Matching procedures in autism research: evidence from meta-analytic studies.
Match and report age, IQ, sex, and birth order or risk fake treatment wins.
01Research in Context
What this study did
The authors looked at three earlier meta-analyses on autism. They checked how well each study matched kids on age, IQ, sex, and birth order.
The goal was to see if poor matching could fake treatment effects.
What they found
Many papers did not report matching steps. When they did, the groups still differed on mental age or birth order.
These gaps make it hard to know if gains come from treatment or from pre-existing differences.
How this fits with other research
Lord et al. (2005) widened the same warning. They said autism psychosocial studies need RCTs and shared outcome tools.
Jonsson et al. (2016) later showed that even RCTs skip details like setting, dosage, and trainer skill. Together the three papers trace one thread: weak reporting hides real-world meaning.
Sevlever et al. (2010) applied the idea to imitation work. They found clashing results simply because teams defined and measured imitation differently.
Why it matters
Before you compare two groups, list age, IQ, sex, and birth order in your intake sheet. Run stats to prove the groups do not differ. Add this table to every report, poster, or grant. This five-minute step shields your conclusions from easy attack and lifts the whole field.
Want CEUs on This Topic?
The ABA Clubhouse has 60+ free CEUs — live every Wednesday. Ethics, supervision & clinical topics.
Join Free →Add a quick matching table to your next intake packet and check for group differences before treatment starts.
02At a glance
03Original abstract
In this paper, we summarize some of our findings from a series of three meta-analyses and discuss their implications for autism research. In the first meta-analysis, we examined studies addressing the theory of mind hypothesis in autism. This analysis revealed that theory of mind disabilities are not unique to autism, although what may be unique is the severity of the dysfunction in this group. Variables such as the chronological and mental age of the participants, and the matching procedures that the researchers employed, were found to be significant moderator variables. In the next two meta-analyses, data regarding siblings and parents of individuals with autism were analyzed. Type of comparison group (e.g., siblings or parents of individuals with Down syndrome or learning disabilities) and type of outcome measure (cognitive, psychiatric, language) were found to be important moderator variables. Furthermore, method of assessing the psychiatric difficulties (e.g., self-report, clinical measures) was found to be a moderator variable in parents' meta-analysis. Suggestions for future research are discussed, highlighting variables such as type of comparison group, matching procedures, chronological and mental ages, gender, and birth order.
Journal of autism and developmental disorders, 2004 · doi:10.1023/b:jadd.0000018072.42845.83