Situational Bias

Abstract

Situational bias is a systematic error that is caused by the research situation and participants’ reactions to this situation. Situational factors that equally affect E- and C-participants do not cause spurious differences between conditions, but factors that differentially affect E- and C-participants cause bias. Standardization of the research situation equalizes the situation for all participants, and calibration equalizes the research situation across time. Participants may differentially react to conditions, and experimenters and data analysts may differentially affect conditions. Blinding of these persons prevents the differential influence. Random assignment of research persons (e.g., experimenters, interviewers, etc.) to conditions turns their systematic influence on participants into random error. Necessary conditions for a causal effect of an IV on a DV are that the conditions are correctly implemented, and conditions are not contaminated. A manipulation check is a procedure to check the implementation of conditions. Contamination of conditions is prevented by separating conditions in location or time. Random assignment of participants counteracts selection bias , but it may induce randomization bias , for example, if participants dislike their assigned condition. Usually it cannot be prevented in randomized experiments , but it can be assessed by applying a double randomized preference design. Pretest -posttest studies are threatened by pretest effects, which are the effects of a pretest on participants’ behavior. It can be prevented by, for example, replacing the pretest by an unobtrusive proxy pretest , and it can be assessed by using Solomon’s four-group design. Additionally to pretest effects, studies that use a self-report pretest may show a response shift , which is a change in meaning of a participant’s self-evaluating from pretest to posttest. It can be assessed by administering a retrospective pretest .