Bias. You’ve heard the term in the context of clinical research, but bias can also seep seamlessly into our conversations with friends, into the way news stories are presented, and into our nation’s politics. What exactly is bias? How does it influence us? How does it emerge in our everyday lives?
Bias appears as an unfair partiality, preference, or prejudice for or against a thing, person, group, or idea.⁶ ⁸
Unconscious bias, also called “implicit bias”
This refers to stereotypes that a person holds about certain groups of people. These beliefs are formed outside of the individual’s conscious awareness⁶ but have real-world consequences on his/her behavior.⁷
Everyone holds unconscious beliefs about social and identity groups, as biases arise from our use of categorization to organize our social worlds.⁶ Certain scenarios can prompt unconscious attitudes and beliefs.
For example, biases may become more apparent when a person is multitasking or working under pressure,⁶ and research demonstrates that unconscious bias has a substantial impact on the criminal justice system, education, and healthcare.⁷
Conscious bias, or “explicit bias”
This refers to bias demonstrated by a person who is clear about his/her feelings and attitudes, and his/her related behaviors are performed intentionally.⁹ This form of bias is processed at a conscious level as words through declarative, semantic memory.
Conscious bias can take the form of overt negative behavior such as physical and verbal harassment or subtler forms such as exclusion.⁹ It is important to note that unconscious bias is far more prevalent than conscious bias and often incompatible with one’s conscious values.⁶
With bias so abundant, what exactly can we do to spot it?
How to watch for bias
Consider the following when you consume information:
- What facts has the author or speaker withheld or overlooked? When biased ideas are presented, they are often worded in a way that oversimplifies or overgeneralizes.⁸
- What additional information is needed? When expressing a bias, the speaker or writer offers a limited view of the topic.⁸
- What were the words used? Oftentimes, the language used will appeal more to your emotions than your logic.⁸
- What words create positive or negative impressions? The language used is extreme; statements have all-or-nothing connotations.⁸
- What impression would I have if different words had been used?⁸ Consider how the message would have been altered if neutral language had been used. What impression would that have made instead?
- Who benefits? What does the author or speaker gain from having you adopt his/her ideas?⁸
- Money talks. Who is funding the speaker, writer, platform, and outlet of information?⁸
- Does the author or speaker present opposing points of view? If so, are those views presented objectively or scornfully?⁸
With all forms of bias, it is important to consider two questions: What is the magnitude of the bias? What is the direction of the bias?¹ These two questions are not only important when encountering bias in everyday situations but also when assessing bias in clinical research. The best research accounts for sources of bias systematically from experimental design to analysis.
When reviewing a study and the reliability of its results, it is important to consider the risk of bias and the extent to which sources of bias have been avoided.
Bias in clinical studies
What are some ways in which bias appears in clinical studies?
