Topic 13: systematic error in research

Question 1

Give 2 types of bias in research

Answer 1

1) Selection bias = from mistakes during selection of study sample
2) Information bias = from mistakes during measurment of variables

Question 2

Define information bias

Answer 2

• Wrong/inaccurate assessment of exposure/disease outcome variables
• Due to researcher = unintentionally
• Due to participant = un/intentionally

Question 3

Describe information bias based on researcher

Answer 3

• Wrong diagnosis = clinical error
• Wrong measurement = faulty intrument
• Wrong measurement = poor training
• Mistake recording/transferring data = paper to electronic

Question 4

Describe information bias based on participant

Answer 4

• Wrong answer:
  > Misinterpret question
  > Sensitive issue
  > Poor recall
  > Intentionally

Question 5

Define response bias

Answer 5

• Information bias from participant action

Question 6

Give 6 types of information bias

Answer 6

1) Misclassification
2) Recall
3) Interviewer
4) Hawthorne effect
5) Observer
6) Surveillance

Question 7

Define misclassification bias

Answer 7

• Error in classifying patient as being exposed/having disease
• Due to instrument inaccuracy
• In any study design

Question 8

Define recall bias

Answer 8

• Report past exposures wrong
• May over/under report
• In case-control studies

Question 9

Define interviewer bias

Answer 9

• Researcher’s knowledge + perception = influnce questions
• May ask leading questions/inconsistent between cases
• In any study design = more when researcher not blinded to participants status

Question 10

Define Hawthorne effect

Answer 10

• Participants act differently when they know they are being observed

Question 11

Define observer bias

Answer 11

• Researcher’s preconceived expectations of what they should find
• More if not blinded

Question 12

Define surveillance bias

Answer 12

• Group with known exposure = followed more closely than comparison group

Question 13

Give methods to reduce bias

Answer 13

1) Design study = minimize chance for bias
- Use more than 1 control
2) Mutually exclusive criteria to define exposed = no overlap between exposed/non-exposed
3) Strict guidlines for data collection
- Train assessors to obtain data consistently
- More than 1 assessor = identical training

Question 14

How can information bias affect study?

Answer 14

• To investigate association = factors measured accurately
• Bias = association distorted
• From participants > researchers
• Affects studies relying on self-report = assume information bias to some extent

Question 15

Describe validity

Answer 15

• Extent of measuring tool measures accurately what it’s meant to measure
• Criterion validity = result from tool compared to gold standard

Question 16

Define internal validity

Answer 16

• Findings had no bias/error = conclusions correct

Question 17

Define external validity

Answer 17

• If a study has no bias/error doesnt mean findings can be generalized to other circumstances
• E.g. a study for effect of medication on men cannot be extrapolated to women

Question 18

Describe reliability

Answer 18

• Overall consistency of measure = producing same results under same conditions again
• Intra-observer reliability = consistency in results from tool used by same researcher on many occasions
• Inter-observer reliability = consistency in results of same group between 2+ researchers