Please enable JavaScript.
Coggle requires JavaScript to display documents.
Bias - Coggle Diagram
Bias
identificatiion of bias
Confounding
The presence of a third variable that
affects exposure and outcome distorts the obs association between exposure and outcome
-
diarrheal disease (Y) in children --> P (Y) . family with cows P (Y=1| A=1) and family without cows P (Y=1|A=0). Children who belong to families with cows have lowes rates diarrhea. P (Y|A=a) > P (Y)
Criteria confounder: The variable must be an extraneous risk/protective factor for outcome, he variable must be associated with exposure of interest in the
population being studied, The variable must not be caused by the outcome or exposure (e.g.,
cannot be on the causal pathway between exposure and outcome
-
Exchangeability: comparison and exposure group must be identical in all factors except exposure status --> We want an individual’s potential outcomes under an exposure
to be independent of the exposure assignment
Selection bias
exposure and outcome influences whether individual will make it into final analytic sample. Estimated measure of occurance measure of association obtained from study population differs systematically from the estimated that obatined information from source pop available
result from selection into study, loss to follow up or analytic choices (missing data)
Reason
Studies of disease occurrence (e.g., estimating risk or prevalence). he outcome, or ancesters of the outcome, affects participation.
-
-
Both the exposure and the outcome of interest are related to entry into the study and/or continued participation in the study
-
-
reason
systematic error--> sampling design, data collection process or analysis that effect how close observed effect is to the true effect (bias) or accuracy
random error--> random sampling, repeat the study unlimited times, using random sample. --> see CI and about precision.
-
-
