New Research Shows Concern with Multiple Types of Bias
According to new unpublished research from the University of Texas at Austin, methodological problems have affected studies on the value of early autism intervention for the last three decades. The study’s authors reported these findings at last month’s International Society for Autism Research (INSAR) annual meeting. The study investigated this topic by evaluating the reliability of many studies which examined the effectiveness of early intervention. The team found that the evidence supporting the conventional autism therapy principle of “the earlier and more therapy a child receives, the better the outcome” is weak, citing that many early intervention studies included multiple types of bias and an overreliance on caregivers to report on outcomes. Three types of bias were identified by the team: selection bias, when experimental groups and control groups are not randomly assigned; detection bias, when the same person administers an intervention and judges its efficacy; and attrition bias, when participants from an experimental or control group drop out over time. The encouraging news is that these researchers found that the number of studies at high risk for selection and attrition bias has decreased from 1989 to 2017. However, both types of bias were still common among many of the studies they reviewed. Detection bias is also a problem for up to 50 to 75% of studies on the topic. The team also reported that many studies relied heavily on parental and caregiver reports on the effectiveness of a particular intervention instead of relying on clinical observations. Between 25 to 50% of studies from 2011 to 2017 used parental reports, down from approximately 90% in 2000. The study’s authors said the main goal of their report was to inspire scientists to improve the quality of their research.
