Replication Research, Publication Bias, and Applied Behavior Analysis
Fix your graph axes before submission so your data picture tells the truth.
01Research in Context
What this study did
The team looked at every single-case graph in three top autism journals from 2013-2017.
They checked if the y-axis to x-axis length and the number of data points followed the rules that stop false positives.
Only graphs that showed one behavior of one child were counted.
What they found
Most graphs broke the y:x ratio rule. The vertical axis was often too tall compared with the horizontal axis.
The good news: the vertical axis was rarely stretched so much that it would trick readers into seeing a big effect that was not there.
In short, our graphs look tidy, but they do not match the math we teach in grad school.
How this fits with other research
Szempruch et al. (1993) already showed that ABA papers were moving toward natural settings. Tincani’s team now says we still carry old graph habits with us.
Gillberg et al. (1983) warned that weak methods hide true treatment effects. Today’s poor axis scaling is the same problem wearing new clothes.
Scahill et al. (2015) picked five tools as trial-ready. If we submit graphs that break scaling rules, even the best RRB measure will look shaky.
Why it matters
Reviewers and editors notice sloppy axes. A stretched y-axis can make a tiny change look huge and waste your client’s time. Before you hit submit, measure your y:x ratio and count your data points per x-to-y length. Fix them in five minutes and your data story stays honest.
Want CEUs on This Topic?
The ABA Clubhouse has 60+ free CEUs — live every Wednesday. Ethics, supervision & clinical topics.
Join Free →Open your last graph, measure the y-axis and x-axis in centimeters, and check the 2:3 ratio.
02At a glance
03Original abstract
In single-case research designs (SCDs) to determine a functional relation a time-series graph is constructed. Preliminary evidence suggest the approach used to scale the vertical axis and the data points per x- to y-axis ratio (DPPXYR) impact visual analysts' decisions. We conducted a systematic review to evaluate time-series graphs published in the last decade in four premier journals in the field of autism. We included 348 articles including 2675 graphs. We identified large variation across and within types of SCDs when evaluating the lengths of the vertical and horizontal axis using the y:x ratio and the DPPXYR, with few adhering to current recommendations. A majority of graphs used an appropriate method to scale the vertical axis that would not increase Type I error rates.
Perspectives on Behavior Science, 2019 · doi:10.1007/s40614-019-00191-5