ERIC Number: ED078086
Record Type: RIE
Publication Date: 1973-Mar
Reference Count: 0
The Effects of Observer Bias in Field-Experimental Settings. Final Report.
O'Leary, K. Daniel
A series of studies on observer biases revealed that simply informing observers of experimental hypotheses does not produce observational data consonant with those hypotheses. However, questionnaire responses following an experiment with different induced expectations does produce global data consonant with experimental hypotheses. In addition, if the observers are informed of the experimental hypotheses and the investigator provides daily feedback to the observers indicating how well their data support his hypotheses, the observers will report data consonant with those hypotheses. The method of investigation in the studies reported involved having observers watch specially prepared video tapes of children who exhibited significant amounts of disruptive behavior. Following a pre-treatment or baseline period, observers were then asked to watch video tapes on which children displayed no change or marked reductions in disruptive behavior during a "treatment period." While observer biases per se did not result in confounded data in any of the studies, an unanticipated problem of observer drift or changing observational criteria can result in seriously confounded data where groups of observers initially trained together are later assigned to different treatment conditions. (Author/CK)
Publication Type: N/A
Education Level: N/A
Sponsor: National Center for Educational Research and Development (DHEW/OE), Washington, DC. Regional Research Program.
Authoring Institution: State Univ. of New York, Stony Brook. Dept. of Psychology.