Key takeaways
- Ten trials per session is a common default because each trial is worth 10 points and the math is easy.
- With 5 trials, one answer moves the percentage 20 points, so small sessions swing hard.
- Combine sessions by adding correct and total trials, not by averaging percentages.
- "80% across 3 consecutive sessions" usually means each of three sessions in a row, not an average of three.
- Use a yes/no sheet when the skill comes up once per occasion instead of in repeated trials.
Trial-by-trial data records each attempt at a skill as correct or incorrect, then reports the result as correct out of total, such as 7 of 10, or 70%. It is one of the most common ways IEP academic and communication goals are measured. The practical questions are how many trials to run, how to turn sessions into a percentage without distorting it, and what a criterion like "80% accuracy across 3 consecutive sessions" actually asks. The short answers: about 10 trials per session if you can, combine sessions by adding trials rather than averaging percentages, and read "3 consecutive sessions" as three separate sessions in a row that each meet the bar.
What counts as one trial
A trial is one chance for the student to show the skill, scored against a definition decided in advance. It has three parts: the cue (the question, card, or direction), the student's response, and your score. "Show me the main idea" followed by a response is one trial. A whole worksheet is not one trial; each item on it is.
Before the first session, write down two things:
- What a correct response looks like. "Names the main idea in their own words" and "points to the main idea sentence" are different skills.
- Whether a prompted response counts. If the goal says "independently," many teams score only independent responses as correct. Prompt hierarchy and prompt-level data covers how to record the level of help without losing the correct-or-not score.
How many trials per session
There is no single required number, and districts and teams vary. A few practical points hold regardless:
| Trials per session | One trial is worth | What that means |
|---|---|---|
| 4 | 25 points | A single slip moves the student from 100% to 75%. Very noisy. |
| 5 | 20 points | 80% means 4 of 5. Workable if you run sessions often. |
| 10 | 10 points | A common default. Easy math, and one error does not swing the result much. |
| 20 | 5 points | Steady numbers, but hard to fit into a lesson and tiring for many students. |
Keep the number of trials about the same from session to session, and keep the difficulty constant. If Monday's ten items are easier than Thursday's, the percentage is measuring the items, not the student. Most importantly, record the trials you actually ran. A session of 6 trials is real data. Six trials rounded up to ten is not.
Calculating percent correct
For one session: correct ÷ total × 100. Eight correct out of 10 is 80%. Six out of 8 is 75%.
For a week, a month, or a reporting period, add the trials, then divide. Do not average the session percentages. Averaging gives a 4-trial session the same weight as a 20-trial session.
Two sessions in one week on a sight-word goal:
| Session | Correct | Total | Percent |
|---|---|---|---|
| Tuesday | 4 | 4 | 100% |
| Friday | 5 | 10 | 50% |
Average of the two percentages: (100 + 50) ÷ 2 = 75%
Pooled: (4 + 5) ÷ (4 + 10) = 9 ÷ 14 = 64%
The student answered 14 items and got 9 right. The pooled figure, 64%, is what happened. The average makes the week look 11 points better because a short session counted as much as a full one. When Evident builds monthly accuracy for a progress narrative, it pools trials this way, adding correct and total trials across the month before dividing.
Reading "80% across 3 consecutive sessions"
Mastery criteria like this are everywhere in IEP goals, and they hide several decisions. Take them apart piece by piece.
- 80% is the bar for a single session. With 10 trials that is 8 or more. With 5 trials it is 4 or more. With 8 trials it is 7 or more, since 6 of 8 is only 75%.
- Across 3 sessions most commonly means each of the three sessions meets the bar. An average of three sessions is a looser standard: 100%, 100%, and 50% average 83% but include a session well below the bar.
- Consecutive means in a row. One session below 80% resets the count. "3 of 4 sessions" is a different, more forgiving criterion.
- Sessions is not the same as days or weeks. If a goal says "across 3 consecutive weeks," you need to decide how sessions within a week combine; pooling that week's trials is the usual approach.
If the goal wording does not settle these, agree on a reading as a team and write it on the data sheet, so the person who declares mastery in May reads it the same way as the person who wrote it in September.
