How to take trial-by-trial data during instruction
By the IEP Atlas editorial teamLast updated
Short answer
To take trial-by-trial data, define what one trial is, decide how many trials each session gets, and mark every trial the moment it ends with a code: + for independent, a prompt code for prompted, and minus for incorrect or no response. After the session, divide independent trials by total trials. Graph that one number per session.
Trial-by-trial data records the result of every single learning trial, one box per trial. It is the standard record for skills taught in short, repeated chances: naming, matching, answering, producing a sound, following a direction. The method is simple. The hard part is doing it while you teach, without slowing the lesson or scoring from memory at lunch.
This guide covers the mechanics during instruction. For how this method compares with the others, see the IEP data collection guide. For the difference between trials you present and chances that come up on their own, see percent accuracy vs percent of opportunities.
What you need before you start
- The goal, with its criterion copied onto the sheet ("8 of 10 trials, independent, 3 consecutive sessions").
- A target list: the exact items, such as 10 picture cards or 5 two-step directions.
- A prompt code list everyone uses. The prompt hierarchy examples give standard codes.
- An opportunity trial data sheet, one row per session.
- A way to mark fast: a pen on a clipboard, a tally counter, or a sticky note on your sleeve.
Step by step
Step 1: Define one trial
A trial has three parts: your cue, the student's response, and what you do next. Write all three down. "Cue: show a card and say 'What is it?' Response: names the picture within 5 seconds. Next: praise and the next card, or a model and the next card."
The 5-second window matters. Without it, one adult waits 20 seconds and another waits 3, and the same student scores differently.
Step 2: Fix the number of trials
Pick a set number per session, usually 10 for one target set. Ten makes each trial worth 10 percentage points and keeps the math easy. Fewer than 5 makes each trial swing the score too far. More than 20 tires most students and rarely adds information.
Keep the count the same every session. A score of 6 of 8 one day and 9 of 12 the next is hard to compare on a graph.
Step 3: Code every trial when it ends
Mark the box before you present the next cue. Use one code per trial:
| Code | Means | Counts toward an independent criterion |
|---|---|---|
| + | Correct, no help, within the response window | Yes |
| G | Correct after a gesture prompt | No |
| V | Correct after a verbal prompt | No |
| M | Correct after a model | No |
| PP | Correct with partial physical help | No |
| − | Incorrect, even after prompting | No |
| NR | No response within the window | No |
If the goal allows a prompt level, such as "with no more than a gesture," the team counts + and G. Write that rule on the sheet so every collector counts the same way.
Step 4: Separate teaching trials from probe trials
Teaching trials include your help, so they run high. A probe is a short set with no prompts, usually at the start of the session before any teaching. Example: 10 probe trials at 9:00, then 15 minutes of teaching with prompts.
Graph the probe, not the teaching set. The teaching data still matters. It tells you which prompt level the student needs, and whether that level is fading.
Step 5: Mix the targets
Present items in a varied order, not the same card first every time. Students learn the order. Mixing 2 or 3 known items in with new ones also keeps the session moving for a student who stalls on hard items. Mark the known items on the sheet so they do not count toward the target score.
Step 6: Summarize the session in one number
Count the + marks and divide by the total trials. Then note the most common prompt code. A row reading "+ + V + − + + M + +" is 7 of 10 independent, 70%, with verbal and model prompts on the rest.
Graph that 70%. Add the date, the target set and the collector's initials. Then check the result against the criterion row on the sheet.
Time each step takes
| Step | First time | After setup |
|---|---|---|
| Define the trial and the response window | 15 minutes | Done once per goal |
| Build the target list and the sheet | 20 minutes | 2 minutes to copy a new sheet |
| Run and code 10 probe trials | 3 to 4 minutes | 2 to 3 minutes |
| Summarize and graph the session | 3 minutes | 1 minute |
Worked example: Dante's object naming goal
Dante is a fictional 1st grader with a goal to name 10 common classroom objects from pictures. The criterion is 8 of 10 trials independent across 3 consecutive sessions. His para, Ms. Lee, runs the probe every Monday, Wednesday and Friday.
The definition. Cue: show the picture and ask "What is it?" Correct: says the object's name, or a clear version of it, within 5 seconds. Ceiling: one prompt per trial, then the next card.
