Types of Reliability Flashcards
What is reliability?
Refers to consistency.
If a study is repeated using the same method, design and measurements, and the same results are obtained, the results are said to be reliable.
How can reliability be improved?
Developing more consistent forms of measurement.
Using clearly defined operational definitions.
Improving inter-observer reliability.
What is internal reliability?
Concerns the extent to which something is consistent within itself.
For example, a set of scales should measure the same weight between 50 and 100 grams as between 150 and 200 grams.
What is external reliability?
Concerns the extent to which a test measures consistently over time.
What are the 3 ways of assessing reliability?
Split-half method.
Test-retest method.
Inter-rater (observer/interviewer) reliability.
Outline split-half method in assessing reliability.
Measures internal reliability by splitting a test into two and having the same participant do both halves.
If the two halves of the test provide similar results this indicates that the test has internal reliability.
Outline test-retest method in assessing reliability.
Measures external reliability, by giving the same test to the same participants on two occasions.
If the same result is obtained, then reliability is established.
Outline inter-rater (observer/interviewer) reliability in assessing reliability.
Is a means of assessing whether different observers are viewing and rating behaviour in the same way.
This can be achieved by conducting a correlation of all the observers’ scores, with a high correlation indicating that they are observing and categorising behaviour consistently.
Inter-observer reliability is improved by developing clearly defined and separate categories of observational criteria.
How can results be reliable but not valid?
If results are unreliable, they cannot be trusted.
However, results can be reliable, but not be valid (accurate).
For example, if you add up 1+1 several times and each time calculate the answer as 3, then your result is reliable (consistent), but not valid (accurate).
