Brush up M&T - Artikel Z - Bias and confouding
14 important questions on Brush up M&T - Artikel Z - Bias and confouding
What are four explanations for an observed association?
2. Confounding
3. Bias (systematic error)
4. Causality
What are two types of bias?
- information bias = invalid comparison due to incomparable data collection
In which study design can selection bias be found?
2. Cohort study = in either the exposed or unexposed group the disease (outcome) occurs more or less than truly present
--> most present in case-control studies because controls are selected by the researcher
- Higher grades + faster learning
- Never study anything twice
- 100% sure, 100% understanding
In which two study designs does information bias occur?
Cohort study = data collection differs between the exposed and unexposed with regard to the disease/outcome
What is recall bias, and in which study does it occur the most?
Occurs in case-control studies
Solution: objective data collection and comparable data collection
Differential vs. Non differential misclassification
- cohort = misclassification of the outcome depends on the exposure
- case-control = misclassification of the exposure depends on the disease/case status
Non-differential misclassification
- misclassification of outcome (cohort) or exposure (case-control) does not depend on exposure (cohort) or disease (case-control)
- no systematic error, no bias -> dilution of the effect
- occurs in both groups, equal between groups
Selection vs. Information bias
• Invalid comparison through differential selection of groups
• Mainly in case-control studies
• Mainly incorrect selection of the control group
Information bias:
• Invalid comparison through differential data collection
• Recall bias: mainly in case-control studies
• Observer bias: both in case-control as in cohort studies
What is the defenition of confounding?
What is the difference between confounding and bias?
Confounding concerns the causal interpretation of the association
What are the three properties of a confounding factor?
2. The confounding factor is associated with the exposure
3. The confounding factor may not be a consequence of the exposure: may not lie in the causal pathway between exposure and outcome
How to prevent confouncint/adjust confounding?
Then, you prevent/adjust for confounding:
- In the design of the study (prevent)
- randomisation = all known and unknown factors are equally distributed
- restriction = known confounding factors
- matching = known confounding factors
- In analysis (adjust)
- stratification (known)
- multivariate regression analysis (known)
What is the disadvantage of matching?
How is stratificatino used in practice?
2. Calculate effect for each stratum
3. Pool effects of the strata
4. Compare corrected effect estimate with the crude estimate
Which models can be used for regression analysis (to adjust for confounding)?
- logistic regression
- cox regression
The question on the page originate from the summary of the following study material:
- A unique study and practice tool
- Never study anything twice again
- Get the grades you hope for
- 100% sure, 100% understanding
![Trustpilot Logo](/images/design_2020/lp/clp-trustpilot-en.png)