Accurate assessment of cognitive ageing is critical to improve healthcare for older adults with cognitive decline. While bio-physiological assessments, such as laboratory, physical, and neurological examinations, offer high precision and interval-level scaling, they are limited by not reflecting perceptual experiences of individuals. In other words, it is impossible to find out what an individual is experiencing without asking questions, which requires psychometric measures, often using ordinal level scales. Psychometric measures are widely used to assess, diagnose, and research cognitive ageing but their reliability and validity may bias the results. Firstly, differentiating between enduring and dynamic patterns captured by a measure helps to distinguish an enduring condition while monitoring changes over time. This type of differentiation is also important to determine the reliability and validity of ordinal scales focused on enduring or dynamic patterns. Secondly, all measures should be ideally at least interval-level to meet the basic assumptions of parametric statistics for comparative analyses with interval level variables such as neuroimaging data. Thirdly, the cognitive components assessed using psychometric measures of cognitive ageing interact with each other, and this may impact on the validity of the outcomes measured at the same time, as well as the ability to predict the future outcomes of cognitive ageing. Together, these issues challenge the reliability and validity of the current psychometric measures of cognitive ageing. The present thesis addresses these reliability and validity issues by using three modern theories of measurement, namely, Generalisability theory (G-Theory), Rasch analysis, and Network analysis. The first part of this thesis focuses on the application of G-Theory to examine dynamic and enduring aspects of cognitive ageing as measured by the Memory Assessment Clinic Questionnaire (MAC-Q), the 16-item Informant Questionnaire on Cognitive Decline in the Elderly (IQCODE-16), the Mini Mental State Examination (MMSE), and the Modified Telephone Interview for Cognitive Status (TICS-M). The findings of these studies highlighted the importance of differentiating between dynamic and enduring aspects of cognition and revealed the specific benefits of each measure in assessing cognitive ageing. The second part of the thesis included three studies that applied Rasch methodology to the IQCODE-16 and the TICS-M. These studies generated algorithms that allow the transformation of ordinal scales scores into interval-level data, which improved the precision of the measures. The final two studies used network analysis to evaluate the validity of the cognitive domains of TICS-M and MMSE in relation to neuropsychological test performance, advancing the understanding of global network mechanisms involved in assessments of cognitive ageing and relations between cognitive functions. Overall, this thesis involved analyses of three longitudinal datasets, totalling 2705 participants. The reported findings enhanced reliability and validity of widely used assessment tools for older adults and contributed to better understanding of cognitive ageing processes.