You need a defensible way to define what your test will even cover. This starts with a job analysis. A committee of subject matter experts (SMEs), will generate a list of tasks in the profession, then we obtain quantitative data from incumbents on which tasks are most frequent and important. This data is invaluable.
We then convert the job analysis results into a blueprint for the exam. You’ve seen those… 16% from Domain 1, 22% from Domain 2, etc. But these numbers aren’t just arbitrary – they will be defined by our data on what tasks or knowledge is most important and frequent for the profession.
A high quality item is the basic building block of any exam. You’ll need to train your writers, have them write items, then put the items through a QA workflow – and be able to document this all for Accreditation.
Your exam needs a cutscore that defines who passes and fails. If you simply pick a round number like 70% with no justification, you are in for a heap of trouble. You’ll need a panel of SME to complete what is known as a modified-Angoff study, or similar approach, to provide a cutscore with extensive quantitative data loaded with SME expertise.
All exam forms need a comprehensive statistical report done to check on the quality of every single item. Moreover, you are required to submit summary reports every single year. This work requires a PhD psychometrician – or software designed to automate one.