The impact of response-guided baseline phase extensions on treatment effect estimates.
Letting visual analysis lengthen the baseline will not bias your multilevel effect estimate and can make it sharper.
01Research in Context
What this study did
The team ran 5,000 fake multiple-baseline studies on a computer. Each fake study had three subjects. They let the computer "watch" the baseline and add extra sessions only if the data looked flat or noisy.
They wanted to know if this common real-life habit warps the final treatment-effect number. They checked bias (did the effect shrink or grow?) and precision (did the estimate jump around?).
What they found
Extending the baseline by eye did not push the effect size up or down. The average estimate stayed right on the true value.
The extra sessions actually tightened the error bars. Standard errors dropped by about 10 percent. In plain words, you get a clearer picture, not a twisted one.
How this fits with other research
Lloveras et al. (2025) extends this idea. They tell us to stagger baselines across two dimensions (like kid AND room) so we never start treatment on a drifting line. Their plan prevents the very dilemma Seang-Hwane green-lights.
Kestner et al. (2018) sounds a warning about repeated exposures: resurgence shrinks the second time around. That caution reminds us that any within-subject tweak (baseline extension included) can carry order effects, even if Seang-Hwane’s simulation did not detect bias.
McGonigle et al. (1982) showed experts mostly ignore variability and look for pattern. Seang-Hwane gives those visual extenders numbers to back up their gut: keep looking, keep extending, your effect stays clean.
Why it matters
You can stop worrying that adding extra baseline points will spoil your multilevel model. If the data look ugly, add sessions until they calm down. You will not cheat the effect size; you will just see it better. Monday morning, open your graph, trust your eye, and collect those few extra points with confidence.
Get CEUs on This Topic — Free
The ABA Clubhouse has 60+ on-demand CEUs including ethics, supervision, and clinical topics like this one. Plus a new live CEU every Wednesday.
If the baseline is rocky, add 3-5 more sessions before starting treatment—your stats stay honest.
02At a glance
03Original abstract
BACKGROUND: When developmental disabilities researchers use multiple-baseline designs they are encouraged to delay the start of an intervention until the baseline stabilizes or until preceding cases have responded to intervention. Using ongoing visual analyses to guide the timing of the start of the intervention can help to resolve potential ambiguities in the graphical display; however, these forms of response-guided experimentation have been criticized as a potential source of bias in treatment effect estimation and inference. AIMS AND METHODS: Monte Carlo simulations were used to examine the bias and precision of average treatment effect estimates obtained from multilevel models of four-case multiple-baseline studies with series lengths that varied from 19 to 49 observations per case. We varied the size of the average treatment effect, the factors used to guide intervention decisions (baseline stability, response to intervention, both, or neither), and whether the ongoing analysis was masked or not. RESULTS: None of the methods of responding to the data led to appreciable bias in the treatment effect estimates. Furthermore, as timing-of-intervention decisions became responsive to more factors, baselines became longer and treatment effect estimates became more precise. CONCLUSIONS: Although the study was conducted under limited conditions, the response-guided practices did not lead to substantial bias. By extending baseline phases they reduced estimation error and thus improved the treatment effect estimates obtained from multilevel models.
Research in developmental disabilities, 2018 · doi:10.1016/j.ridd.2017.12.018