University of Twente Student Theses

Login

Item response theory model with testlet effects: A simulation study investigation of effectiveness in small sample sizes

Wenzel, J.H.M. (2018) Item response theory model with testlet effects: A simulation study investigation of effectiveness in small sample sizes.

[img] PDF
2MB
Abstract:In this article, two different item response theory (IRT) models with testlet effects are compared. Testlets structures account for item interrcorrelations in sets of items, so that conditional independence for a test can be assumed, even if groups of related items are used in the testconstruction. The conditional testlet model adds a testlet parameter to the basic IRT model, that models the dependencies among response observations to the same testlet items. The marginal testlet model, by contrast, models them directly as a covariance parameter in the covariance matrix of the distribution of the error. This leads to better parameter estimation for small dependencies. The precision of the parameter estimates under the two models are evaluated for data simulated under the conditional testlet model. The simulation study uses different conditions, varying in number of participants, number of items included per testlet and amount of variance in the testlet effects. It will be shown that under the marginal testlet model, parameter estimates can be obtained with the same precision as under the conditional testlet model for large sample sizes. Furthermore, under the marginal testlet model, better parameter estimates are obtained for conditions with small testlet variances than under the conditional testlet model.
Item Type:Essay (Bachelor)
Faculty:BMS: Behavioural, Management and Social Sciences
Subject:77 psychology
Programme:Psychology BSc (56604)
Link to this item:http://purl.utwente.nl/essays/75259
Export this item as:BibTeX
EndNote
HTML Citation
Reference Manager

 

Repository Staff Only: item control page