Item Response Theory (IRT) has a lot of advantages when comparing it to Classical Test Theory (CTT):
- IRT supports adaptive testing, where the most informative questions are presented based on your previous responses
- IRT supports item banking, which means lower exposure for each question and better coverage of the latent trait being measured
- IRT scoring increases the accuracy of the results, by taking item characteristics into account. This means that fewer questions are needed to get accurate results.
While CTT deals with a fixed number of questions to form a test scale, IRT deals with each question, or item, separately. This makes the process of continuously improving a test simpler - replacing one question doesn't change the entire scale.
For Alva's personality test, a specific IRT model is used called the Graded Response Model (Samejima, 1969; Muraki, 1990). This model provides a mathematical formula for the likelihood of selecting a given response option, given a latent trait of the test taker and characteristics of the question. A modification of the model is specifically developed for Likert-type data, as described by Muraki (1990).
Read more about IRT in this Wikipedia article.
Muraki, E. (1990). Fitting a Polytomous Item Response Model to Likert-Type Data. Applied Psychological Measurement, 14(1), 59-71. doi:10.1177/014662169001400106
Samejima, F. (1969). Estimation of latent ability using a response pattern of graded scores. Psychometrika Monograph Supplement, 34(4), 100.