Glossary of Biases

Below is a brief glossary of cognitive biases you are likely to encounter in your own research and your sources.

Acquiescence bias: Applies particularly to surveys.  Refers to respondents’ tendency to agree with whatever is said or to respond positively when in doubt; sometimes called yea-saying or the “yes” bias. This type of bias can affect a variety of question types, but agree/disagree and Likert-type questions are especially vulnerable.

Confirmation bias: Applies to research more generally.  Refers to a tendency to seek, interpret, and remember information in a way that supports a previously held belief or hypothesis.  This type of bias is especially strong with emotionally charged or polarizing issues.

Observer effect: This term, which originated in physics, has come to serve as a shorthand expression for a wide variety of similar effects in different fields.  Refers to the notion that the act of observation will alter the outcome of the experiment in some way.  Subjects may alter their behavior if they know they are being observed; the instruments used to measure a phenomenon will inevitably change it in some way.

Recall bias: Affects studies that rely on memory and self-reporting.  Refers to differences in the accuracy with which participants recall past experiences.  This type of bias can be amplified by flaws in study design (questions which lead respondents to classify past experiences differently) or by differences in respondents’ motivation to remember past experiences.  Interacts with other cognitive biases.

Selection bias: Affects a wide variety of study types.  This type of bias occurs when the subjects or data selected for analysis are not sufficiently random to be representative. Often unavoidable; demonstrating awareness of this type of bias is sometimes the best strategy.

Social desirability bias: Applies particularly to surveys, but can affect any study that relies on self-reporting of behavior not confirmed by objective measures.  Refers to respondents’ tendency to under-report socially undesirable behaviors (i.e., smoking) and over-report socially desirable behaviors (i.e., recycling).