- Selection bias
- Performance bias
- Recall bias (also called reporting bias)
- Detection bias
- Reporting bias
Selection bias refers to systematic differences between the baseline characteristics of groups that are compared.¹ When selection bias occurs, the study population is not representative of the target population; thus, the study’s external validity is questionable, and its conclusions should not be generalized to the greater population.²
How to avoid selection bias: Randomization. When applied, randomization prevents selection bias in the allocation of the intervention to study participants. Allocating interventions to participants should be based on a random process, such as using a simple and random (and thus unpredictable) sequence.¹
Performance bias refers to systematic differences between the study groups with respect to the care that is provided or exposure to factors other than the study’s interventions.¹
How to avoid this performance bias: Blinding. After enrolling participants into the study, participants and personnel are blinded to reduce the risk that knowledge of which intervention was received would affect the outcomes. Effective blinding can also ensure that the compared groups receive similar attention, treatment, and investigation.¹
Recall bias describes the phenomenon where study participants are systematically more or less likely to recall information on exposure depending on their outcome status. Recall bias is likely to occur, for example, when exposure and disease status are both known during the study and can also occur when patient interviews are considered primary data sources.⁵
How to avoid recall bias: When patient-reported data are used, the study can be designed to mask the intent of questions in structured interviews or surveys, and/or validated scales can be used for data acquisition.⁵
Detection bias refers to systematic differences between groups in how outcomes are determined.¹
How to avoid detection bias: Blinding. Blinding of outcome assessors could reduce the risk that knowledge of which intervention was received would affect outcome measurement, especially subjective outcomes.¹
Reporting bias refers to systematic differences between reported and unreported findings. Within a report, analyses with statistically significant differences between intervention groups are more likely to be reported than non-significant differences. This type of “within-study publication bias” (also called outcome reporting bias or selective reporting bias) is a substantial source of bias affecting study results.¹
How to avoid reporting bias: Transparency. Transparency safeguards the integrity of research. Reporting guidelines such as CONSORT can help researchers improve their reporting and establish transparency in their process.³ ⁴ Researchers can also mitigate reporting bias by publishing a protocol in which they pre-specify the outcomes assessed in their research. If any of these outcomes is left unreported in their findings, reporting bias is likely present.¹⁰
Biased information -- in research and in everyday life -- attempts to change your mind and how you think. It is therefore an important skill to be aware of bias and know how to identify, analyze, and assess biased information properly. In research, from inception to publication, it is critical that studies are conducted in a manner that is reliable and accurate, unaltered by the influence of biases.
The consequences of bias in everyday life are far more wide-ranging. It can be light in one expression but lofty in the next. It can be loudly present in the banter between two sports fans, or it can quietly yet powerfully influence life-altering decisions in our health or criminal justice systems. Knowing how biases affect your world puts you in charge of how you think and consume the information that surrounds you.
1. Higgins, Julian and Sally Green. “8.4 Introduction to Sources of Bias in Clinical Trials.” The Cochrane Collaboration, March 2011. Accessed October 2020. https://handbook-5-1.cochrane.org/chapter_8/8_4_introduction_to_sources_of_bias_in_clinical_trials.htm
2. Lambert, Jerome. “How to Assess Bias in Clinical Studies?” Clinical Orthopaedics and Related Research 469, no. 6 (2011): 1794-1796. doi: 10.1007/s11999-010-1538-7.
3. Richards, GC and IJ Onakpoya. “Reporting Biases.” Catalogue of Bias, 2019. Accessed October 2020. https://catalogofbias.org/biases/reporting-biases/
4. “Consort: Transparent Reporting of Trials.” The CONSORT Group. Accessed October 2020. http://www.consort-statement.org/
5. Pannucci, Christopher J and Edwin G Wilkins. “Identifying and Avoiding Bias in Research.” Plastic and Reconstructive Surgery 126, no. 2 (2010): 619-625. doi: 10.1097/PRS.0b013e3181de24bc.
6. “Unconscious Bias.” UCSF, Office of Diversity and Outreach. Accessed October 2020. https://diversity.ucsf.edu/resources/unconscious-bias
7. “State of Science on Unconscious Bias.” UCSF, Office of Diversity and Outreach. Accessed October 2020. https://diversity.ucsf.edu/resources/state-science-unconscious-bias
8. “How to Evaluate Information Sources: Identify Bias.” New Jersey Institute of Technology, October 2020. Accessed October 2020. https://researchguides.njit.edu/evaluate/bias
9. “Two Types of Bias.” Georgetown University, National Center for Cultural Competence. Accessed October 2020. https://nccc.georgetown.edu/bias/module-3/1.php
10. Higgins, Julian and Sally Green. “8.14.2 Assessing Risk of Bias from Selective Reporting of Outcomes.” The Cochrane Collaboration, March 2011. Accessed October 2020. https://handbook-5-1.cochrane.org/chapter_8/8_14_2_assessing_risk_of_bias_from_selective_reporting_of.htm