Factors Impacting Data Reliability: Rate and Total Behaviors

Workman et al. (2025) watched new staff try to record behavior. They looked at two things: how many behaviors the sheet asked for, and how fast the behavior happened.

The goal was to see when rookie observers start making shaky data. No kids or clients were treated; this was pure measurement homework.

More items on the data sheet lowered reliability. Very fast responses also lowered reliability. The team did not give numbers, but the pattern was clear: keep it short and slow or expect errors.

LeBlanc et al. (2003) showed that rich reinforcement makes both rate and accuracy go up. Workman flips the coin: high rate later hurts the person holding the clicker. Same variable, two opposite headaches.

K et al. (1994, 1995, 1996) tracked how responding drifts within a session when schedules are dense. Their lab work warns that rate jumps around; Workman warns that such jumps are hard for new staff to count.

Together the papers make a simple chain: dense schedules raise rate, raised rate invites observer error.

When you write a program with five target behaviors or a vocal that can top 200 responses per minute, split the job. Give the new RBT one behavior, or add a second observer, or use a timer to break the session into short chunks. Clean data early, clean data always.