T5: Selection Standards (p.224) ( Reliability (Consistency of test…
T5: Selection Standards (p.224)
Consistency of test results
Is the measure free from random error?
Questions to test or method, not the result.
Need for reliability is dependent on the decision being made and the people being decided between.
'Correlation coefficient' statistical measure can be used to assess reliability in a test-retest model on the same test
Extent of connection between test and job performance
- test relationship to job performance (statistical approach)
Predictive validity study
- empirical relationship ee test results and eventual job performance (superior but time consuming to use this study method)
Concurrent validity study
- compares current employees scores against current performance - then uses this to assess applicants (inferior results to predictive version, should be used in advance of, not instead of)
'Correlation coefficient' can be used to assess the link between the measure and the relevant job aspects
- does test represent actual situations/problems in the job (judgemental approach)
Must determine reliability, then move to validity
Is the test outcome transferable?
To different samples of people?
To different jobs / organisations?
Validity generalisation can be used where other validity measures are unavailable.
How much does the test help the selection process?
Key for managers
Is the method helping find the right person for the job?
The more reliable, valid and generalizable a measure is, the more utility it will have.
How much does this measure support the bottom-line?
Does the method align with the legal requirements?