How School SLPs Collect Data During Group Therapy

1. Define the measure before the group arrives

A shared game or book can create opportunities for several goals. It does not make every turn an equivalent trial. Start with the exact goal, the condition, the response that counts and the support permitted. Choose one measure you can collect consistently during that activity.

For an accuracy goal, write what counts as a scored opportunity and a correct response. For a frequency goal, define the event and observation period. For a rubric, keep the descriptors available. Do not convert every goal to percent correct just because a tally sheet has a percentage box.

Set a comparable starting point with the baseline tracker. Record task level and support alongside the number so a later probe can use the same conditions.

2. Use one activity with different targets

These are illustrative scoring plans, not universal targets or individualized treatment recommendations. Adapt them to each student’s IEP, access needs and communication preferences.

• Articulation: during a picture game, probe a specified sound in a specified word position and task level. Count productions meeting the goal’s definition out of actual target attempts. A sentence production is not interchangeable with an isolated word. Note models or cues.
• WH questions: during a shared story, select a defined question type and comparable items. Track correct responses out of presented opportunities, recording whether choices, a visual cue or repetition were provided. Keep who and why results separate if their demands differ.
• Pragmatics: define an observable, meaningful communication behavior—such as responding to a peer’s invitation using the student’s available communication mode. Count responses out of identified opportunities, or count events during a defined period if that matches the goal. Avoid vague ratings such as “good participation” and do not equate eye contact with engagement.
• AAC: choose a communication function and acceptable response modes, such as requesting a turn with the available AAC system. Record whether the system was accessible and whether modeling or prompts occurred. Allow response time. A missing device or inaccessible materials are context barriers, not proof of inability.

A language sample may be useful for a language goal, but MLU is not a measure of every group target. If appropriate, use the reviewable MLU calculator and analyze language beyond utterance length.

3. Separate a probe from teaching

A probe samples performance under defined conditions. Teaching may include models, feedback, repeated practice and more support. Both are valuable; they answer different questions. Label them before tallying so coached practice does not silently become an independent baseline.

For each opportunity, record the response and the support: independent, visual cue, verbal cue, model or another defined condition. “Prompted” without a type or amount may be too broad to compare sessions. Use the goal’s actual support criterion rather than a universal prompt hierarchy.

If the goal permits one verbal cue, a successful response after that cue may meet that goal. It still should not be relabeled independent. If you change the prompt level or task difficulty, mark the change and interpret the results accordingly.

4. Rotate short probes without losing the group

1. Before the session, assign each student a brief probe window and a target within the shared activity.
2. Probe one student while the others have meaningful participation opportunities. Record that student’s trials immediately, not from memory after the game.
3. Rotate to the next student’s target. Do not score incidental turns unless they meet your prewritten opportunity definition.
4. Move into teaching and label those trials separately. Note changes in cueing, materials or task level.
5. Close with a quick record check: student, goal, raw counts, support and whether the planned probe actually occurred.

There is no universally defensible “five trials per student” rule. Plan enough opportunities for your clinical question, note when the sample is small, and avoid claiming mastery from one short probe. If a target was not sampled, plan when to revisit it rather than filling in a number.

5. Keep absent, declined and not addressed distinct

Absent: the student did not attend. Record attendance and relevant service context in your required system; there are no performance trials.

Declined/refused: describe the offered opportunity and observable response neutrally, including relevant access or participation barriers. Do not infer motivation or score it as a skill error unless your established measure actually warrants that scoring. Follow local attendance/service rules separately.

Not addressed: the student attended but that goal was not probed. Note what was addressed and, when useful, the next probe plan. None of these statuses means 0% accuracy. Zero percent means scored opportunities occurred and no responses met the definition.

A fictional mixed-goal group session

Invented example: three student codes share a picture-based turn-taking activity. These are not real students, norms or treatment prescriptions.

