General Principles for Reducing Bias
The following are general principles for writing about all people and their personal characteristics without bias.
Guideline 1: Describe at the appropriate level of specificity
Precision is essential in scholarly writing; when you refer to a person or persons, choose words that are accurate, clear, and free from bias or prejudicial connotations. Bias, like inaccurate or unclear language, can be a form of imprecision. For example, using “man” to refer to all human beings is not as accurate or inclusive as using the terms “individuals,” “people,” or “persons.”
Focus on relevant characteristics
Be mindful to describe only relevant characteristics. Although it is possible to describe a person’s age, disability, gender identity, participation in research, racial and ethnic identity, sexual orientation, socioeconomic status, or other characteristic without bias, it is not always necessary to include all of this information in your report. For a more detailed discussion of participant characteristics to report in your papers, see the journal article reporting standards for quantitative studies and qualitative studies. For example, you would be unlikely to mention participants’ sexual orientation in a study of cognition because sexual orientation is not relevant to cognition; however, you would likely mention participants’ gender in a study of stereotype threat because gender is relevant to the examination of stereotype threat. Furthermore, there may be multiple relevant characteristics to discuss; when this is the case, address the ways in which the characteristics intersect, as appropriate.
Acknowledge relevant differences that do exist
Part of writing without bias is not only recognizing that differences should be mentioned only when relevant but also acknowledging relevant differences when they do exist. Evaluate the meaning of the word “difference” carefully in relation to the target population, not the dominant group. For example, a researcher who wants to generalize the study results to people overall, or students overall, and so forth, should assess and report whether the sample studied is different from the target population and, if so, describe how it is different.
Be appropriately specific
Once you have determined which characteristics to describe, choose terms that are appropriately specific, which will depend on the research question and the present state of knowledge in the field. Do not mention characteristics gratuitously; however, when in doubt, be more specific rather than less because it is easier to aggregate data than to disaggregate them. Consider the appropriate level of specificity early in the research process—such as when designing the study—because it may not be possible to gather more data once the study is underway or finished. Using specific terms improves readers’ ability to understand the generalizability of your findings and other researchers’ ability to use your data in a meta-analysis or replication.
Examples of specificity by topic
Next, we present examples of specific language for the topics covered in these bias-free language guidelines; again, the proper choice will depend on the situation, and these examples represent just some of the possible options.
- When writing about age, exact ages or age ranges (e.g., 15–18 years old, 65–80 years old) are more specific than broad categories (e.g., under 18 years old, over 65 years old). Also include the age mean and median in addition to the range of ages to increase the specificity of the reporting.
- When writing about disability, names of conditions (e.g., Alzheimer’s disease) are more specific than categories of conditions (e.g., types of dementia) or general references such as “people with disabilities.”
- When writing about gender identity, descriptors with modifiers (e.g., cisgender women, transgender women) are more specific than descriptors without modifiers (e.g., women) or general nongendered terms (e.g., people, individuals; see the bias-free language page for gender for how to differentiate between gender and sex).
- When writing about people who took part in research, terms that indicate the context of the research (e.g., patients, participants, clients) are more specific than general terms (e.g., people, children, women).
- When writing about racial or ethnic groups, the nation or region of origin (e.g., Chinese Americans, Mexican Americans) is more specific than a generalized origin (e.g., Asian Americans, Latin Americans).
- When writing about sexual orientation, the names of people’s orientations (e.g., lesbians, gay men, bisexual people, straight people) are more specific than broad group labels (e.g., gay).
- When writing about socioeconomic status, income ranges or specific designations (e.g., below the federal poverty threshold for a family of four) are more specific than general labels (e.g., low income).
Guideline 2: Be sensitive to labels
Respect the language people use to describe themselves; that is, call people what they call themselves. Accept that language changes with time and that individuals within groups sometimes disagree about the designations they use. Make an effort to determine what is appropriate for your study or paper, particularly when these designations are debated within groups. You may need to ask your participants which designations they use and/or consult self-advocacy groups that represent these communities to research the issue if you are not working directly with participants. However, note that some individuals may use slurs or stigmatizing language to refer to themselves; researchers should use extreme caution before repeating this language because doing so can propagate that stigma (see the bias-free language pages on age and disability for more on the use of stigmatizing language when talking about these topics).
Acknowledge people’s humanity
Choose labels with sensitivity, ensuring that the individuality and humanity of people are respected. Avoid using adjectives as nouns to label people (e.g., “the gays,” “the poor”) or labels that equate people with their condition (e.g., “amnesiacs,” “schizophrenics,” “the learning disabled,” “drug users”). Instead, use adjectival forms (e.g., gay men, older adults) or nouns with descriptive phrases (e.g., people living in poverty, people with learning disabilities, people who use drugs). Some groups (e.g., the Deaf) have chosen to use a capitalized label to identify and promote a sense of unity and community (Solomon, 2012); use the label that the community uses, even when that label is adjectival (note, however, that not everyone who has hearing loss identifies as Deaf). In particular, the use of labels where disability is concerned is evolving, and people may disagree about the preferred approach. When writing about disability, person-first language (e.g., “a person with paraplegia” rather than “a paraplegic”), identity-first language (e.g., “an autistic person” rather than “a person with autism”), or both may be acceptable depending on the group you are writing about.
Provide operational definitions and labels
If you provide operational definitions of groups early in your paper (e.g., “participants scoring a minimum of X on the Y scale constituted the high verbal group, and those scoring below X constituted the low verbal group”), the best practice is to describe participants thereafter in terms of the measures used to classify them (e.g., “the contrast for the high verbal group was statistically significant”), provided the terms are not inappropriate. A pejorative label should not be used in any form. Abbreviations or series labels for groups usually sacrifice clarity and may be problematic: “LDs” or “LD group” to describe people with specific learning difficulties is problematic; “HVAs” for “high verbal ability group” is difficult to decipher. “Group A” is not problematic, but it is also not descriptive. Instead, ensure that operational group labels are clear and appropriate (e.g., “group with dysgraphia”).
Avoid false hierarchies
Compare groups with care. Bias occurs when authors use one group (often their own group) as the standard against which others are judged (e.g., using citizens of the United States as the standard without specifying why that group was chosen). For example, usage of “normal” may prompt readers to make the comparison with “abnormal,” thus stigmatizing individuals with differences. Likewise, contrasting lesbians with “the general public” or “normal women” portrays lesbians as marginal to society. More appropriate comparison groups for lesbians might be straight individuals, straight women, or gay men. Use parallel designations for groups, especially when presenting racial and ethnic identity information.
Be aware that the order of social group presentation may imply that the first-mentioned group is the norm or standard and that later-mentioned groups are abnormal or deviant. Thus, the phrases “men and women” and “White Americans and racial minorities” subtly reflect the perceived dominance of men and White people over other groups (furthermore, listing specific racial minority groups is preferable to writing about racial minorities in general) when talking about racial and ethnic identity. Similarly, when presenting group data, placing socially dominant groups such as men and White people on the left side of a graph or at the top of a table may also imply that these groups are the universal standard (Hegarty & Buechel, 2006). When referring to multiple groups, thoughtfully consider the order in which to present them. Do not put groups in order of social dominance by default; instead, consider options such as alphabetical order or sample size order. For ease of comprehension, list groups in the same order consistently throughout a paper.
Hegarty, P., & Buechel, C. (2006). Androcentric reporting of gender differences in APA journals: 1965–2004. Review of General Psychology, 10(4), 377–389. http://doi.org/10.1037/1089-26220.127.116.117
Solomon, A. (2012). Far from the tree: Parents, children, and the search for identity. Scribner.