CDIS 564 • Methods Reference

Fidelity

Showing the intervention was delivered the way you said it would be.

Procedural fidelity, also called treatment fidelity, is the degree to which the intervention was delivered as designed. Without it, a change in behavior cannot be tied to the intervention, because you cannot show the intervention was the thing that happened. Expand each panel for how to build the checklist, who observes, how it is calculated, and the standard to meet.
Use the Single-Case Design Rigor Planner to plan your fidelity, reliability, and validity decisions for the proposal.
1 What procedural fidelity is ▼

Procedural fidelity measures whether the person delivering the intervention did what the protocol says, step by step. It protects internal validity directly: if the intervention was not delivered as planned, a change cannot be credited to it, and the absence of a change cannot be blamed on it either.

Fidelity is separate from reliability. Reliability asks whether your measurement of behavior was consistent. Fidelity asks whether your delivery of the intervention was accurate. A study needs both.

Fidelity is measured against your standard operating procedures. You write one SOP for the baseline condition and one for the intervention, each as specific, observable steps, before the study starts. The baseline fidelity checklist is built from the baseline SOP, and the intervention fidelity checklist is built from the intervention SOP.
2 Building the checklist from your SOPs ▼

A fidelity checklist turns each core step of your intervention into an item an observer can mark as done or not done. Build it directly from your standard operating procedures.

A well-constructed checklist usually includes:

  • Core intervention components: the essential steps that define the intervention.
  • Adherence measures: a yes or no, or a rating, for whether each component was delivered.
  • Quality ratings: how well each component was delivered, when quality matters, not just whether it happened.
  • Dosage information: the amount or frequency of delivery, when dose is part of the intervention.
  • Participant responsiveness: the participant's engagement or reaction.
  • Comments: space for notes on anything the checklist items do not capture.

To build it: identify the core components from your protocol, write each as an observable action, set clear criteria for adherence and quality, add sections for dosage and responsiveness, then pilot the checklist and refine it before the study.

How a checklist is built from SOPs Boster and colleagues (2024) turned each core step of their intervention into a single observable item. An observer watched each video and marked every step complete or incomplete. The full intervention checklist is below. Each item is a concrete action taken straight from the procedure, so two observers scoring the same video would mark it the same way.
Example Checklist
Intervention (B) phase procedural fidelity
Built from the intervention procedure. Adapted from Boster et al. (2024). An observer marks each step complete or incomplete for each monitored session.
  • Seats both children together at a table.
  • Provides the child with CCN access to the AAC device.
  • Introduces the activity and tells the children they will be working together.
  • Provides the children with a shared kit and camera.
  • Has the iPad set up as a visual for the children on the table.
  • Models the word of the day for the child with CCN.
  • Provides the "surprise ingredient" for the craft activity.
  • Instructs the children to work together to build their monster.
  • Provides praise for completing the activity at the end of the session, or when the children have completed the craft.
  • Maintains both children in camera view for the duration of the session.
3 Who observes, and how much ▼

An observer other than the person delivering the intervention scores fidelity from a sample of sessions.

  • An independent observer, ideally someone not otherwise involved in delivering or assessing the study, scores fidelity.
  • At least 20% of sessions are monitored for fidelity, selected at random.
  • The observer is trained on the checklist to a set agreement standard before scoring study sessions, the same way a reliability observer is trained.
Example In the worked narrative study, at least 20% of each participant's randomly selected sessions were coded for procedural fidelity by a coder not involved in the study procedures, with sessions chosen using a random number generator. Namasivayam and colleagues (2024) had treatment fidelity assessed from video by a speech-language pathologist experienced with the approach who was external to the site and not involved in the study's assessments or interventions.
4 How fidelity is calculated ▼

The basic calculation is the percentage of required steps that were completed.

fidelity = steps completed / total steps required x 100

Depending on your intervention, you may report fidelity in more than one way:

TypeFormula
Adherence percentage(components delivered / total components) x 100
Quality-weighted adherence(sum of adherence x quality rating / total possible score) x 100
Dosage fidelity(actual dosage / intended dosage) x 100
Composite fidelity(adherence % + quality % + dosage %) / 3
Worked example In the worked narrative study, procedural fidelity was calculated by dividing the number of completed intervention procedures by the total number expected and multiplying by 100. Across nine assessed videos, fidelity averaged 93%, with a range of 91% to 100%. The two recorded errors were named specifically: the researcher once did not review the visual tool, and once prompted the child during the measurement portion. Naming the errors, not just the average, is good practice.
Watch out Report fidelity with more than one rater on the monitored sessions and check agreement between them, just as you do for the dependent variable. A fidelity score is only as trustworthy as the agreement behind it.
5 The 80% standard and corrective action ▼

Interpret fidelity against these levels:

  • High fidelity: 80% or above.
  • Moderate fidelity: 51% to 79%.
  • Low fidelity: 50% or below.

The minimum you aim for is 80%. If fidelity falls below the standard, retrain the person delivering the intervention before continuing, and use the fidelity data to guide that retraining. Fidelity is not only a number to report at the end; it is a signal you act on during the study.

Example of a training standard In the worked narrative study, the fidelity coder practiced until reaching 90% agreement on every fidelity variable across three consecutive practice videos before scoring study sessions. When practice agreement fell below 90% on any variable, the researcher and coder held a discrepancy discussion, reviewed the video, and resolved the disagreement before continuing.
6 Fidelity in baseline and intervention ▼

Fidelity is not only for the intervention phase. The baseline condition also follows a procedure, and you have to show it was carried out as planned. If baseline sessions drifted, the comparison against intervention is weakened. Monitor fidelity in both phases.

Because the steps differ between phases, each phase has its own SOP, and you build a separate fidelity checklist from each one. The baseline fidelity checklist is built from the baseline SOP, which states how a baseline session is run with the intervention withheld. The intervention fidelity checklist is built from the intervention SOP, which states how the intervention is delivered.

Example of phase-specific checklists Boster and colleagues (2024) built one checklist for the A (baseline) phase and a separate checklist for the B (intervention) phase. The two share several setup steps but differ where the phases differ, for example whether children worked on individual kits (A) or a shared kit together (B). Matching the checklist to the phase is what lets you show fidelity everywhere it matters.
Example Checklist
Baseline (A) phase procedural fidelity
Built from the baseline procedure. Adapted from Boster et al. (2024). An observer marks each step complete or incomplete for each monitored session.
  • Seats both children together at a table.
  • Provides the child with CCN access to the AAC device.
  • Introduces the activity and tells the children they will be working on their own kit.
  • Provides the children with individual kits.
  • Has the iPad set up as a visual for the children on the table.
  • Models the word of the day for the child with CCN.
  • Provides the "surprise ingredient" for the craft activity.
  • Instructs the children to build their monster.
  • Provides praise for completing the activity at the end of the session, or when the children have completed the craft.
  • Maintains both children in camera view for the duration of the session.

The intervention (B) phase checklist for the same study is in the "Building the checklist from your SOPs" panel above. Comparing the two shows how the shared setup steps stay the same while the steps that define each phase differ.

7 What to write in your proposal ▼

Your fidelity section should let a reader see the intervention was delivered as designed. Include each of these:

  • A fidelity checklist for each phase, the baseline checklist derived from your baseline SOP and the intervention checklist from your intervention SOP, with the core steps as observable items.
  • Who observed fidelity and how they were trained, including the training criterion.
  • The percentage of sessions monitored (20% or more), in baseline and in intervention, selected at random.
  • How fidelity was calculated and the values obtained, including any errors.
  • The minimum you set (80% or higher) and what you would do if a session fell below it.

The SCRIBE reporting guideline, included in this course, lists the fidelity details reviewers expect a single case report to state.

□ Sources ▼