• Student A, articulation: 6/8 independent word-level target productions met the scoring definition (75%). In a separate teaching block, 3/4 productions after a model met the target. Keep the model-supported block separate.
• Student B, WH questions: 3/5 independent responses to who questions were correct (60%). A later block using picture choices yielded 2/3 correct. Do not report that as independent 5/8.
• Student C, AAC: initiated a request using the accessible AAC system in 2/4 defined opportunities without an added cue (50%). Record aided modeling and any access adjustments as context, not as hidden independent trials.

If Student D was absent, record absence—no denominator. If Student B’s second goal was not addressed, record that separately. The group had a shared activity, not a shared outcome.

6. Convert trials to percentages without hiding counts

Percent = correct responses ÷ actual scored opportunities × 100. Student A’s 6 ÷ 8 × 100 = 75%. Write “6/8 (75%), independent, word level,” not just “75%.” Use the accuracy tally to check the arithmetic and keep prompting visible.

With unequal denominators, an unweighted average can tell a different story from a pooled count. For example, 2/3 and 8/10 total 10/13 (about 76.9%); averaging 66.7% and 80% gives about 73.3%. Say which calculation you used. Pool only comparable tasks and support conditions, never different students’ results as an individual measure.

A small number of opportunities limits the conclusion. Do not create a denominator for an unmeasured goal, inflate a short sample to an imagined ten trials, or call a goal met unless its own consistency and support criteria are documented.

7. Write the session note from the evidence

Follow your required note format. Useful content commonly includes the date and service context, actual targets, task level, individual responses and counts, support provided, relevant observations and the next clinical step. Add required duration, attendance, signatures or payer fields in the approved system—not because this guide declares them universally sufficient.

Fictional note excerpt: In a picture-based group activity, Student A produced the target at word level in 6/8 independent opportunities (75%). A separate model-supported practice block yielded 3/4 target productions. Next probe will retain the same word-level task and independent condition; teaching will continue with documented models as needed.

Describe what you observed. A note should not invent generalization, educational impact, diagnosis or reasons for behavior. A progress draft can summarize recorded evidence, but you still check its counts and interpretation. See progress drafts from session data.

Common mistakes to avoid

• Changing the response definition halfway through the session without documenting it.
• Combining modeled and independent responses into one unexplained percentage.
• Using a group total for every child or forgetting the actual denominator.
• Logging an absent or unaddressed goal as a failed performance sample.
• Treating an engaging activity as evidence that every goal improved.
• Reconstructing detailed trials from memory or allowing a generated note to add observations.
• Using a generic template as proof of legal or billing compliance.

A reusable group-data template

No form, account or email required. This blank template is a starting point—not a district-approved record format.

GROUP THERAPY DATA — use fictional codes for practice
Date / duration / setting:
Activity and materials:
Student code:
Goal / target:
Measure and scoring definition:
Probe or teaching?:
Support permitted by goal:
Independent correct / independent opportunities:
Prompted correct / prompted opportunities:
Prompt type / amount:
Incorrect responses and observations:
Status if unmeasured: absent / declined / not addressed / other (describe)
Interpretation limited to this sample:
Next probe or teaching step:

Repeat the student block for each student and goal.
Percent = correct / actual opportunities × 100; keep raw counts.
No opportunities means unmeasured, not 0%.
Follow district privacy, documentation and retention rules.

Requirements, sources and privacy

34 CFR §300.320(a)(3) requires an IEP description of how progress toward annual goals will be measured and when periodic reports will be provided. It does not prescribe this template, a fixed trial count or a particular software product. This guide is general clinical workflow information, not legal advice.

ASHA’s Documentation in Schools discusses school documentation and varying requirements. Consult applicable district, state and payer rules. ASHA does not endorse SLP Draft through this citation.

Keep identifying information out of public examples, clipboard tools and unapproved systems. Initials may still identify a student when combined with other details. Follow district agreements, access controls and retention rules before entering or saving records.

For the product workflow, see SLP Draft, live individual/group data tracking and the 14-day no-card trial. This guide is useful with paper or another approved tool as well.