Thursday, May 19, 2011

Ambiguous Evidence

Gilbert Chin Two of the methodologies used to compare the efficacies of treatments are randomized placement of participants in experimental studies and statistical or propensity matching of participants on confounding variables in observational studies. Both approaches have their proponents and might plausibly carry comparable weight in evidentiary assessments. Bastardi et al. developed a scenario in which a matching study (in comparison to a randomized one) was described as favoring home care of children to one group of participants, whereas a second group of participants received the information that the matching study favored day care. All of the participants were soon to become parents, and their preexisting belief was that home care was superior, although half of them were intending to use day care. When confronted with these scenarios in a crossed design, those who intended to use home care for their children did not judge one methodology to be more reliable than the other and maintained their stance about the superiority of home care. On the other hand, the internally conflicted participants (that is, those who intended to use day care) did regard the matching protocol as better when it favored day care, and this sufficed to shift their belief to the point where home care and day care were viewed as equivalent.
Psychol. Sci. 22, 10.1077/0956797611406447 (2011).