Construct · 6 fields · 8 citations
One name. Different field. Different construct. One map.
● jingle · one name, six constructs Nearest construct by meaningresponsecompare the meaningsEvery construct on the map is compared with every other by the meaning of its definitions, and the closest one is surfaced here. The nearest construct to bias is response.
Hold this as a prompt, not a verdict. Two constructs sitting close in meaning is a reason to read them side by side and decide for yourself whether they are one idea under two names. It is not proof that they are. Compare the meanings in the Collider.
Spokes are fields; dashed arcs are shared lineage. Tap a field for its definition, unit, and measures.
Drag to watch the word get claimed. Each field appears at the publication date of the earliest source cited for it. Tap a year to see which.
The difference between the expected value of an estimator and the quantity it estimates. Bias here is a property of a procedure averaged over samples, and it is not a fault: Geman, Bienenstock, and Doursat showed in 1992 that expected squared error decomposes into bias plus variance, so an estimator with some bias and low variance routinely beats an unbiased one. In this meaning a model can be too unbiased. No person, group, or value judgement appears in the definition.[1]
A systematic difference in a system's outputs or errors across social groups that produces unfair disadvantage. Barocas and Selbst showed in 2016 how it arises without intent (from the target variable chosen, the training labels, the features, and the base population) and mapped it onto disparate-impact doctrine in American antidiscrimination law. Mehrabi and colleagues catalogued the sources and the competing fairness definitions in 2021 and noted that several cannot be satisfied at once. A model can be unbiased in the statistical meaning and biased in this one, and the reverse.[2][3]
A systematic, predictable departure of human judgement from a normative standard, probability theory, logic, expected utility. Tversky and Kahneman defined the program in 1974: people judge under uncertainty with a few heuristics such as representativeness, availability, and anchoring, and each heuristic produces a characteristic bias. The bias belongs to the judgement process, is shared by nearly everyone, and is diagnosed by comparing answers with what the norm says they should be. It carries no implication about attitudes toward anyone.[4]
An automatic association between a social category and an evaluation or stereotype, held without endorsement and measured without asking. Greenwald, McGhee, and Schwartz introduced the Implicit Association Test in 1998: the speed with which a person pairs, say, a group with pleasant versus unpleasant words is taken as the strength of the association. Unlike the cognitive meaning, there is no normative answer being missed; unlike the fairness meaning, the object is an individual mind, not a system's outputs.[5]
Error in a population estimate that arises because the units observed differ systematically from the units not observed. Groves showed in 2006 that nonresponse bias depends on the covariance between the propensity to respond and the survey variable, not on the response rate itself, so a low response rate is neither necessary nor sufficient for bias. Heckman's 1979 treatment recast selection as a specification error that can be modelled and corrected. The bias is in the estimate of a population quantity, and its cause is who got in.[6][7]
Covariance between measured variables that comes from the measurement procedure rather than from the constructs, most often because one respondent rated both the predictor and the outcome on the same questionnaire at the same time. Podsakoff and colleagues catalogued the sources in 2003 (single source, item wording, context, mood, social desirability) and the procedural and statistical remedies, and reviewers in management, marketing, and information systems now demand them. It inflates correlations rather than shifting a population mean, and nothing about the sample or the respondent's attitudes needs to be wrong.[8]
Next
See how bias sits against neighbouring constructs in the Collider, or export every reference on this page in one click.
Open in ColliderExport BibTeXReport this collisionCited on this page?
If your work is cited behind one of these entries, you know better than we do whether we captured the meaning right. Confirm it, suggest an edit, or dispute it, and your name goes on the entry as an author who has checked it, linked to your ORCID.
Claim your construct