Anchoring decides where a judgement starts. Confirmation bias decides what evidence gets in afterwards. It is quiet, it feels like ordinary careful thinking, and it is the reason two people can look at the same situation and walk away certain of opposite things.
📘 What you need to know
Confirmation bias is the tendency to notice information that supports your existing view and overlook or ignore information that does not.
Like anchoring, it is a heuristic — a shortcut that saves effort but costs accuracy.
It affects everyday life, but it is most dangerous in professional judgement: medicine, law, diagnosis.
In research it means selectively recording results that fit the hypothesis and ignoring those that challenge it, which lowers validity and reinforces stereotypes.
Pavkov & Lewis (1989) is the key study, on race and the diagnosis of schizophrenia.
The bias does not feel like bias. It feels like being right. Evidence that fits is accepted quickly and remembered; evidence that clashes is questioned, explained away or simply not noticed at all.
The grey box never disappears — the clashing evidence still exists. It just never gets weighed. That is why confirmation bias is so hard to spot from the inside.
Definition to learn
Confirmation bias = seeking, noticing and remembering information that supports an existing belief, while overlooking or dismissing information that contradicts it.
A quick test on yourself. Think of a person you have decided is unreliable. How many times have they actually been on time? You probably do not know, because those occasions never got filed. Only the late ones did.
Confirmation bias inside research
This is where the topic gets serious. A researcher who expects a result may record supporting findings carefully and treat challenging ones as noise. The result looks like evidence, but the study has effectively agreed with itself.
Note step 3. Ignoring inconvenient data is not fraud — it usually happens without the researcher noticing, which is what makes blind procedures necessary rather than insulting.
Key study: Pavkov & Lewis (1989)
🧩 The study in four steps
Aim — to investigate whether race or ethnicity and confirmation bias influence clinicians’ decisions when diagnosing schizophrenia.
Participants — patients from four mental health hospitals in Chicago, USA. Two thirds were male and two thirds aged 18–34. They came from neighbourhoods representing both Black-dominant and White-dominant populations.
Procedure — participants were interviewed in hospital. One interview was diagnostic, carried out by an expert who had not been told the aim of the research. The researchers ran a second interview on social-psychological measures, such as how socially integrated the patient was and how aware they were of their condition.
Findings — Black patients were more likely than White patients to receive a diagnosis of schizophrenia. This was especially pronounced in hospitals located in Black-dominant neighbourhoods, and it happened even where patients had previously been diagnosed with a different mental illness.
Conclusion: some patients may be misdiagnosed on the basis of skin colour. Clinicians may be relying on stereotypes rather than objective evidence — evidence of confirmation bias.
Why the blind expert matters. The diagnostic interviewer did not know the aim of the study, which removes one obvious source of researcher bias. Mention that detail — it shows you can evaluate a design, not just describe it.
Evaluating confirmation bias
Strengths
Understanding it can change behaviour. Knowing about confirmation bias helps people avoid labelling or stereotyping those outside their own cultural or social group.
Rich data. Interviews give detailed material with strong explanatory power and real insight into participants’ experiences.
Serious real-world stakes. Research in a clinical setting shows the bias mattering where it does most damage.
Limitations
Extraneous variables. Using four separate hospitals means differences in staff, patients, equipment and location, which makes it hard to draw a clean conclusion.
Social desirability bias. Interview participants may answer as they think they should, lowering validity.
Alternative explanations. Apparent bias may partly reflect clinician fatigue, limited time and resources, or understaffing. Basing misdiagnosis on a single interview may also be an unrealistic expectation for accurate diagnosis in the first place.
Link to concepts
Causality — confirmation bias is a subjective, ill-defined variable, hard to measure directly. Findings that look like bias may reflect other factors, so causal claims should be cautious.
Bias — the study itself may suffer sampling bias, drawn from four Chicago hospitals only, which limits representativeness. Replication in other states and countries with more diverse samples would be needed to assess the effect properly.
Exam practice
WORKED EXAMPLE
Explain how confirmation bias may affect decision-making. [3 marks]
1. Define
Confirmation bias is favouring information that supports an existing belief and dismissing information that contradicts it.
2. MechanismExpectation shapes what is noticed, so the decision is made on filtered evidence.
3. Evidence
Pavkov and Lewis found Black patients were more likely to be diagnosed with schizophrenia, even when previously diagnosed with something else.
Define + mechanism + evidenceReport the finding factually. No editorialising needed — the finding speaks.
WORKED EXAMPLE
Evaluate one study of a cognitive bias. [ERQ paragraph]
Strength
Pavkov and Lewis used a diagnostic interviewer blind to the aim, reducing researcher bias in the key measure.
LimitationHowever four different hospitals introduce extraneous variables in staffing, resources and patient population.
Alternative
Fatigue, understaffing or time pressure could produce the same pattern without confirmation bias.
Judgement
The finding is important and troubling, but the design cannot isolate bias as the cause.
Important finding, uncertain mechanismOffering a rival explanation is worth more than listing three weaknesses.
💡 Exam tip
Give an everyday example and a professional one. The contrast shows the bias scales up.
Say that confirmation bias is difficult to operationalise — it is the strongest methodological criticism available.
Link it to the 44 thieves study from Human Development, where Bowlby interviewed his own participants. Cross-topic links score well.
Use “may be” language throughout. This study describes an association in one city.
The blind interviewer is a design strength worth one sentence in any evaluation.
Reflexivity is the standard remedy — name it.
⚠️ Common mix-up
Confirmation bias is not lying. It usually happens without awareness.
It is not the same as anchoring. Anchoring sets a starting value; confirmation bias filters later evidence.
This is not an experiment. There is no manipulated independent variable.
Do not overstate the sample. Four hospitals in one city is not the USA.
Rich data is not the same as valid data. Interviews can be detailed and still distorted by social desirability.
Bias in a clinician is not necessarily prejudice. The study shows a pattern in outcomes, not the intentions behind it.
Up next: Classical Conditioning: Learning by Association — we leave thinking behind and turn to learning, starting with the simplest mechanism psychology has: two things happening together, over and over.
Want this explained one-to-one?
Book a free session with an experienced IB Psychology tutor and get your trickiest topics made simple.
