Psychometric Test Reliability
Rob Williams Assessment Ltd has considerable experience in assessing the internal reliability of psychometric tests – a key stage in any psychometric test design. Many of the UK’s and US’s leading test publishers have used Rob Williams Assessment’s to calculate and prove the reliability of their newly developed psychometric tests.
There are two main types of reliability study: internal reliability and test re-retest reliability. Both represent best practice in psychometric design. Hence both reliability measures are key stages of a psychometric test development project. The statistical internal reliability tests are conducted during the item selection phase of psychometric test development (post-trial). Tests of re-retest reliability are typically produced as one of the last stages, or even subsequent to a version of the test having a limited release.
Internal reliability is typically measured by Cronbach’s Alpha as part of the trial item analysis. The internal reliability value is maximised as part of the trial item selection and deselection process. A widely accepted “cutoff” value of 0.7 is deemed acceptable for aptitude tests / ability tests. A higher internal reliability value of 0.8 is considered acceptable for personality scales. There is academic debate about whether situational judgement tests can be expected to have internal reliability. Regardless of this Rob Williams Assessment ltd has developed SJTs with Cronbach’s Alpha values in excess of 0.8.
Test-retest reliability is also needed as one of the final stages of test development. The final item set is administered twice over a three to two month period to the same group in excess of 150+ respondents.