Planning, Implementation, and Evaluation of Intervention: Goals, Dose, and Data
Domain 3 covers turning assessment findings into measurable goals, delivering intervention with defensible intensity, and using data to continue, modify, or discharge. Items target goal quality, dose calculation, cueing, and generalization planning.
Writing goals that survive scrutiny
A behavioral objective has four parts: condition (context, materials, cue level), behavior (observable and countable), criterion (accuracy plus consecutive sessions or probes), and a measurement plan.
- "Will improve receptive language" fails: no observable behavior, no criterion.
- "Will identify 8 of 10 pictured verbs from a field of three given a verbal model, across 3 consecutive probes" passes.
- State baseline and aim numerically so progress is auditable. Goal attainment scaling is an alternative when discrete counts misrepresent change.
Intensity and dose
Cumulative intervention intensity = dose (teaching episodes or trials per session) × dose frequency × total duration. Minutes per week is a weak proxy: two 30-minute sessions with 15 trials each are not comparable to the same minutes containing 80 trials. Speech sound intervention leans on high production counts and distributed, variable practice; blocked practice may speed acquisition while random/variable practice better supports retention and transfer.
Cueing and approaches
- Retreat through a least-to-most hierarchy: independent → verbal model → phonemic cue → visual/gestural → tactile → imitative, then fade cues. Count only unprompted responses as correct.
- Language approaches differ in who initiates and how contingent the response is: focused stimulation, recasts, expansions, milieu/incidental teaching, peer-mediated intervention.
- Service delivery: pull-out versus push-in/classroom-based, individual versus group, and RTI tiers that precede special education referral.
Worked example
Baseline: /s/ at 20% in phrases, 0% in conversation, 80% in isolation. Goal: "Given a picture scene, Jordan will produce /s/ in conversational speech with 80% accuracy across three consecutive weekly probes, with no more than one verbal cue per 10 utterances." Reasoning: condition, behavior, criterion, measurement schedule, and a cue ceiling are all specified, so the data determine continuation, modification, or discharge.
Traps that cost marks
- Goals with unobservable verbs ("know," "understand").
- Counting prompted responses as correct.
- Changing criterion and cue level in the same session, rendering data uninterpretable.
- Confusing more minutes with greater intensity.
- Discharging with no maintenance or generalization probe outside the therapy room.