3.3, pp. 145–160), then offers the rest of the taxonomy as collapsible accordions.
3. Mode of Delivery — specify who delivers, when, how often, in what setting, individual vs group, face-to-face vs broadcast vs internet etc. (BCW Guide Step 8 / Box 2.9).
4. Save — the COM-B Analysis is anchored to its parent Behavior Analysis via linked_behavior_id; both client and server enforce the link.

Full canonical reference for the COM-B model, the 9 intervention functions, the 7 policy categories, APEASE, and the 8-step BCW intervention design process is in the BCW Guide³ and the companion ABC of Behaviour Change Theories⁵. Those two books document the academic foundation; this page documents the upstream objective behavior description that feeds them.

Quick Reference: The Live Tool’s Data Model

For developers, integrators, and analysts who want to map their own work onto the structure:

BehaviorAnalysis {
  title: string
  description: string
  source_url?: string

  location_context: {
    geographic_scope: 'local' | 'regional' | 'national' | 'international' | 'global'
    specific_locations: string[]   // indexed
    location_notes?: string
  }
  behavior_settings: {
    settings: ('in_person' | 'online' | 'hybrid' | 'phone' | 'mail' | 'app')[]
    setting_details?: string
  }
  temporal_context: {
    frequency_pattern?: FrequencyPattern
    custom_frequency_number?: number
    custom_frequency_unit?: 'minutes' | 'hours' | 'days' | 'weeks' | 'months' | 'years'
    time_of_day?: ('morning' | 'afternoon' | 'evening' | 'night' | 'any_time')[]
    duration_typical?: string
    timing_notes?: string
  }
  eligibility: {
    has_requirements: boolean
    age_requirements?: string[]
    legal_requirements?: string[]
    skill_requirements?: string[]
    resource_requirements?: string[]
    other_requirements?: string[]
  }
  complexity: 'single_action' | 'simple_sequence' | 'complex_process' | 'ongoing_practice'

  timeline?: TimelineEvent[]            // events with sub-steps and forks
  environmental_factors?: any[]
  social_context?: any[]
  consequences?: ConsequenceItem[]      // timeframe + valence + who_affected
  symbols?: SymbolItem[]                // type + media (image/audio)
  observed_patterns?: any[]
  potential_audiences?: any[]           // candidates for COM-B analyses

  is_public?: boolean
  tags?: string[]
  category?: string                     // Health, Civic, Economic, Social, Environmental, etc.
}

Source: src/types/behavior.ts in the researchtoolspy repository².

---

Pre-Filled Template (Copy & Replace)

Use this skeleton when starting a new behavior analysis on a wiki page (or in any markdown context). Replace bracketed placeholders. Do not add audience-specific framing — that is for a downstream COM-B Analysis.

# [BEHAVIOR TITLE]

**Location**: [SPECIFIC LOCATIONS, e.g., "Cook County, Illinois, USA"]
**Geographic scope**: [local | regional | national | international | global]
**Description**: [Concise, audience-agnostic description.]

## Basic Information

- **Settings**: [in_person, online, hybrid, phone, mail, app — pick all that apply]
- **Frequency**: [daily | weekly | … | as_needed | custom: every N units]
- **Time of day**: [morning, afternoon, evening, night, any_time]
- **Typical duration**: [e.g., "5 minutes", "ongoing"]
- **Complexity**: [single_action | simple_sequence | complex_process | ongoing_practice]

### Eligibility (omit if `has_requirements = false`)
- **Age**: [list]
- **Legal**: [list]
- **Skill**: [list]
- **Resource**: [list]
- **Other**: [list]

## Behavior Timeline

1. **[Event label]** — *[time / T+...]* at *[location]*
   - Sub-steps: [if simple_sequence or complex_process]
   - Decision point? [yes/no]
   - Forks: [if complex_process — list alternative paths with conditions]
2. **[Event label]** — …

## Environmental Factors

- Physical infrastructure: [...]
- Resources / equipment: [...]
- Accessibility: [...]
- Physical constraints / enablers: [...]
- Environmental conditions: [...]

## Social and Cultural Context

- Norms: [...]
- Social influences (family, peers, community): [...]
- Influencers / community leaders: [...]
- Group dynamics / social pressures: [...]
- How the behavior is talked about: [...]

