Lecture 5: Bias & Misclassification

Question 1

what 3 aspects of a study, must researchers evaluate for internal validity before they can declare a true association?

Answer 1

1. confounders and effect modifiers
2. bias and misclassification
3. statistical significance

Question 2

definition of bias

Answer 2

systematic error in study design or conduct, resulting in false results

Question 3

how do we control for bias after a study has been conducted?

Answer 3

we can’t

must be done before study is implemented

Question 4

what are the 3 elements of bias impact?

Answer 4

1. source or type
2. magnitude
3. direction

Question 5

the 2 main categories of bias

Answer 5

selection-related

measurement-related

Question 6

selection-related bias

Answer 6

any aspect in the way the researcher selects study subjects differently

Question 7

measurement-related bias

Answer 7

any aspect in the way the researcher collects data differently

Question 8

types of selection bias

Answer 8

healthy-worker bias

self-selection bias

Question 9

other names for responder bias

Answer 9

self-selection bias

participant bias

Question 10

what type of bias is easily seen in prospective cohort studies?

Answer 10

healthy-worker bias

Question 11

measurement bias can be ……?

Answer 11

subject related
observer related
screening related

Question 12

subject related bias

Answer 12

type of measurement bias
recall bias
hawthorne effect
contamination bias
compliance bias
lost to follow up bias

Question 13

groups being studied have different withdrawal rates

Answer 13

lost to follow up bias

Question 14

groups being interventionally studied have different adherence with study protocols

Answer 14

compliance/adherence bias

Question 15

members of the control group, receive treatment or are exposed to the intervention being studied

Answer 15

contamination bias

take drugs they shouldn’t

Question 16

individuals alter their behavior because they know they are apart of a study under observation

Answer 16

hawthorne effect
or
observer effect

Question 17

a differential level of accuracy in provided information between study groups

Answer 17

recall bias or reporting bias

Question 18

observer-related

Answer 18

type of measurement bias
interviewer bias
surveillance bias

Question 19

a systematic difference in interpreting done by the researcher

Answer 19

interviewer bias

not always consciously done

Question 20

different evaluation or diagnosis between study groups due to an observers expectations

Answer 20

diagnosis/surveillance/expectation bias

Question 21

screening-related

Answer 21

type of measurement bias

lead-time bias

Question 22

an apparent benefit from a healthcare screening due to the early detection of disease despite an unchanged clinical outcome

Answer 22

lead-time bias

Question 23

source of measurement bias, especially observer-related bias

Answer 23

misclassification bias

Question 24

error in classifying either disease or exposure status, or both

Answer 24

misclassification

Question 25

the 2 types of misclassification

Answer 25

differential | non-differential

Question 26

misclassification of exposure which is unrelated to the other exposure

Answer 26

non-differential = error in both groups equally

Question 27

not different

Answer 27

non-differential

Question 28

in non-differential bias, the measure of association is always....?

Answer 28

type of misclassification | OR is always moved towards 1.0

Question 29

misclassification of exposure which is related to the other exposure

Answer 29

differential misclassification

Question 30

in differential bias, the measure of association is.....?

Answer 30

can be inflated or depreciated

Question 31

ways to control for bias

Answer 31

blinding/masking multiple data sources randomly allocate observers for data collection minimize loss to follow up bias

Question 32

random allocation and randomization are ways to control for....?

Answer 32

``` random = bias control randomization = confounding control ```

Question 33

ways to control for confounding variables

Answer 33

``` restriction randomization matching stratification multivariate analysis ```