Understanding Bias in Research
Study Validity: Is it True & Applicable?
Validity asks if a study's findings are trustworthy. Click each concept to learn more.
Internal Validity
Are the study's conclusions correct for the participants within this study?
Reflects how accurately the study results show the true situation for the specific group studied.
High Internal Validity means: The observed effects (e.g., treatment difference) are likely due to the intervention itself, not flaws or errors.
- Threatened by: Bias (selection, measurement, analysis), Confounding, Random Error (though bias/confounding are key).
- Example: A well-designed RCT carefully measuring outcomes in its 100 participants has high internal validity if it shows the drug worked *in that group*.
External Validity (Generalisability)
Can the study's findings be applied to other people or settings beyond this study?
Refers to the extent results can be applied to other populations, settings, or times.
High External Validity means: The results are likely relevant to your patients or the broader population.
- Threatened by: Highly selective sample (e.g., only young healthy males), artificial study setting, specific time period.
- Example: If the anxiety drug RCT only included young men, its findings might not apply well (low external validity) to elderly women with multiple health issues.
Random Error vs. Systematic Error (Bias)
Errors affect study accuracy and precision. See the difference using the target analogy. Adjust sample size and introduce bias.
🎯 Random Error (Chance/Imprecision)
- Unpredictable scatter around the true value.
- Due to sampling variability.
- Reduces Precision.
- Decreased by larger sample size & replication.
- Affects internal & external validity (less certainty).
🎯 Systematic Error (Bias)
- Consistent deviation in a specific direction.
- Due to flaws in design, conduct, or analysis.
- Reduces Accuracy (validity).
- NOT fixed by larger sample size.
- Mainly affects internal validity (results are wrong).
Exploring Types of Bias
Bias arises from flaws in how a study is designed, conducted, or analysed. Filter by category or click cards to learn about specific types.
Direction of Bias: Pushing the Results
Bias doesn't just cause errors; it pushes results in a specific direction relative to the 'null' (no effect) and the 'true' effect.