When to use it
Use this sheet for discrete trial teaching (DTT). Each trial has a clear discriminative stimulus (SD), a response from the learner, and a consequence, followed by a short intertrial interval. Recording every trial shows you, target by target, how independent responding is developing and how much prompting is still needed.
Some programs use first-trial (cold probe) data instead, where only the first trial of each target in a session is scored. Probe data is faster to collect and can predict mastery well. Trial-by-trial data gives more detail on prompt dependence and within-session learning. Many practices use both: trial-by-trial during acquisition and probes for mastery or maintenance checks. Follow the method written in the learner’s program.
How to fill it in
- Write the program details at the top. Include the program name, the SD (exact instruction), the target response definition, the prompt strategy (for example, errorless, most-to-least, least-to-most, or prompt delay), the reinforcement schedule, and the mastery criterion.
- One row per target. Write the target in the first column (for example, “touch dog,” “mand: juice”).
- Deliver the SD and wait for the response window in the program (often 3 to 5 seconds, but follow the written program).
- Record each trial in the next box using the codes below.
- + correct and independent (no prompt)
- - incorrect or no response within the response window
- P prompted correct. If your program tracks prompt level, write the level instead (for example, G, V, M, PP, FP).
- Follow the consequence steps in the program: reinforcement for correct responses, and the error-correction procedure after an incorrect response. Error-correction trials are usually recorded as prompted, but check your program’s rule.
- Total each row at the end of the session: number of independent correct responses, number of trials, and percent correct.
Common mistakes:
- Scoring prompted responses as correct. For mastery decisions, most programs count only independent correct responses. Prompted trials show how much support was needed, not what the learner can do alone.
- Delivering the SD differently across trials or staff (for example, “Touch dog” one time and “Where’s the dog?” the next) without a planned reason.
- Filling in data at the end of the session from memory.
- Running the same target many times in a row (massed trials) when the program calls for interspersal, or the reverse.
The sheet
Client:Date:Instructor:Setting:
Program:SD:Prompt strategy:Reinforcement schedule:
Mastery criterion:
Code: + independent correct | - incorrect or no response | P prompted (or write the prompt level: G, V, M, PP, FP)
| Target | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | Independent correct / trials | % correct |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
Session notes (reinforcers used, behavior, changes to try next session):
Summarizing the data
Percent correct (independent) = independent correct responses divided by total trials for that target, times 100.
Worked example: target “touch dog,” 10 trials: + + P + - + + + P +
- Independent correct: 7
- Prompted: 2
- Incorrect: 1
- Percent correct = 7 / 10 x 100 = 70%
If the program’s mastery criterion is “80% or higher independent across 2 consecutive sessions with 2 different instructors,” this session does not count toward mastery.
Mastery criteria vary. Common examples include 80% to 90% independent across two or three consecutive sessions, sometimes with a requirement for more than one instructor or setting. Each program should state its criterion in writing before teaching begins. Some research suggests higher criteria may support better maintenance, but the right criterion depends on the skill and the learner.
It can also help to graph percent prompted alongside percent independent. A falling prompted line with a flat independent line can point to prompt dependence.
Tips
- If percent correct is flat or falling for three or more sessions, flag it for the supervising BCBA. Common causes are reinforcer quality, prompt fading steps that are too big, an unclear SD, or target sets that are too similar.
- Mix mastered targets in with acquisition targets to keep reinforcement rates up and reduce errors.
- Record data right after each trial. It’s one mark, and it’s much more accurate than recording from memory.
- Once a target meets mastery, move it to a maintenance schedule and probe it at the intervals in the program, rather than dropping it.