Bias, Confounding and Interaction

Question 1

Random Error

Answer 1

• “accidental” error that occurs because of variability in measurement

Question 2

Systematic Error

Answer 2

• error in measurement that occurs in the development of an instrument
• no matter how many times you measure it will always be wrong
• or repeated mis-use of that instrument

Question 3

Selection Bias

Answer 3

• happens when some feature of data collection makes the participation of some people in the study, less likely than others

Question 4

Non-Response

Answer 4

• Subjects declining to participate, or don’t further agree with the study

Question 5

Information Bias

Answer 5

• the data that we collect leads to research participants being misclassified
• also called misclassification bias

Question 6

Types of Misclassification Bias (2)

Answer 6

1. Differential Misclassification

2. Nondifferential Misclassification

Question 7

Differential Misclassification

Answer 7

• happens when one study group or the other is misclassified MORE frequently than the other

Question 8

Nondifferential Misclassification

Answer 8

• errors in both/any study group (ie both case + controls)
• mainly a problem to do with data collection
• less likely to find an association between groups

Question 9

Reporting Bias

Answer 9

• participants are less likely to report an exposure that may-well be related to their disease, but which is socially undesirable
• “covering up” or “creating” a new one

Question 10

Information Bias comes from (6)

Answer 10

• interviews
• surveillance
• recall bias
• reporting
• data taken from records
• surrogate bias (info from close family on behalf of loved one)

Question 11

Confounding

Answer 11

• is not an error found in the study, but a phenomena that is identified
• an extraneous factor identified

Question 12

How can you control for confounding? (4)

Answer 12

1. Randomization
2. Restriction
3. Stratification
4. Matching

Question 13

Randomization

Answer 13

• to ensure equal distributions of characteristics between groups

Question 14

Restriction

Answer 14

• remove potential influences of a confounder by eliminating that characteristic in our sample population

Question 15

Stratification

Answer 15

• create study groups that are less vulnerable to confounding
  Ex. if age is a confounder, use study groups with a smaller age range

Question 16

Matching

Answer 16

• match subjects in groups according to the value of the known confounding variables to ensure equal distribution in each group.

Question 17

Interaction

Answer 17

• is concerned with the effects of multiple exposures, such that the effect of any given exposure is amplified / mitigated by the presence of another factor

Question 18

T/F - Exposure to a single causal factor, incidence rates will be higher than the background risk

Answer 18

True

Question 19

T/F - If interaction is not happening, the incidence rates for those with both exposures will simply be the risk

Answer 19

True

Question 20

T/F - If the incidence rate in those exposed to both factors is higher than the risks added together, we can conclude that interaction is occurring, such that exposure to both factors multiplies the risk

Answer 20

True

Question 21

Synergism

Answer 21

• a situation in which the combined effect of two or more factors is greater than the sum of their solitary effects

Question 22

Antagonism

Answer 22

• combined effect of two or more factors is SMALLER or WEAKER than that expected from the effect of one factor in the absence of the others