Use this URL to cite or link to this record in EThOS:
Title: Non-parametric item response theory applications in the assessment of dementia
Author: McGrory, Sarah
ISNI:       0000 0004 5923 7248
Awarding Body: University of Edinburgh
Current Institution: University of Edinburgh
Date of Award: 2015
Availability of Full Text:
Access from EThOS:
Full text unavailable from EThOS. Please try the link below.
Access from Institution:
This thesis sought to address the application of non-parametric item response theory (NIRT) to cognitive and functional assessment in dementia. Performance on psychometric tests is key to diagnosis and monitoring of dementia. NIRT can be used to improve the psychometric properties of tests used in dementia assessment in multiple ways: confirming an underlying unidimensional structure, establishing formal item hierarchical patterns of decline, increasing insight by examining item parameters such as difficulty and discrimination, and creating shorter tests. From a NIRT approach item difficulty refers to the ease with which an item is endorsed. Discrimination is an index of how well an item can differentiate between patients of varying levels of severity. Firstly I carried out a systematic review to identify applications of both parametric and non-parametric IRT to measures assessing global cognitive functioning in people with dementia. This review demonstrated that IRT can increase the interpretive power of cognitive assessment scales and confirmed the limited number of IRT analyses of cognitive scales in dementia populations. This thesis extended this approach by applying Mokken scaling analysis to commonly used measures of current cognitive ability (Addenbrooke’s Cognitive Examination-Revised (ACE-R)) and of premorbid cognitive ability (National Adult Reading Test (NART)). Differential item functioning (DIF) by diagnosis identified slight variations in the patterns of hierarchical decline in the ACE-R. These disease-specific sequences of decline could serve as an adjunct to diagnosis, for example where learning a name and address is a more difficult task than being orientated in time, late onset Alzheimer’s disease is a more probable diagnosis than mixed Alzheimer’s and vascular dementia. These analyses also allowed key items to be identified which can be used to create briefer scales (mini-ACE and Mini-NART) which have good psychometric properties. These scales are clinically relevant, comprising highly discriminatory, invariantly ordered items. They also allow sensitive measurement and adaptive testing and can reduce test administration time and patient stress. Impairment of functional abilities represents a crucial component of dementia diagnosis with performance on these functional tasks predictive of overall disease. A second aspect of this thesis, therefore, was the application of Mokken scaling analyses to measures of functional decline in dementia, specifically the Lawton Instrumental Activities of Daily Living (IADL) scale and Physical Self-Maintenance Scale (PSMS). While gender DIF was observed for several items, implying the likelihood of equal responses from men and women is not equal a generally consistent pattern of impairment in functional ability was observed across different types of dementia.
Supervisor: Austin, Elizabeth ; Shenkin, Susan ; Starr, John Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID:  DOI: Not available
Keywords: dementia ; item response theory ; Mokken scaling ; hierarchy