Revelian is continuously looking for ways to improve our products and services to ensure our clients receive best of breed psychometric assessments and technology platforms.
In 2012 Revelian’s Research and Development team, consisting of Organisational Psychologists and Software Engineers, embarked on a project that would see a significant change in how the Revelian Cognitive Ability Test (RCAT) and our separate abilities tests for verbal, numerical an abstract reasoning are delivered. Specifically, this involved moving from fixed-form test delivery to Linear on the Fly (LOFT) to improve the candidate experience, psychometric properties, and test security.
LOFT is a well known and respected method of assessment item delivery often used in high stakes settings including education and recruitment, where test security, validity and fairness are of critical importance. Under the LOFT model, each test taker receives a unique set of items that are equivalent in terms of content variation and difficulty.
Considered to be a significant improvement to fixed-form delivery where candidates are streamed one of several predetermined parallel forms, LOFT ensures that no two candidates receive the same set of items. This heightens the security of the test and ensures that candidates do not gain from sharing item content and responses.
When a candidate logs in to begin the RCAT, 51 items are selected from a large item bank and streamed to the candidate. Both item type and item difficulty are controlled from candidate to candidate.
The items are content balanced such that each individual receives 17 of each item type (verbal, numerical, abstract) with the order of item type being randomised. That is, the first item streamed to a candidate will be a random selection of either a verbal, numerical or abstract item. The second item will be randomly selected from the two remaining item types, with the third item therefore representing the item type not yet selected. Having now established the ordering of these 3 general item types, this same systematic ordering is retained throughout the assessment.
Each item is randomly selected from carefully constructed item groupings to ensure difficulty levels are controlled and equivalent across candidates. Similar to the fixed-form approach, item groupings are generally ordered from easiest to most difficult so that the RCAT becomes more difficult as candidates progress through the assessment.
The sophisticated controls described above ensure that while each candidate experience is unique in terms of the specific items streamed, equivalence in content and difficulty are maintained so that appropriate performance comparison across candidates is possible.
This LOFT model of item streaming is presented graphically in the figure below.