Consequence given no response probability equals valid nonresponse units containing the defined consequence within the selected window divided by all valid nonresponse units. Define what counts as a response opportunity, the complement state, window, and overlap rule. Keep raw cells and coverage. This comparison helps separate response-linked occurrence from the consequence's ordinary availability in the sampled environment.

Make "no response" an observable comparison state

The complement must be more precise than everything left over after a target response is coded. Define a valid unit in which the response had a reasonable chance to occur and could be observed. Then classify the target response as present or absent. A unit with blocked visibility, an unavailable communication system, no relevant opportunity, or an interrupted clock is missing or ineligible. Calling any of those units "no response" turns a measurement problem into apparent client behavior.

The response and consequence definitions also need compatible clocks. If a consequence is counted within a fixed interval after a response, state what starts the comparison window for a nonresponse unit. Interval-based ABC data can use identical interval boundaries when that fits the question. Event-based records may require predeclared control windows sampled independently of the outcome. Select the comparison worklist before revealing which windows contain the consequence.

Build the four raw cells before calculating

Create one row for each valid unit with response present or absent and consequence present or absent. Preserve the timestamp, setting, opportunity, window eligibility, observer, and any uncertainty code. The four counts should reconcile to the valid cohort:

  • response and consequence;
  • response and no consequence;
  • no response and consequence; and
  • no response and no consequence.

This table catches denominator errors early. It also lets another reviewer reproduce P(consequence | response), P(consequence | no response), and a later contingency difference without recoding narrative notes. Keep invalid, missed, and truncated units in a separate coverage table rather than forcing them into one of the four cells.

Calculate Yara's nonresponse probability

Yara has 60 valid intervals under one interval and consequence rule. The target response occurs in 12. The consequence follows the response in 9, leaving 3 response intervals without the consequence. The other 48 intervals are valid nonresponse units. Six contain the consequence and 42 do not.

For the requested conditional probability, the denominator is all 48 valid nonresponse intervals:

P(consequence | no response) = 6 / 48 = 0.125, or 12.5%.

The response-linked comparison is 9 / 12 = 0.75, or 75%. Both fractions belong in the record because 75% and 12.5% conceal very different exposure counts. Yara's cells reconcile as 9 + 3 + 6 + 42 = 60. The calculation describes the sampled units and does not identify why the pattern occurred.

Check exposure, coverage, and comparability

Report how the 60 valid units relate to the planned observation frame. If 70 units were planned and 10 lacked visibility, coverage would be 60 / 70, or 85.7%, while the conditional denominator would remain 48. A high valid-unit percentage can still be unrepresentative when the missing units cluster in one routine, staff condition, time, or access state. Review coverage by context instead of relying on one overall percentage.

Response and nonresponse windows should be comparable in length, opportunity, observation quality, and consequence code. Overlapping windows can count one consequence for several anchors unless the protocol says how to assign it. Repeated responses may reset a window, be grouped into an episode, or make a unit ineligible. Choose the rule before calculation and test it on sample records.

Handle zero, missing, and truncated states

If valid nonresponse units exist but none contains the consequence, the result is 0 divided by that positive denominator, or zero. If there are no valid nonresponse units, the probability is undefined, not zero. A unit that ends before its consequence window closes is truncated. Exclude or score it under a predeclared partial-window method and show that decision in coverage. Uncertain consequence codes remain unresolved until adjudication; they are not automatically absent.

Sparse data deserve visible caution. A single consequence among two nonresponse units produces 50%, but that fraction has little stability. Report session and context distributions, raw counts, and repeated observation rather than treating the percentage as a stand-alone threshold.

Interpret the pattern without assigning function

The 12.5% result estimates how often the defined consequence appeared during valid sampled nonresponse units. Its comparison with 75% after response may identify a sequence worth further assessment. Descriptive association can also reflect routine schedules, staff behavior, shared antecedents, event duration, coding rules, or unequal opportunities. It does not demonstrate reinforcement, prove a functional relation, or authorize treatment.

Clinical decisions belong to the qualified clinician using the full assessment, direct participation from Yara, health and safety information, implementation evidence, and other relevant data. Observation should never create, delay, or withhold food, water, bathroom access, medical care, communication, mobility, rest, relationships, or emergency support to complete a window.

Graph and report the denominator clearly

A four-cell table is the clearest audit display. A second graph can show P(consequence | response) and P(consequence | no response) as paired points or bars, with 9/12 and 6/48 printed beside them. Include coverage and the observation dates in the caption. Do not use a bar alone when its denominator could be mistaken for time, trials, or people.

Suitable report wording is: "Across 60 valid intervals, the consequence occurred after 9 of 12 response intervals (75.0%) and during 6 of 48 nonresponse intervals (12.5%). Ten planned intervals were excluded for recorded observation limitations [if applicable]. These descriptive probabilities summarize the defined sample and do not establish behavioral function." Name the actual exclusion reasons rather than inserting a generic missing-data phrase.

Review access and meaning with Yara

Show Yara the question, event definitions, sample records, result, and proposed next step in an accessible form. Use Yara's established communication method, allow processing time, and record corrections, assent, dissent, discomfort, and priorities. ASHA's AAC guidance supports continuous access to communication tools and devices. A partner's delayed interpretation or a device access barrier should be labeled as context, not recoded as no response.

Store only the source detail needed for review. Limit access to identifiable ABC records, label observer notes separately from calculated fields, and preserve the raw-to-summary audit trail. A summary percentage should not erase Yara's account of what occurred.

Keep the evidence in its proper role

The BACB ethics resources, CASP practice-guideline summary, and BCBA Test Content Outline frame professional assessment, measurement, and review responsibilities. Studies of antecedent and consequent predictors, contingency-space analysis, and response-stimulus sequences support retaining conditional and background cells. Research on precursor sequences illustrates direction-specific denominators, while comparative descriptive methods reinforces the nonexperimental boundary. None supplies a universal clinical cutoff for Yara.

Clinician review checklist

For consequence given no response probability, confirm the decision question, response opportunity, response and consequence definitions, comparison anchor, window length, overlap rule, planned units, valid units, four cells, missing and truncated states, 6/48 arithmetic, rounding, context coverage, graph label, report wording, accessible review, safety actions, qualified decision owner, next step, and review date. Recalculate from raw cells before releasing an updated summary.

Practical limits

This conditional probability depends on the selected unit, complement, timing rule, sample, and observation quality. It cannot recover unseen events, correct a weak response definition, show consequence duration or quality, establish treatment integrity, predict an individual event, or prove cause. Changes across phases may reflect exposure or coding changes. Preserve raw records and repeat assessment when the comparison state, setting, access conditions, observers, or clinical question changes.

Related resources

Sources