I am struggling to interpret the location parameters in generalised partial credit models. Say you have location parameters $a_1$,$a_2$ and $a_3$. My professor said that for the item to be accepted then they must be monotonic i.e. $a_1$<$a_2$<$a_3$ which I think has to do with the monotonicity assumption of such models however I am not sure.


1 Answer 1


Your understanding is on the right track. In the context of the Generalized Partial Credit Model (GPCM), a type of Item Response Theory (IRT) model, the location parameters $\alpha_1, \alpha_2, \alpha_3$ are important for interpreting item characteristics.

Here's a brief explanation:

Location Parameters: In the GPCM, these parameters represent the difficulty of achieving a certain score level on an item. For instance, $\alpha_1$ might represent the difficulty of moving from a score of 0 to 1, $\alpha_2 $from 1 to 2, and so on.

Monotonicity: The requirement that $\alpha_1 < \alpha_2 < \alpha_3$ is indeed related to the monotonicity assumption. This assumption states that as a person's ability increases, their probability of achieving a higher score on the item also increases. In practical terms, it wouldn't make sense for an item to be easier to get a score of 2 than a score of 1, which is why these parameters need to be ordered (monotonic).

In summary, your professor's emphasis on the monotonic nature of these parameters is well-founded as it relates to the logical progression of difficulty within an item and the fundamental assumptions of the GPCM. For more information on the assumptions of IRT models, I highly recommend Thissen & Steinberg (1986).


Thissen, D., & Steinberg, L. (1986). A taxonomy of item response models. Psychometrika, 51(4), 567-577.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.