Bayotas, Pauline Wanda
(2022)
A simulation study of computer-adaptive testing for measuring treatment-related change in confrontation naming.
Undergraduate Thesis, University of Pittsburgh.
(Unpublished)
Abstract
Computer adaptive testing (CAT) is an approach that can be used to shorten assessments without sacrificing their psychometric properties. Recent studies (Fergadiotis, Kellough, & Hula, 2015; Hula, Kellough, & Fergadiotis, 2015; Fergadiotis, Hula, Swiderski, Lei, and Kellough, 2019; Hula, Fergadiotis, Swiderski, Silkes, & Kellough, 2020) produced a CAT with an item bank consisting of the Philadelphia Naming Test (PNT; Roach et al., 1996). The main advantage of CAT is to maximize the precision of the test, requiring fewer testing items while having the same or better level of accuracy as traditional brief naming assessments. However, before a CAT can be used to measure change in anomia severity, it is important to understand how the algorithm interacts with commonly used aphasia interventions and whether it is as responsive to treatment-related change as standard static assessments. This simulation study investigated the sensitivity of a computer adaptive version of the Philadelphia Naming Test (PNT-CAT) to treatment-related change in three different treatment conditions: item-general, item-specific and partially item-specific. For each condition, we simulated responses using a one-parameter logistic item response theory model and computed pre- to post-treatment change scores for the PNT-CAT and the full PNT. For the item-general condition, both tests performed similarly well. However, the PNT-CAT overestimated the effects of item-specific and the partially item-specific treatment relative to the full test. These results provide useful information about the conditions in which CAT can be validly used to measure treatment outcomes. Present results suggest that when treatment affects underlying naming ability the PNT-CAT30 is appropriately responsive to treatment and provides an efficient alternative to the administration to the full PNT. On the other hand, when treatment effects are item-specific, the PNT-CAT30 may overestimate or underestimate treatment effects depending on the baseline ability level and the number of treated items included in post-treatment CAT.
Share
| Citation/Export: |
|
| Social Networking: |
|
Details
Metrics
Monthly Views for the past 3 years
Plum Analytics
Actions (login required)
 |
View Item |
![[feed]](/images/feed-icon-32x32.png){kind=icon}