To align observation start and stop rules for IOA, synchronize clocks, define the shared start and stop events, predeclare pause and interruption rules, and record each observer's actual exposure. Calculate agreement only across matched valid units. Keep late arrivals, lost visibility, device failures, and excluded minutes visible so a shorter shared window never appears to represent the full session.

Choose observable boundary events

Use an event such as the activity starting, the timer sounding, or both observers confirming readiness.

Synchronize time

Record device clocks, time zone, interval timer, and any drift or manual correction.

Predeclare pauses

Define how bathroom breaks, emergencies, lost visibility, technical failures, and interruptions affect the window.

Retain unmatched exposure

Show planned, observed, shared-valid, and excluded duration with reasons.

Build a window reconciliation table

Create one row for every planned interval or clock segment. Record whether Observer A had valid exposure, Observer B had valid exposure, both shared the same condition, and the segment enters the agreement calculation. Add an exclusion reason, responsible owner, and source timestamp. This table makes partial exposure auditable, prevents a missing segment from becoming agreement, and lets the reviewer reproduce both the coverage percentage and the exact units used in IOA.

Reconcile clocks and boundary events

Store the planned start, actual start, planned stop, actual stop, time zone, and device clock for each observer. A one-minute device difference can move an event into a different interval. Record any correction without overwriting the original timestamp. For human-entered times, use the shared observable boundary event as the primary reference and the clock as supporting evidence.

Define how partial intervals are handled before scoring begins. The plan may discard a partial interval, retain it as a shorter named unit, or use another method appropriate to the measurement system. Apply the same rule at late arrivals, pauses, resumptions, and early stops. Silent prorating creates a denominator that another reviewer cannot reproduce.

Keep coverage and agreement separate

Cora's four late-arrival minutes plus three interruption minutes account for all seven excluded minutes. The reconciliation is 23 shared-valid + 4 B-absent + 3 both-without-visibility = 30 planned minutes. Coverage is therefore 23 of 30, or 76.7%. Agreement uses only eligible units inside the 23-minute shared-valid window and requires its own numerator and denominator.

The 76.7% figure measures matched exposure, not observer agreement. Report both values together so a high agreement score cannot obscure limited observation coverage.

Audit transitions and partial units

Review the unit immediately before and after every start, stop, pause, and resumption. Confirm both observers saw the same condition and applied the same boundary rule. Preserve clock drift, device failure, bathroom or health pauses, and lost visibility as explicit states. When the first or last unit cannot be reconciled, hold that unit outside the agreement calculation and explain why.

Apply the field control to one defined question

The matched-window worksheet for Cora begins with the exact decision, the responsible qualified role, and the evidence needed. Preserve original observer records, matched exposure, the selected calculation, raw numerator and denominator, unrounded result, and every excluded or unresolved unit.

Work the example for Cora

Cora's sample is scheduled for 30 minutes. Observer B arrives four minutes late, and both lose visibility during a three-minute interruption. The matched valid window is 23 minutes. Report 23 of 30 minutes, or 76.7% coverage, and calculate IOA only within those 23 minutes.

Audit the denominator for Cora

The denominator for Cora is the eligible cohort defined before review. Report planned, due, observed, matched, invalid, missing, and open units separately. A percentage gains meaning from the unit, coverage, dates, conditions, and disagreement pattern.

Protect access and safety during the sample for Cora

The observation for Cora keeps AAC, interpreters, mobility, food, water, bathroom use, prescribed care, health supports, rest, relationships, and emergency help available. Observer procedure never delays immediate safety or mandated action. Record changes in access, health, assent, distress, visibility, staff behavior, or routine.

Use current sources within their scope for Cora

For Cora, the BACB ethics hub points to current ethics materials. The CASP public summary supplies high-level autism-treatment scope. The BACB Ethics Code addresses competence, client involvement, consent and assent when applicable, documentation, supervision, risk, and evaluation for covered behavior analysts. The BCBA Test Content Outline includes measurement, reliability, sampling, and data-based decisions as examination content.

For Cora and the matched observation exposure question, Vollmer, Sloman, and St. Peter Pipkin discuss practical implications of data reliability and treatment-integrity monitoring. Essig, Rotta, and Poling review IOA and fidelity reporting in an identified research corpus. Reed and Azulay describe agreement calculations and a calculation tool. ASHA says AAC users should always have access to their tools or devices. The responsible clinician still selects a case-specific measurement and review plan.

Close the IOA review for Cora

Review the matched-window worksheet with Cora through accessible communication and with the responsible clinician. Record the question, sample, method, arithmetic, disagreement types, limits, repair, owner, next observation, and review date. Reopen the review when the definition, observer, setting, system, exposure, access, prevalence, procedure, or decision changes.

Before release, verify that every planned minute or unit appears exactly once in the window table and that included plus excluded exposure equals the planned sample. Confirm that the agreement calculation points to the same shared-valid units. Any unexplained difference remains an open reconciliation item.

Related resources

Sources