Observer Bias

Errors and biases can occur in all types of social science research, whether it is qualitatively driven or quantitatively led. One type of bias that may occur [Page 577]during observational methods is observer bias. This term refers to the ways in which errors may unconsciously occur when gathering and analyzing observational data. The observer's age, gender, social class, values, schemas, and expectations may lead certain observations to be recorded as significant and others to be ignored if the observer regards them as unimportant. Further, observers may associate meanings to the observed behavior that are different from the meanings associated by the person(s) displaying the behavior.

Psychological studies have shown that bias in human perception and expectation can encourage findings that correspond with those perceptions and expectations. In one study, Miriam Goldstein, J. Roy Hopkins, and Michael J. Strube found that students who were led to believe that an individual's performance would decrease as a result of consuming alcohol reported corresponding observations. The decrease in performance was not evident in the observed behavior as the individual observed—a trained confederate who drank a nonalcoholic drink—maintained consistent performance throughout the observation.

Although the researcher cannot alter some influencing characteristics, such as age, observer bias can be reduced if the researcher adopts specific strategies during the observation, for example, using standardized forms. Standardized forms, as used in structured observations, can help to reduce the impact of selective perception, as the forms direct the observer's attention to those behaviors or activities that are closely related to the central research project. The use of multiple observers can also highlight deviations from the main research topic, misinterpretations, omissions, and overemphasis. Further, triangulation—the use of multiple methods or data sources to study the observational individual(s)—enables findings to be cross-checked and any biases or errors to be highlighted.

Observer bias can also manifest itself in more subtle ways, which are more difficult to eliminate. Those being observed may modify their behavior, disguise their actions, and alter their opinions because they do not wish to share them with the researcher. Behavior may be changed according to how those being observed think they are expected to behave. Such responses are often referred to as the Hawthorne effect, taken from the name of a factory where the effect was first identified—the Western Electric Company's Hawthorne Works in Chicago. As completely neutral observations may be difficult to achieve, when reporting observational findings researchers should acknowledge and discuss the factors that may have contributed to observer bias and how these factors may have altered the interpretation of the findings. This discussion will enable readers to make an informed judgment when assessing the value of the observational findings.