APCAT cheat sheet

PDCA (Plan-Do-Check-Act) is the underlying structure of most management-system standards: establish objectives, implement, verify against the objectives, then improve and re-baseline.

A logic chain runs inputs → activities → outputs → outcomes → impact; outputs are direct deliverables, outcomes are changes in the target population, impact is the long-term contribution.

Evidence hierarchy: systematic reviews and randomised designs generally outrank observational studies, which outrank expert opinion and anecdote.

Competence is demonstrated ability against a defined standard; capability is the broader capacity to adapt. Assessing competence always requires an explicit criterion and evidence.

Define terms once in a glossary and use them identically throughout; in formal documents 'shall' denotes a requirement, 'should' a recommendation, and 'may' a permission.

Every profession separates three artifacts: a body of knowledge (what is known), a standard of practice (what must be done), and a code of conduct (how one must behave).

Systems-thinking rule: an intervention in one element of a system produces effects elsewhere, so secondary and delayed effects must be predicted before acting.

Validity means measuring the right construct; reliability means measuring it consistently. High reliability does not imply validity, and validity sets the ceiling on usefulness of any measure.

Qualitative methods answer 'why' and 'how'; quantitative methods answer 'how many' and 'how much'. Mixed-methods designs use each to check the other's blind spots.

Root-cause tools: 5 Whys for a linear causal chain, fishbone/Ishikawa for a categorised cause inventory (people, process, equipment, environment, management), FMEA to rank failure modes by severity × occurrence × detection.

Gap analysis compares current state with desired state; the gap statement must name the metric, the current value, the target value, and the evidence source for each.

Probability sampling (random, systematic, stratified, cluster) supports inference to the population; non-probability sampling (convenience, purposive, snowball) does not.

Margin of error shrinks with the square root of sample size, so quadrupling the sample roughly halves the error, all else equal.

Triangulate across at least two independent sources, methods, or reviewers before treating a finding as established.

Baseline data must be captured before the intervention; retrospectively reconstructed baselines are weaker evidence of change.

SMART objectives are specific, measurable, achievable, relevant, and time-bound; an objective without a measure cannot be evaluated later.

The scope statement fixes what is in and out of scope; unmanaged scope creep is a leading cause of schedule and cost overrun.

WBS decomposition rule: each work package should be estimable, assignable to a single owner, and completable within one reporting period (commonly 8–80 hours).

The critical path is the longest sequence of dependent tasks; float = latest start minus earliest start, and zero float identifies the critical path.

In a RACI chart exactly one person is Accountable per activity; multiple Responsible parties are allowed but each needs a defined boundary.

Risk score = likelihood × impact; response options are avoid, reduce, transfer, accept, plus exploit, enhance, and share for opportunities.

Each design criterion must trace to a requirement or stakeholder need, and each requirement needs a named verification method: test, analysis, inspection, or demonstration.

Resource levelling trades schedule for stable resource use; resource smoothing holds the end date fixed and only reallocates work within available float.

Scope, schedule, and cost baselines are approved before execution; all variance is measured against them and every change routes through change control rather than informal adjustment.

Variance formulas: cost variance = earned value − actual cost; schedule variance = earned value − planned value; negative values indicate adverse performance.

Quality assurance is process-focused and preventive; quality control is product-focused and detective. Both are required and neither substitutes for the other.

Change control decision rule: assess impact on scope, schedule, cost, risk, and compliance before approval, then record the decision and update the affected baseline.

Milestones are zero-duration checkpoints with defined acceptance criteria; a milestone counts as met only when the criteria are evidenced.

Contingency reserves cover identified and assessed risks; management reserves cover unidentified risks and normally require sponsor approval to release.

Document control requires a unique identifier, version, date, author, approver, and change history for every controlled document.

Handover of a deliverable requires documented acceptance against stated criteria, not verbal sign-off.

Hierarchy of controls, most to least effective: elimination, substitution, engineering controls, administrative controls, then personal protective equipment. PPE is the last line, never the first.

Risk = likelihood × severity; residual risk is assessed after controls are applied and the result must be recorded alongside the original assessment.

Nonconformity sequence: contain the effect, correct the defect, investigate the root cause, implement corrective action, then verify effectiveness.

A regulation is legally binding and enforced by an authority; a standard is a voluntary consensus document that binds only when adopted by contract or regulation; guidance is non-binding advice.

An audit systematically examines a management system against criteria using sampling and documented evidence; an inspection directly examines a specific item, location, or activity.

Management-system standards share a common structure: scope, normative references, terms, context of the organisation, leadership, planning, support, operation, performance evaluation, improvement.

Records must be legible, traceable, retrievable, and retained for the defined period; the retention obligation rests on the organisation holding them.

Incident-reporting thresholds should be set on actual or potential severity, not solely on whether harm actually occurred.

Informed consent requires four elements: capacity to decide, disclosure of purpose, risks, benefits and alternatives, voluntariness, and documented agreement.

Conflicts of interest are disclosed and managed; where management is insufficient to protect the decision, the professional recuses from it.

Common confidentiality exceptions are legal compulsion, imminent risk of serious harm, and authorised oversight; each exception exercised should be documented.

Practise only within demonstrated competence; where a matter exceeds it, obtain supervision, consult, or refer onward rather than proceeding alone.

Data minimisation: collect only what the stated purpose requires, retain only as long as needed, and restrict access to those who need it to do the work.

Record integrity rule: never alter or delete an original entry; corrections are added as a dated, attributed, additive note that preserves the original.

Ethical decision test: is the action lawful, is it consistent with the applicable code, would it survive disclosure to those affected, and are affected parties' interests represented?

Stakeholder grid: high power/high interest → manage closely; high power/low interest → keep satisfied; low power/high interest → keep informed; low power/low interest → monitor.

A communication plan specifies, per audience, the message, format, channel, frequency, owner, and feedback mechanism.

Tailor content and channel to audience technical literacy; the same facts need different framing for executives, practitioners, and the public.

Close the loop: every significant message needs confirmation of receipt and understanding, not merely transmission.

Escalation triggers are defined in advance by threshold (cost, time, risk, safety) rather than by in-the-moment judgement.

Minutes should record decisions, actions, owners, and due dates; verbatim transcripts of discussion are not the record of value.

Feedback should be designed as two-way: schedule review points with affected stakeholders instead of only broadcasting updates.

A KPI needs a specific definition, a data source, a target, and a review cycle; without a target it is monitoring data, not a performance indicator.

Leading indicators predict future results (training completion, defect rate at intake); lagging indicators confirm results already realised (incident rate, satisfaction score).

Formative evaluation runs during delivery to adjust the work; summative evaluation runs at the end to judge overall merit and worth.

Control-chart rule of thumb: a point beyond a control limit, or a run of seven consecutive points on one side of the mean, signals a non-random cause.

Benchmarking is valid only when definitions and measurement methods match between the organisation and the comparison reference.

Improvement cycle: measure baseline → identify gap and causes → select intervention → implement → re-measure → standardise if effective, or iterate if not.

ROI = (benefit − cost) ÷ cost, expressed as a percentage; always state the time period and whether benefits are monetised or estimated.

Lessons learned must be captured at defined checkpoints during the work, because at closure participants have dispersed and recall is degraded.