Rate

Calculate count per unit of time

The formula is:

Rate = response count divided by valid observed time.

If 12 responses occur during 30 minutes, the rate is 12 divided by 30, or 0.4 responses per minute. This can also be expressed as 24 per hour. State that 24 is a conversion if a full hour was not observed.

Rate makes unequal sessions comparable

Ten responses in 20 minutes equal 0.5 per minute. Fifteen in 60 minutes equal 0.25 per minute. The longer session contains the higher count and lower rate.

Raw count can still matter for workload, total access, or absolute exposure. Report count, valid observation time, and rate together.

Define the time denominator

Decide whether the denominator is the whole session, time when the response could occur, active task time, or another defined exposure. Excluding breaks can be appropriate for one question and misleading for another.

Predefine treatment of late starts, interruptions, blocked observation, travel, device downtime, and missing records. A denominator selected after seeing the data can distort comparison.

Opportunity may explain a rate

A response can occur only when relevant conditions arise. Help requests may be low because few problems occur. Conversation turns depend on partners. Completed deliveries depend on assigned items.

Report eligible opportunities and partner performance when they help interpretation. Rate alone cannot separate limited opportunity from limited responding.

Compare like exposure conditions

Rates from school, home, clinic, and community settings may differ because opportunity, activity, partners, observation coverage, and response effort differ. Segment the data before pooling. If the goal is general use across settings, show each setting and a combined result with explicit weights. An overall rate dominated by a long clinic observation can hide a small community sample.

Keep calendar time distinct from actual observed time. A week with two observed hours does not provide a response-per-week rate unless the collection plan defines and supports that denominator.

A fictional community example

Dev is a fictional 16-year-old who wants to participate in a makerspace and ask for missing tools through speech or AAC. During three 40-minute observations, Dev makes 4, 7, and 5 independent requests. Total count is 16 across 120 minutes.

Pooled rate is 16 divided by 120, or 0.13 requests per minute, equal to 8 per hour after rounding. Nine actual missing-tool opportunities occur, and Dev requests in 8 of 9, or 88.9%.

Partners respond within 20 seconds to 6 of 8 requests, or 75%. Rate, opportunity percentage, and partner response describe different parts of the routine. They support review and no causal conclusion.

Select a useful reporting unit

Choose a unit readers can understand and that avoids tiny decimals. Per minute may fit rapid responses. Per hour may fit work routines. Per day may fit rare events when observation coverage is clearly defined.

Do not compare “0.2 per minute” with “12 per hour” until converting them to the same unit. Store the original count and time so software conversions can be checked.

Direction needs context

A higher rate can indicate greater access, fluency, participation, or repeated difficulty. A lower rate can indicate reduced interruption, fewer opportunities, fatigue, unavailable communication, or greater efficiency. Direction is useful only in relation to the person’s goal.

Review accuracy, quality, effort, comfort, independence, prompt use, health, and the person’s report. Speed can be harmful when it pressures processing or sacrifices safety.

Rare and bursty events need care

An average rate can hide clusters. Three events in the first minute and none in the next 29 produce 0.1 per minute across 30 minutes. Preserve timestamps or interresponse times when clustering matters.

For rare health or safety events, state the observation coverage and individual events. A small denominator can make rate appear unstable.

Protect access and measurement integrity

Keep AAC, interpreters, mobility and sensory supports, breaks, food, water, bathroom access, prescribed care, relationships, and emergency help available. Missing supports change the conditions and should be recorded.

Useful quality checks include valid observed minutes divided by planned minutes, records with complete count and time, clock failures, exclusions by reason, and sampled observer agreement.

The Association for Behavior Analysis International basic-principles page situates measurement in higher-education behavior-analysis content. The BACB BCBA Test Content Outline, Sixth Edition covers occurrence, temporal dimensions, validity, reliability, and representative measurement selection. Neither source sets an individual rate target.