A composite second grader we will call Jordan. Goal: "Jordan will solve two-digit addition problems with regrouping with 80% accuracy across 3 consecutive sessions." The team reads it as each session at 80% or higher, three in a row.
| Session | Correct | Total | Percent | Meets 80%? | Run count |
|---|---|---|---|---|---|
| 1 | 5 | 10 | 50% | No | 0 |
| 2 | 6 | 10 | 60% | No | 0 |
| 3 | 8 | 10 | 80% | Yes | 1 |
| 4 | 7 | 10 | 70% | No | 0 |
| 5 | 8 | 10 | 80% | Yes | 1 |
| 6 | 9 | 10 | 90% | Yes | 2 |
| 7 | 6 | 8 | 75% | No | 0 |
| 8 | 8 | 10 | 80% | Yes | 1 |
| 9 | 9 | 10 | 90% | Yes | 2 |
| 10 | 10 | 10 | 100% | Yes | 3 |
Criterion met at session 10.
Two traps show up here. Session 7 had only 8 problems because an assembly cut it short. At 6 of 8 it fell just below the bar, and the run reset. That is correct under this criterion; the team recorded the real total rather than guessing what 10 would have looked like. Second, if the team had used a three-session average, sessions 5 to 7 average 81.7% and would have "met" the goal at session 7, even though session 7 itself was below 80%. Deciding the reading in advance is what keeps those two outcomes from being argued after the fact.
Probes vs teaching trials
Many teams separate two kinds of trials, and it helps to decide which ones count as data.
Teaching trials happen during instruction, with prompts, feedback, and corrections. They are where the learning happens, and scores from them tend to run high because help is built in.
Probe trials are a short, separate check, often the first few trials of a session or a quick set at the end, run without prompts or feedback. They show what the student can do on their own that day.
A common approach is to record probe trials as the goal data and use teaching trials only for your own planning. If you do record teaching trials, write the prompt level for each one so the data does not overstate independence. Either way, use the same approach every session. A September of teaching-trial data and a January of probes will look like a drop that never happened.
Trial data or yes/no data?
Not every skill comes in repeated trials. If the skill comes up once per natural occasion, such as following the arrival routine, using a break card when upset, or bringing materials to class, you cannot run ten trials of it in a session. Record one yes or no per opportunity on the yes/no data sheet and report the percent of opportunities met.
The math is the same: yes count divided by opportunities. The difference is where the denominator comes from. On a trial sheet, you decide how many trials to present. On a yes/no sheet, the day decides how many opportunities there were, so write each one on its own line rather than summarizing a day as a single yes.
Setting up the sheet
The free trial data sheet has columns for activity, correct, total, prompt level, and notes. You can type in up to three goals before printing, and the QR code on the printed sheet reopens it with the same goals next time. A few habits make it hold up at the annual review:
- One line per session. Two sessions of 4 trials are not the same as one session of 8, so keep them on separate lines.
- Write the total even when it is 10. A blank total forces someone to guess later.
- Leave the percentages for later. Do the math at your desk, not at the table.
- Date every line. Undated rows cannot become a progress report.
If you would rather log a session from the laptop you are already teaching on, the Evident Capture Chrome extension records correct and total trials, with an optional prompt level, as a single data point. When enough sessions are in, graphing IEP progress data shows how to plot them against an aim line, and the measurable goal wording bank has examples of criteria written clearly enough to score.
Common questions
How many trials should I run per IEP data session?
Many teams run 5 to 10 trials per session, and 10 is a common default because each trial is worth 10 percentage points. What matters most is running about the same number every session and recording the number you actually ran.
What does 80% accuracy across 3 consecutive sessions mean?
It most commonly means the student scored 80% or higher in each of three sessions in a row. An average of 80% across three sessions is a looser standard, because one strong session can cover for a weak one. If the goal does not say which, the team should agree and write it down.
How do I calculate percent correct for a month of sessions?
Add all correct trials, add all total trials, and divide. Averaging the session percentages gives a small session the same weight as a large one and can misstate the month.
When should I use a yes/no data sheet instead of a trial sheet?
Use yes/no when the skill comes up once per natural occasion, such as following the arrival routine or using a break card when upset. Use a trial sheet when you present the skill repeatedly in a session and score each attempt.