Biases in study design

Question 1

random error

Answer 1

Non-differential misclassificaiton

Question 2

Systematic error

Answer 2

Differential misclassication
any process at any stage of inference
which tends to produce results or
conclusions that differ systematically
from the truth”
- fundamental threat to the internal validity of the study

Question 3

Selection bias

Answer 3

Any deviation from the true value . unknowingly selecting participants so that the results of the study do not represent the underlying population.
Can arise during recruitment - selection in as well as retaining - selection out (high number of drop outs, died etc..)

*
The E→O relationship in the study population differs from that in the source population

direction of bias can be away or towards the null, no way of knowing. also known as differntial misclassiciation. or systematic error because it changes the exposure outcome relationship among the cases and controls one more than another.

Question 4

how does Selection bias occur in case-control study

Answer 4

Selection of control, exposure status may vary leading to over/uner representation. we cannot assess the direction of bias it can go either way. closer to or away from the null. Controls did not represent the exosure distribution as the source population;

Question 5

how does selection bias occur in a cross-sectional study?

Answer 5

Healthy worker effect; comparing a working population to non-working population

Question 6

how does selection bias occur in a cohort study

Answer 6

loss to follow up / unable to retain participants eventually leading up to unbalanced distribution of diseased and non diseased.

Question 7

how does selection bias occur in a RCT?

Answer 7

not following allocaiton consealment; the researcher knows in advance who gets the treatment and who does not.

Question 8

Non Differential misclassification

Answer 8

misclassiciation does not differs in exposed vs unexposed as well as cases vs controls ; (because misclassification is similar and happens in both groups. is

Question 9

Differential

Answer 9

Exposure or outcome are misclassified and misclassifction is different in exposed vs unexposed as well as case and control

Question 10

Social desirability bias

Answer 10

participant want to appear socially desirable.. abortion rates will be lower in a conservative state, as they want to say they did not have it

Question 11

Recall bias

Answer 11

patients who experienced traumatic events may recall the event more. mothers giving birth to spina bifida child… may likely to remember exposures.. but for healthy babies they are not likely to remember the exposures.

Question 12

Case - control/cohort Interviewer bias

Answer 12

leading question, treatign unexposed/exposed askign about disease, cases/control differntly when asking abotu exposure

Question 13

In RCT Participant expectation bias

Answer 13

here the participant is aware of the benfits of the treament vs placebo

Question 14

In RCT observer/detection bias

Answer 14

here the participant is aware of the benfits of the treament vs placebo Lack of blinding can cause detection bias (knowledge of intervention can influence assessment or reporting of outcomes) ; exaggerated outcome.

Question 15

Assessment bias

Answer 15

physicians misclassifying patients when they know the history of exposure FPs might normally take family history into
consideration when making a dx of migraine (i.e., they are not
‘blinded’ to the exposure). Those without the family hx of
migraines might get misclassified as those without migraine