Bias

Bias refers to a predisposition or partiality. In qualitative research, bias involves influences that compromise accurate sampling, data collection, data interpretation, and the reporting of findings. Researchers may show bias when they reach conclusions that ignore contradictory data or when the collection and analysis of data are designed to lead to predetermined conclusions. Publication bias occurs when researchers and journals avoid reporting insignificant findings.

The traditional scientific method says that researchers should revise a theory when data fail to fit the theory, or they should abandon the theory and look for new explanations. Nevertheless, Thomas Kuhn has shown that most scientists attempt to make the data fit the theory. Kuhn's work helps to explain that scientists are products of their environments and therefore bring their assumptions and personal standpoints to the research enterprise.

The potential for bias enters the research enterprise the moment a researcher chooses one topic over another, one research question to the exclusion of another, and one particular theory over another. Researchers, like everyone else, are products of the social world and therefore have values that will be more or less apparent in their research.

Decisions around research method, population sampling, and other design issues can introduce bias. In circumstances where researchers repeatedly follow the methodology of previous studies, they run the risk of reproducing similar findings that are method-dependent. In quantitative research, especially, multiple methods are often used to maximize confidence that findings are reliable and valid. Similarly, it is important to recognize whether a particular sample represents the parent population. In qualitative research, however, biases are often assessed in the context of doing the research, to acknowledge and manage the limitations of the research design. Finally, the wording of interview questions merits careful consideration with regard to the wording of questions so that they are not preordained to elicit biased responses. Wherever possible, pretesting should be employed.

Many researchers anguish over the dilemma of doing research that is either impartial and neutral or firmly grounded in a value position. Howard Becker has argued that this dilemma does not exist because researchers are not value-free, and therefore, personal and political views will enter a research agenda. The real imperative is for researchers to be aware of their values and predispositions and to acknowledge them as inseparable from the research process.

Perception of bias can be most apparent when research challenges a status quo. For example, research that opposes the vested interests of public officials is more likely to be criticized for bias than research that does the opposite. Research that challenges longstanding positions such as drug prohibition, gender discrimination, or ageist policy will often be accused of bias.

Researchers manage bias by being self-aware of their values and assumptions, looking for contradictory data, and being open to alternative interpretations of their data. Although many of the social sciences aspire to objectivity, social scientists should acknowledge their own subjectivity in the research process.