Measurement and bias

Question 1

how is a hypothesis written

Answer 1

exposure to X is associated with an increased/decreased risk of Y

Question 2

2 features of hypotheses

Answer 2

testable

specific

Question 3

two types of errors?

Answer 3

random error

systematic error

Question 4

3 types of random error

Answer 4

sampling error
individual biological variation
measurement error

Question 5

measurement error formula

Answer 5

measured score = true score +/- error

Question 6

example of systematic error

Answer 6

bias

Question 7

selection bias

-what point in study does it arise

Answer 7

allocation stage

Question 8

selection bias: what is it

Answer 8

non random allocation

researcher manipulates allocation

Question 9

selection bias: how to control

Answer 9

random allocation

allocation concealment

Question 10

performance bias

• when
• what

Answer 10

when interventions are being delivered

either

• experimenter treats either group differently
• participant realises which they’ve been allocated to and acts differently

Question 11

performance bias

-controlling

Answer 11

double blind

placebo

Question 12

detection bias

• when
• what

Answer 12

at the data collection stage

either

• some differences in how outcomes are determined between groups
• researcher question’s individuals outcome variable, causing them to change it

Question 13

detection bias

-controlling

Answer 13

blind outcome assessors

Question 14

measurement bias

• when
• what

Answer 14

when measuring either variable

either

• measurement tool is wrong
• one researcher consistently measures something wrong
• different tools per group

Question 15

measurement bias

-fixing

Answer 15

• correct the tool
• train the researcher
• use consistent tools/protocols
• collect data from multiple sources

Question 16

recall bias

-what

Answer 16

participant forgets or omits information

Question 17

recall bias

-controlling

Answer 17

have a good research question
have a good questionnaire
have a good researcher

Question 18

attrition bias

-what

Answer 18

differential attrition between groups

Question 19

attrition bias

-controlling

Answer 19

may use intention to treat analysis

Question 20

publication bias

-what

Answer 20

research that has new/expected findings is more likely to be published than research with null/unexpected findings

Question 21

describe intention to treat analysis

Answer 21

use results of all people regardless of attrition/not following guidelines

Question 22

opposite of intention to treat analysis?

Answer 22

per protocol analysis

Question 23

what % of attrition is acceptable to lead to little bias?

Answer 23

<5%