Assessment & Research

Further evaluating interobserver reliability and accuracy with and without structured visual‐inspection criteria

Cox et al. (2021) · Behavioral Interventions 2021
★ The Verdict

Give every new BCBA the Roane visual-inspection checklist—accuracy shoots up even if agreement only inches forward.

✓ Read this if BCBAs who train staff to interpret functional-analysis graphs.
✗ Skip if Practitioners who already use automated FA-interpretation software.

01Research in Context

01

What this study did

Cox et al. (2021) asked new BCBAs to look at functional-analysis graphs.

Half got the Roane visual-inspection checklist. Half got no form.

The team then scored how close each novice came to expert answers.

02

What they found

Accuracy jumped when the checklist was in hand.

Agreement between novices only nudged upward.

In short, the form fixed more wrong answers than it created matches.

03

How this fits with other research

Guerrero et al. (2022) repeated the idea with mealtime graphs and saw the same pattern: accuracy up, but only small reliability gains.

Morantz et al. (2022) went further and coded the checklist into an R script. The robot agreed with experts 81% of the time, showing the rules still work when a human is not holding the pencil.

Machado et al. (2021) trained observers to score behavior videos at 5× speed. Both studies show that brief, clear rules can slash error without long lectures.

04

Why it matters

You can hand the Roane sheet to a new hire today. One page lifts their FA-call accuracy right away. Keep using it yourself, or let the free R script do the math. Either way, you spend less time fixing graphs and more time treating clients.

Free CEUs

Want CEUs on This Topic?

The ABA Clubhouse has 60+ free CEUs — live every Wednesday. Ethics, supervision & clinical topics.

Join Free →
→ Action — try this Monday

Print the Roane checklist, slide it into each trainee’s binder, and have them use it on the next five FA graphs you review together.

02At a glance

Intervention
not applicable
Design
single case other
Sample size
15
Finding
mixed

03Original abstract

AbstractVisual inspection is the primary method of interpreting functional analysis (FA) outcomes, even though it has occasionally been criticized for producing low levels of interobserver agreement. Researchers have addressed this issue by creating structured visual‐inspection criteria to guide visual inspection of FA outcomes (e.g., Hagopian et al., 1997, https://doi.org/10.1901/jaba.1997.30‐313; Roane et al., 2013, https://doi.org/10.1002/jaba.13). The purpose of the current study was to systematically replicate and extend Study 1 of Roane et al. (2013, https://doi.org/10.1002/jaba.13). We did this by evaluating the reliability and accuracy of 15 novice participants’ visual inspection of 84 FA graphs with and without the modified visual‐inspection criteria developed by Roane et al. Accuracy was markedly higher when participants used the modified visual‐inspection criteria relative to when they used traditional visual‐inspection strategies, while we observed more modest increases in reliability coefficients. Results are discussed in the context of practical and clinical implications of the modified visual‐inspection criteria and suggestions for future research.

Behavioral Interventions, 2021 · doi:10.1002/bin.1793