Week 1 filled-in sheet (probe trials only):
| Date | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | Independent | Notes |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Mon 9/8 | + | V | − | + | M | NR | + | V | − | + | 4 of 10 (40%) | Missed scissors, glue twice |
| Wed 9/10 | + | + | V | + | M | + | − | V | + | + | 6 of 10 (60%) | First correct "stapler" |
| Fri 9/12 | + | V | + | + | V | + | + | M | + | − | 6 of 10 (60%) | Verbal prompts mostly the first sound |
What the week says. Dante moved from 40% to 60%. Most prompts were verbal, and most of those were the first sound of the word. That is a clue: he knows the object but struggles to retrieve the word. Ms. Lee adds a first-sound cue during teaching and plans to fade it.
Week 4. Monday 8 of 10, Wednesday 8 of 10, Friday 9 of 10. That is 3 consecutive sessions at criterion. The team marks the goal met for this target set under its mastery criteria and starts a maintenance probe every 2 weeks.
Measurement: probe frequency and the sheet
Probe 2 to 3 times a week for skills taught daily. Probe once a week for skills taught 2 or 3 times a week. Fewer than one probe a week makes a 3-consecutive-sessions criterion take a month to confirm.
On the opportunity trial data sheet, each box is one trial. Mark NA if a trial did not happen, such as a fire drill after trial 6. A session with 6 of 10 trials run is scored out of 6 and flagged, not scored as 6 of 10. The opportunities measurement guide covers this denominator rule in more detail.
The IEP must state how progress will be measured and when it will be reported.[1] A goal written in trials should name the number of trials, so the sheet and the goal match.
Common mistakes
- Weak: Scoring the whole set from memory after the session. Better: Mark each box before the next cue; it takes 1 second.
- Weak: Counting a prompted correct answer as +. Better: Code the prompt level and count only what the goal allows.
- Weak: 10 trials one day and 6 the next. Better: Fix the count and mark NA for trials that did not happen.
- Weak: Graphing teaching trials. Better: Graph the unprompted probe at the start of the session.
- Weak: Same card order every day. Better: Shuffle the cards or follow a printed random order.
- Weak: No response window. Better: Write "within 5 seconds" on the sheet and use it every trial.
Tools for the next step
The trial-by-trial entry in the glossary defines the term for team members. If you run trials for several students at once, the guide to group data collection shows how to rotate turns and keep one sheet per group. For checking that two adults score trials the same way, see interobserver agreement.
Frequently asked questions
What is the difference between trial-by-trial data and opportunity data?
In trial-by-trial data you present every trial, so you set the count, such as 10 cards. In opportunity data, the chances come up during the day, such as a peer greeting the student. Both use the same sheet, but trials are planned and opportunities are counted as they happen.
How many trials should I run per session?
Ten is a good default for one target set. It keeps the math simple and takes 2 to 4 minutes. Use at least 5 so one trial does not swing the score too far, and rarely more than 20.
Do prompted trials count as correct?
Only if the goal says so. Most goals count only independent responses. Code every prompted trial with its prompt level anyway, because that record shows whether the student needs less help over time.
Can a para take trial-by-trial data?
Yes. Train the para on the trial definition, the response window and the codes, then score one session side by side and compare. Once you agree on nearly every trial, the para can collect alone.
Use it in IEP Atlas
- Opportunity trial data sheet
- Opportunities measurement guide
- Trial by trial data in the IEP glossary
- Free IEP data sheets
- Data sheet in the IEP glossary
Related guides
- IEP data collection: methods, sheets and schedules that hold up
- Prompt hierarchy examples for IEP data
- Collecting data on several students during group instruction
- Percent accuracy vs percent of opportunities: what the denominator means
- Cold probe data vs trial-by-trial data
Sources
- 34 CFR §300.320: Definition of individualized education program, Legal Information Institute, Cornell Law School
- Progress Monitoring: Data & Goal Setting, National Center on Intensive Intervention
Written and checked by the IEP Atlas editorial team under our editorial policy. This guide describes general practice and federal rules. It is not legal advice; your state may add requirements, and the IEP team makes the decisions for each student.