Assessing And Improving Reliability

Question 1

Reliability is a measure of?

Answer 1

Consistency

Question 2

How can you test if a tool is reliable

Answer 2

Test retest method
(Same group of people asked to undertake the research measure e.g. questionnaire, on different occasions)

Question 3

(Test retest method) what happens after the measure has been completed on 2 separate occasions

Answer 3

Two scores are then correlated

Question 4

What’s a good correlation

Answer 4

+0.8
(Perfect is 1)

Question 5

What is a significant correlation an indication of

Answer 5

Good reliability

Question 6

What’s an issue with the re-test method

Answer 6

If you test on two occasions that are too far apart of course ur gonna get diff results that make it unreliable?
E.g. testing for depression a year apart - sb might recover by then

Question 7

What’s inter obser reliability/ inter rater

Answer 7

The extent to which two or more observers record and observe behaviour in a consistent manner

Question 8

How to improve the reliability of questionnaires

Answer 8

• it’s possible to identify which questions that are having the biggest impact upon the reliability and then adjust them as necessary e.g. rewriting them so that they’re in a manner that reduces the potential for them to be incorrectly interpreted

Question 9

How could we improve the reliability of interviews

Answer 9

• ensure the same interviewer conducts everything (reduces researcher bias) so the questions are asked the same way + some researchers might do leading questions and others won’t so keep it the same for everyone
• change the interview from unstructured to structured to limit researcher bias

Question 10

How to improve the reliability of experiments

Answer 10

• follow standardised procedures
• take more control over extraneous variables (make em confounding)

Question 11

How to improve the reliability of observations

Answer 11

• observations can lack objectivity as they rely on the observer right
  So we have to improve reliability would be to operationalise the behavioural categories, so they’re clear and there’s no overlap and no need for personal interpretation