What bias changes

Bias is a systematic tendency in thinking or research that favours particular judgments, observations or conclusions over others. The key word is systematic. Bias pushes in a particular direction; it doesn't just produce unpredictable mistakes.

Dishonesty isn't required. Researchers may sincerely believe they are being fair even as their expectations shape what they notice. A participant, too, may give a misleading answer to make a good impression, rather than to undermine the research.

The supplied guide extract identifies the terms below as supporting ideas, not separate assessable terminology–learning-objective rows. They help explain where bias comes from and what it affects, as well as how researchers can address it.

There are several points at which bias can enter a study: the assumptions behind it, the selection of participants, interactions between researcher and participant, the interpretation of evidence, or the decision to publish findings. Improving one part of the research process does not automatically protect the others.

Assumptions about what psychology should explain

Positivism is a philosophical approach that treats systematic observation and measurement as the basis for establishing knowledge about the world. In psychology, this encourages researchers to investigate behaviour through evidence that others can check, rather than relying only on personal impressions.

A recorded observation can be scrutinised in a way that an unsupported opinion cannot. That can constrain bias. But measurement doesn't automatically make research neutral: researchers still decide what to measure and how to measure it, as well as which explanations to consider. The feature that is easiest to measure is not necessarily the most important one.

Reductionism is an explanatory approach that accounts for a complex phenomenon through its simpler components or a more basic level of explanation. An explanation of irritability, for example, might focus on a biological process and set aside the person's relationships and circumstances. The approach can isolate a useful mechanism. It can also privilege one explanation while overlooking relevant influences.

Holism is an explanatory approach that examines a phenomenon as an integrated whole, including interactions between its components. A holistic account of the same behaviour might examine how biological processes work together with personal interpretations and social circumstances. Attention to these interactions is its strength. The difficulty is testing such a broad explanation precisely.

Neither approach is automatically biased. What matters is whether the approach excludes evidence relevant to the conclusion. A tightly focused question may call for a tightly focused explanation; presenting that explanation as a complete account without justification is where it becomes misleading.

Determinism is a philosophical position that regards behaviour as caused by preceding conditions, rather than as independent of those conditions. Biological determinism places those determining causes in biological factors. Environmental determinism, by contrast, places them in environmental influences and experience.

These positions can draw attention to different evidence. Researchers committed to one may dismiss findings that support the other. Don't confuse a deterministic position with evidence that one factor completely explains behaviour. Showing that a biological or environmental factor has an influence does not establish that all other influences are irrelevant.

Bias in everyday thinking and interpretation

Cognitive biases are systematic tendencies in mental processing that influence judgment, memory or decision-making in particular directions. They help explain how people reach distorted conclusions without consciously choosing to be unfair.

Confirmation bias is a cognitive bias in which people preferentially seek, interpret or remember information that supports an existing belief. Imagine a tutor who believes students sitting near the back are disengaged. A back-row student looks away, and the tutor notices; the same behaviour near the front goes unnoticed. The belief directs attention towards selected evidence, which then appears to confirm it.

Participants can show confirmation bias, as can researchers and readers of research. One useful check is to deliberately consider evidence that would count against the preferred explanation. Collecting more evidence isn't enough if only evidence that agrees is collected.

Bias within a study

Researcher bias is a systematic influence of researchers' expectations, preferences or assumptions on the conduct or interpretation of a study. It may shape how prompts are worded or observations are recorded. It can also affect how an ambiguous response is interpreted.

Suppose a researcher expects an activity to improve cooperation. They might interpret an uncertain gesture as helpful in the activity condition, but not in the comparison condition. Clear recording criteria and consistent instructions can reduce this influence. Where feasible, concealing condition information from observers can help too. None of these measures guarantees that every research decision is free from bias.

Participant bias is a systematic influence arising from participants' expectations, motives or perceptions of the research situation that changes their responses. If a researcher appears to value considerate behaviour, a participant might report being more considerate than they really are. Their response then reflects the research situation as well as the characteristic under investigation.

