Objective The first objective was to assess the psychometric properties of the 92-item Dutch Naming Test (DNT-92), developed to assess word finding difficulties in people with aphasia, using Item Response Theory (IRT). The second objective was to select suitable items for a short version with a discriminative purpose. Method This study has a retrospective, psychometric research design, in which 510 DNT-92-forms of people with aphasia and 192 DNT-forms of healthy participants were used for analyses. An IRT analysis was performed and information on the item- and person parameters was obtained. Item selection for the short version was based on a combination of the discriminative ability of the items and their estimated theta or difficulty. Items with the highest information load, and a difficulty parameter in the range of overlap between the sample of people with aphasia and healthy participants were selected. Results A 2-PL IRT analysis showed best fit to the data. Assumptions of unidimensionality, local independence, and monotonicity were met. Items were removed incrementally, whilst checking sensitivity and specificity of the remaining short form. A selection of six items proved optimal in terms of sensitivity and specificity, with an area under the curve value of 0.85. Differences were found between participants younger than 70 and older. Conclusions The IRT assumptions for the DNT-92 were met, indicating that the test has good psychometric properties. A reduction of items to just six items proved possible, leading to a reliable six item short form with a discriminatory purpose.
LINK
To identify and critically appraise the evidence for instruments assessing depression in stroke patients with aphasia.
LINK
Spontaneous speech is an important source of information for aphasia research. It is essential to collect the right amount of data: enough for distinctions in the data to become meaningful, but not so much that the data collection becomes too expensive or places an undue burden on participants. The latter issue is an ethical consideration when working with participants that find speaking difficult, such as speakers with aphasia. So, how much speech data is enough to draw meaningful conclusions? How does the uncertainty around the estimation of model parameters in a predictive model vary as a function of the length of texts used for training?
