inst/app/www/intro.md

In rbcavanaugh/pnt: Computer adaptive forms of the Philadelphia Naming Test

THIS IS JUST A ROUGH DRAFT OF WHAT MIGHT GO HERE

Welcome to the computure adapative testing form of the Philadelphia Naming Test (PNT-CAT). All versions of the PNT-CAT were developed within a Item Response Theory (IRT) Framework. The advantage of calibrating an assessment within the IRT framework is functionality to estimate a participants naminig ability based off the items difficulty.

The intention of this application is to provide users with the ability to deliver a lesser amount of the total items of the PNT (175 pictures), without sacrificing the ability to estimate naming ability of people with aphasia.

The assessment was developed with a 1 Pararmeter logistic model (see: Fergadiotis, Casilio, Hula, and Swiderski, 2021 and https://aswiderski.shinyapps.io/IRTapp/). In breif, the 1PL model predicts the probablity of a correct response of an item by the examinees ability. The computer adaptive tests leverage the properties of this psychometric model by delivering items that are only proximal to the participants ability. The PNT-CATs do not, deliver items that are either too easy or too dificult for the examinee.

This application permits administration of the computer-adaptive PNT using a 30-, 60-, and 100-items. A full length PNT is also variable in the original order as published by (Roach et al., 1996) and in a random order. Support for the use of the 30 item PNT-CAT (PNT-CAT30) was provided by Fergadiotis, Hula, Swiderski, Lei, and Kellough (2018) who demonstrated that naming ability estimates derrived from the PNT-CAT30 were highly correlated with naming ability estimates derrived from the full PNT (r = 0.95). Continuing the PNT-CAT30 to either 60 or 100 items may provide more measurement percision of naming ability. Increasing the number of items delivered will also allow the clinician to observe more error types to aid in the understanding the locus of the naming impairment.

Two additional testing instruments are available. The Variable length PNT-CAT (PNT-CATVL) can be utilized to assess naming ability after a version of the PNT-CAT has been given, but will exclude overlapping items. Using the PNT-CATVL Hula and colleagues (2019). Hula and his colleagues demonstrated naming ability estimates from 25 persons with aphasia estimated with the PNT-CAT30 and then again with a non-overlapping PNT-CATVL were highly correlated (r = 0.89). These results suggest that the PNT0CATVL may be utilized to assess the severity of anomia of people with aphasia with an alternate subset of items not delivered in the PNT-CAT30. At the administred assessment, users have the option of downloading performance data and a pdf-report of the results.

In order to exclude items delivered from the PNT-CAT30 or PNT-CAT60 in the PNT-CATVL you must provide the application with a .csv file generated from the initial test administration. To complete this step, upload this file on the following page. Then, select the Variable length PNT-CAT option. At the end of the test, users have the option of downloading performance data and a pdf-report of the results.

In the cases where the PNT was either adminstered or scored offline, results can be uploaded to obtain IRT-based ability estimates. This application provides a simple user interface to estimate these parameters. Simply click the score-offline test button below. Users should also use this option to rescore assessments where responses were coded erroneously be the examiner.

Please note, that the intended population for this assessment are individuals with acquired language impairments and the testing instrument may be unreliable for other populations.The app does not save any data beyond a single user session.

To minimize server costs, the app will time-out after 15 minutes of non-use and all data will be deleted.

Special thanks to...

rbcavanaugh/pnt documentation built on Oct. 20, 2024, 7:21 p.m.