Neutral wording and credible privacy arrangements can reduce the pressure to give an approved answer. Researchers need to attend to how participants experience these arrangements. Saying that a response is confidential is less useful if participants don't believe it.

Who enters the sample?

Sampling techniques are procedures used to select participants from a population for a study. How researchers select people determines whose experiences can contribute to the findings.

Sampling bias is a systematic distortion caused when the selection process makes some members or characteristics of the population more likely to be represented than others. Suppose researchers use only an online invitation to investigate access to community activities. They may miss people with limited internet access, even though those people's experiences are directly relevant.

Selecting people based on availability is practical. It may, however, repeatedly favour those who are easy to reach. Letting people volunteer can favour those with particularly strong interests or opinions. Giving each population member an equal chance of selection reduces selection bias only if the selection list adequately covers that population. Non-participation can still change the final sample.

When examining sampling, link the recruitment method to the likely distortion. Ask who could enter and who was unlikely to enter. Does that difference matter to the conclusion? A large sample cannot repair a systematically selective recruitment process.

Gender and cultural bias

Gender bias is a systematic tendency in research or explanation that privileges, misrepresents or inadequately represents people on the basis of gender. It can arise when findings from one gender group are treated as universal. Interpreting behaviour by different standards for different groups can introduce it too.

For example, calling the same assertive behaviour leadership in one group but hostility in another introduces bias through interpretation. Including different groups helps, but representation alone isn't enough. The measures and explanatory assumptions need scrutiny as well. An observed difference between groups is not, by itself, evidence of bias.

Cultural bias is a systematic tendency in research or explanation that privileges one cultural perspective or misrepresents people whose cultural circumstances differ. A measure may take behaviour valued in one cultural setting as the universal standard for psychological functioning.

Imagine a questionnaire that assumes speaking frequently in a group always indicates confidence. Where listening before speaking is valued, the scores may reflect cultural expectations rather than the intended characteristic. Translating the words would not necessarily solve this problem. The meaning of the behaviour needs consideration too.

For both gender and culture, ask whether the evidence supports the scope of the claim. Findings from a particular group can be useful without being treated as a description of everyone.

Making interpretations more trustworthy

Credibility is a quality of research findings that makes them believable as an evidence-supported account of the phenomenon investigated. A plausible story isn't enough. Readers should be able to see how the evidence leads to the interpretation.

Researchers strengthen credibility by explaining their decisions and considering evidence that challenges their interpretation. Showing how conclusions relate to participants' accounts or recorded observations also helps. These practices make selective interpretation easier to identify.

Reflexivity is a research practice in which researchers critically examine how their own position, assumptions and interactions may influence the research process and its findings. It involves examining one's influence, not simply announcing that one might be biased.

A researcher might record why they interpreted a response in a particular way. They could then consider how their relationship with the participant shaped that response, revisiting the interpretation if contradictory evidence appears. Reflexivity makes the researcher's role visible; it doesn't remove that role.

Inter-rater reliability is the extent to which different observers or coders agree when independently judging the same material using the same criteria. When two observers classify the same interaction differently, the criteria may be unclear or open to individual interpretation.

To improve agreement, researchers can define categories precisely and practise using them. Observers can also code independently before comparing judgments. Agreement helps establish consistency, but it does not prove that the judgments are correct: two observers may agree because they share the same biased assumption.

Which findings become visible?

Publication bias is a systematic distortion in the available research evidence caused by some findings being more likely to be published than others. Striking results, or results that support a preferred claim, may become more visible than those that do not.

Suppose several investigations examine whether an activity improves concentration. Studies reporting improvement are published, while those finding no clear improvement remain unavailable. The reader then encounters a selectively positive picture. Each published study could be competently conducted, yet the available body of evidence could still be biased.

Publication bias operates beyond a single study. Addressing it requires making research plans and results more visible, including findings that are unremarkable or contradict expectations. Distinguish between what the research community has investigated and what readers are able to find.