Yr 4 - General

Question 1

What is always good to remember about trying to eliminate bias?

Answer 1

That in the attempt to eliminate 1 particular bias, its important to be aware that another + different type of bias may inadvertently be introduced.

Question 2

What are the 2 broad categories of error?

Answer 2

Random

Systematic

Question 3

Random error

Answer 3

Occurs by chance.

Result of small fluctuations.

Can be minimised by increasing sample size.

Question 4

Systematic error (bias)

Answer 4

Tends to lead to an erroneous conclusion.

Caused by a feature of the design or conduct of the study.

Remains regardless of sample size.

Question 5

The larger the size of the study sample, the more closley the sample means will be dispersed around the ….

Answer 5

true population mean.

Question 6

What are the 2 possible unfavourable outcomes of systematic bias/error?

Answer 6

It can create spurious association where no real relationship exists between 2 variables

OR

It can mask a real association between 2 variables.

Question 7

What are the 3 most commonly documented types of systematic bias?

Answer 7

Selection bias

Confounding variable bias

Information bias

Question 8

How can selection bias be minimised?

Answer 8

By properly conducting a random sample for the population of interest.

Question 9

Hawthorne effect

Answer 9

Individuals modify their behaviour because they know they’re being observed.

Question 10

Bias will be present if the stufy population is restricted to only those who volunteer to participate. But there would be no selection bias if a….

Answer 10

random sample of the target population was recruited.

Question 11

What is meant by confounding variable bias?

Answer 11

When a confounding variable can create an erroneous relationship between 2 variables or can hide or mask a real relationship.

Question 12

How can confounding variables be controlled in a study design?

Answer 12

By carefully matching age, gender and other relevant factors with appropriate statistical analysis.

Question 13

What can come under information bias?

Answer 13

Measurement bias

Misclassification of outcomes

Inaccurate information

• All can impact study findings.

Question 14

How can measurement bias be caused?

Answer 14

By non-response and lack of complete follow-ups in studies.

By the use of defective measurement tools i.e poorly caliberated weighing scales.

By use of an inappropriate tool such as a poorly designed interview form or questionnaire.

Question 15

MEASUREMENT BIAS

What can non-response and/or lack of complete follow up in studies result in?

Answer 15

Missing information.

Question 16

How may misclassification bias occur?

Answer 16

If a categorical exposure +/or ouctome is misclassified.

It may occur where there is an element of subjectviity or difficulty in determining classification.

Question 17

List some bias-minimising strategies for misclassification bias

Answer 17

Use of predetermined criteria pertaining to data classification + data-handling training in the standardisation of methods of classification.

Question 18

What could outliers be?

Answer 18

Genuine observations

OR

Typing errors.

Question 19

What should happen if an outlier is detected?

Answer 19

Raw data should be re-examined to check if the outlier is genuine (rather than typographical or a measurement error).

The effect of the outlier can be determined by incl. + excl the outlier in the analysis, recording it in the findings + then brought to the attention of the reader.

Question 20

When might survivor bias occur?

Answer 20

If ind don’t survive long enough to receive treatment.

This can introduce artificial survival advantage among healthier subjects receiving the treatment.

Question 21

When may length-based (prognostic) bias occur?

What can it further introduce?

Answer 21

When screening may detect ind w. less severed disease than the usual route of presenting to a doctor w/ troublesome symptoms.

= Can introduce bias becasue the screened group, on avg, will have less severe disease + consequently may have a better outcome.

Question 22

When can lead-time bias occur?

Answer 22

When earlier detection through screening may lead to the perception that survival is longer simply because the diagnosis was made earlier than it would otherwise have been through a visit to the doctor.

Question 23

What are the 2 types of observer bias?

Answer 23

Variation within the observer.

Variation between observers.

Question 24

OBSERVER BIAS

Variation within the observer

Answer 24

Considered to be random

Question 25

OBSERVER BIAS

Variation between observers

Answer 25

Can be caused by different criteria + is likely to lead to systematic bias.

Question 26

When can observer reporting bias (a.k.a assessment bias) occur?

Answer 26

When 1 observer tends to under or over report a particular variable.

Question 27

How may observer bias be minimised?

Answer 27

By providing thorough training + including the standardiation of data collection methods prior to the commencent of a study.

Question 28

In what type of studies may recall bias occur most commonly?

Answer 28

Retrospective case-control studies as the knowledge of being a case (with disease) or a control may affect how an individual remembers their history.

i.e patients with the disease may be more likely to remember + report events that occurred aroudn the same time the disease first occurred.

– A young female w/ a myocardial infarction may be more likely to report the use of birth control pills than a similar yound woman without myocardial infarction.

Question 29

Strategy to minimise recall bias

Answer 29

To back up the information using multiple sources.

i.e through medical records or by applying triangulation whereby information is gathered + analysed in more than 1 way.

Question 30

What is confirmatory bias?

Answer 30

Providing extra emphasis to information that supported the researchers hypothesis + minimising the information that fails to support this hypothesis.

Question 31

What does avoiding confirmatory bias involve?

Answer 31

Requires impartiality in all stages of the study.

Incl the literature review and the collection of information used as data.

Question 32

How is selection bias often reduced in quantitative studies?

Answer 32

By the random selection of participants.

Question 33

What can not accounting for participants who withdrew from a study or are lost to follow-up result in?

Answer 33

Sample bias.