## Consequences and Outcomes

| Consequence | Timeframe | Valence | Who is affected |
|---|---|---|---|
| [...] | immediate / long_term / generational | positive / negative / neutral / mixed | [...] |

## Symbols and Signals

| Symbol | Type | Description | Context | Media |
|---|---|---|---|---|
| [name] | visual / auditory / social / other | [what it represents] | [when/where used] | [link] |

## Observed Patterns

- Variations in how the behavior is performed: [...]
- Common sequences / typical paths: [...]
- Shortcuts / workarounds: [...]
- Subgroup variations *(audience candidates — see next section)*: [...]
- Adaptations to constraints: [...]

## Potential Target Audiences (Handoff to COM-B Analysis)

> These are **candidates only** — each gets a separate COM-B Analysis.

- **[Audience 1]** — currently performs / could perform — *brief justification, no diagnosis here*
- **[Audience 2]** — …

## Tags & Indexing

- **Category**: [Health | Civic | Economic | Social | Environmental | Security | …]
- **Tags**: `[tag1, tag2, tag3]`
- **Public submission**: [yes / no]

---

References

Additional Resources

• ResearchTools.net — Behavior Analysis — start a new behavior analysis (audience-agnostic)
• ResearchTools.net — COM-B Analysis — downstream target-audience-specific analysis that links to a behavior analysis
• Behaviour Change Wheel official site — book, training, worked examples
• Behaviour Change Theories companion site — searchable index of all 83 theories
• BCT Taxonomy v1 — definitions and worked examples for all 93 BCTs
• Related wiki pages: PMESII-PT · Center of Gravity Analysis Guide

Footnotes

1. Department of the Army. TM 3-53.11 Influence Process Activity: Target Audience Analysis, Chapter 2. The doctrinal basis for the eight-section behavior-analysis structure used by the live tool. (Internal U.S. Army publication; analogous public NATO doctrine: AJP-3.10.1 Allied Joint Doctrine for Psychological Operations.) ↩ ↩²

2. ResearchTools.net. Behavior Analysis framework — Analysis Frameworks dashboard. “Objective documentation of behaviors in specific locations/contexts (based on U.S. Army FM TM 3-53.11).” Source code: src/types/behavior.ts, src/components/frameworks/BehaviorBasicInfoForm.tsx, src/components/frameworks/BehaviorTimeline.tsx, src/config/framework-configs.ts in the researchtoolspy repository. Quote from src/types/behavior.ts: “IMPORTANT: Behavior Analysis = BEHAVIOR + LOCATION. These analyses document specific behaviors in specific contexts/locations. Designed for public submission, indexing, and querying.” ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶ ↩⁷ ↩⁸ ↩⁹ ↩¹⁰ ↩¹¹ ↩¹² ↩¹³ ↩¹⁴ ↩¹⁵ ↩¹⁶ ↩¹⁷

3. Michie, S., Atkins, L., West, R. The Behaviour Change Wheel: A Guide to Designing Interventions. Silverback Publishing, 2014. ISBN 978-1-912141-08-1. “COM-B stands for Capability Opportunity Motivation – Behaviour. The COM-B model is the starting point used by the BCW for understanding behaviour in the context in which it occurs… The BCW consists of three layers. The hub of the wheel identifies the sources of the behaviour that could prove fruitful targets for intervention. It uses the COM-B model for this. Surrounding this is a layer of nine intervention functions to choose from depending on the particular COM-B analysis one arrives at. Then the outer layer, the rim of the wheel, identifies seven types of policy that one can use to deliver these intervention functions.” (Step 4 — pp. 47–60; Steps 5–6 + Tables 2.1, 2.3, 2.7 — pp. 88–116; APEASE Table 1 — pp. 18–20; Step 7 BCTs + Table 3.3 — pp. 145–160.) ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶

4. Michie, S., Richardson, M., Johnston, M., et al. The Behavior Change Technique Taxonomy (v1) of 93 hierarchically clustered techniques: building an international consensus for the reporting of behavior change interventions. Annals of Behavioral Medicine 46(1):81–95, 2013. “93 BCTs hierarchically clustered into 16 groupings via expert card-sort.” Official taxonomy: www.ucl.ac.uk/health-psychology/BCTtaxonomy. ↩ ↩²

5. Michie, S., West, R., Campbell, R., Brown, J., Gainforth, H. ABC of Behaviour Change Theories. Silverback Publishing, 2014. ISBN 978-1-912141-05-0. “83 theories of behaviour change identified through systematic search of six databases (PsycINFO, Econlit, Cochrane, IBSS, EMBASE, MEDLINE), articles published between 1 January 1960 and 11 September 2012.” (Companion volume to the BCW Guide; defines theory and behaviour, lists 9 quality criteria for theory, and catalogues all 83 theories.) ↩