About
How do we know how reliable our measure are.
Raw scores (X) are not perfect. (Classical test|True score) theory asserts that raw scores are influenced by:
- bias,
- chance error,
- and true scores
A Raw score is considered to be reliable as it approaches the true score.
Articles Related
Method
Test Retest and Parallel are really hard to do because they are time consuming. Inter-item is the most commonly used method in the social sciences.
Test Retest
Just measure twice. It should be a strong correlation. The correlation between X1 and X2 is an estimate of reliability
It wouldn't detect the bias.
Parallel test
Just take two different instruments and seeing if those are correlated. The correlation between X1 and X2 is an estimate of reliability.
The bias must be detected.
Inter-item
Reliability in one measurement, in one time point but with two set of measurement (questions for instance) that must be correlated.
The correlation between X1 and X2 is an estimate of reliability.