Bias

Question 1

Status-Quo Bias

Answer 1

• Default option is ranked more highly
• Hesitancy to change due to aversion to loss
• Endowment effect (overvalue what you own)

Question 2

Observed Association Explanations

Answer 2

• Observed association is true
• Due to chance
• Resulted due to bias
• Resulted due to confounding

Question 3

Precision

Answer 3

• Lack of random error (chance)
• Study size
• Study efficiency

Question 4

Validity

Answer 4

• Lack of systemic error
• Internal validity (bias, confounding)
• External validity (generalizability)

Question 5

Internal Validity

Answer 5

• Validity of inferences drawn that relate to study population
• Ability of test to measure what it sets out to measure
• Whether the experimental treatment makes a difference in this specific experimental instance

Question 6

External Validity

Answer 6

• Inferences about the extent to which a casual relationship holds over variations in persons/settings/treatments/etc
• Validity of inferences drawn as they relate to groups OUTSIDE the study population
• Extent to which the study findings can be generalized

Question 7

Bias

Answer 7

• Any trend in the collection, analysis, interpretation, publication or review if data that can lead to conclusions are systemically different from the truth
• Systematic error: incorrect estimate of association (RR or OR is wrong)

Question 8

Why worry about Bias/Confounding?

Answer 8

• Can cause incorrect results
• Over or underestimate association
• Change direction of association
• Quasi-experimental research is prone to bias and confounding
• Due to limited ability to remove the effect of other risk factors, ascertain exposure and outcome, representative sample

Question 9

Validity

Answer 9

• Randomization doesn’t eliminate threats to validity
• Minimizes threats to internal validity
• External validity can still be jeopardized due to stringent criteria

Question 10

Sources of Bias

Answer 10

• Caused by investigators or participants
• During design/conduct of study
• Occur in cross-sectional, case-control, cohort, or interventional studies

Question 11

Major Types of Bias

Answer 11

1. Selection Bias
2. Information Bias
3. Sampling or ascertainment bias
4. Response/Participation Bias
5. Lost to Follow-up Bias
6. Publication Bias

Question 12

Selection Bias

Answer 12

• Distortion in measure of association
• Results form manner in which subjects are selected into study population
• Results differ from if you had enrolled entire population

Question 13

Information Bias

Answer 13

• Flaw in measuring exposure or outcome

- Collection of incorrect information from subjects

Question 14

Sampling/Ascertainmetn Bias

Answer 14

• Healthcare access bias
• Length-bias sampling
• Exclusion
• Detection

Question 15

Response/Participation Bias

Answer 15

• Differences in characteristics between participants in study and non-participants
• Non-participants having more severe symptoms or conditions not responding to surveys for example

Question 16

Exclusion Bias

Answer 16

• Applying exclusions to only one group being studied (controls or cases)
• Can lead to over or under estimation

Question 17

Types of Information Bias

Answer 17

1. Observer/interviewer bias
2. Surveillance/detection bias
3. Recall bias
4. Reporting bias
5. Response fatigue Bias

Question 18

Observer/Interviewer Bias

Answer 18

• Knowledge of hypothesis
• Disease or exposure status knowledge
• Gathering selective data

Question 19

Surveillance/Detection Bias

Answer 19

-Disease ascertainment might be better in a more closely monitored population

Question 20

Recall Bias

Answer 20

• Accuracy problems due to memory

- Differential recall - subject with disease are more likely to remember exposure in the past

Question 21

Reporting Bias

Answer 21

-Selective suppression or revealing of information

Question 22

Response Fatigue Bias

Answer 22

-Incorrect answers given at the end of a long questionnaire

Question 23

Misclassification Bias

Answer 23

• Type of Information Bias
• Subjects are incorrectly catagorized with response to either exposure or outcome
• Placed into the wrong group
• Can introduce bias towards or away from the null

Question 24

Non-differential Misclassification

Answer 24

-Proportion of subjects misclassified is about the same in each group

Question 25

Differential Misclassification

Answer 25

- Information is more accurate in one study group | - Causes bias

Question 26

Methods to Control Bias

Answer 26

- Select groups by similar mechanisms or sources - Choose study population that is easy to follow - Maximize follow-up - Standardize data collection - Interviewers/data collectors should be blinded - Facilitate recall