Bias as an epistemic notion

TY - JOUR

T1 - Bias as an epistemic notion

AU - Bueter, Anke

N1 - Copyright © 2021 Elsevier Ltd. All rights reserved.

PY - 2022/2

Y1 - 2022/2

N2 - Once one abandons the ideal of value-free and impartial science, the question of how to distinguish biased from legitimately value-laden science arises. To approach this “new demarcation problem”, I argue that one should distinguish different uses of “bias” in a first step: a narrow sense of bias in statistics (often described as “systematic deviation from the truth”), and a wider sense of the term covering any kind of tendency that may impact scientific reasoning. Secondly, the narrow sense exemplifies an ontological notion of bias, which understands bias in terms of deviation from an impartial ideal outcome. I propose to replace it with an epistemic notion of bias, which understands biased research as research that we have good reasons to suspect could have been (done) systematically better. From a socio-epistemic perspective, such good reasons to expect better can be found in a lack of responsiveness to conventional standards and/or critical discourse in the scientific community. In short, bias in an epistemic sense consists in a deviation, not from truth but from current best practice. While this turns bias into something that is dependent on time and context, it can illuminate the new demarcation problem by allowing to distinguishing between biased and legitimately value-laden research.

AB - Once one abandons the ideal of value-free and impartial science, the question of how to distinguish biased from legitimately value-laden science arises. To approach this “new demarcation problem”, I argue that one should distinguish different uses of “bias” in a first step: a narrow sense of bias in statistics (often described as “systematic deviation from the truth”), and a wider sense of the term covering any kind of tendency that may impact scientific reasoning. Secondly, the narrow sense exemplifies an ontological notion of bias, which understands bias in terms of deviation from an impartial ideal outcome. I propose to replace it with an epistemic notion of bias, which understands biased research as research that we have good reasons to suspect could have been (done) systematically better. From a socio-epistemic perspective, such good reasons to expect better can be found in a lack of responsiveness to conventional standards and/or critical discourse in the scientific community. In short, bias in an epistemic sense consists in a deviation, not from truth but from current best practice. While this turns bias into something that is dependent on time and context, it can illuminate the new demarcation problem by allowing to distinguishing between biased and legitimately value-laden research.

U2 - 10.1016/j.shpsa.2021.12.002

DO - 10.1016/j.shpsa.2021.12.002

M3 - Journal article

C2 - 34955230

SN - 1871-7381

VL - 91

SP - 307

EP - 315

JO - Studies in History and Philosophy of Science

JF - Studies in History and Philosophy of Science

ER -