Thursday May 27th

Use of Item Response Theory (IRT) for the generation of normative data in neuropsychological tests

Diego Rivera, PhD

Pre-Conference Workshop Description

Having normative data is of vital importance to properly interpret a person’s score on a given neuropsychological test. Currently the use of “direct” scores for the interpretation of neuropsychological test results is seldom used. The reason for this is that direct scores are strongly affected by demographic variables (e.g., age, education, and sex) and their interpretation can create important biases when shaping neuropsychological assessment results. Normative neuropsychological testing data are intended to determine participant performance with respect to a reference group and different methods for their estimation, such as those based on regression and IRT estimators, can now be found.

IRT is a more refined approach to the study of the psychometric properties of a test, allowing more realistic modeling of the responses of a test, since it uses the items as a unit of analysis and allows the description of some psychometric properties of the instrument through invariant indicators. Thus, generating normative data of a test or a bank of items based on its estimated θ does not depend on the specific items administered and the value obtained can be compared with that of another test that could have received even a different set of items from the bank.

The objectives of this workshop are

1. To explain the theoretical bases of IRT in the analysis of databases with dichotomous answers.

2. To know the R packages for parameter estimation in IRT.

3. To put into practice the estimation of the test characteristic curve (TCC) and its value that corresponds in that curve to the θ obtained by the evaluator.

4. Estimation of normative data based on multiple regressions from the estimated θ