This symposium will present and critically discuss common yet consequential methodological challenges in the study of cognitive ageing and dementia. It will feature three complementary presentations addressing: (i) clinical prediction modelling, (ii) the use of age- and education-adjusted norms in neuropsychological assessment, and (iii) bias arising from missing data.
First, we will examine challenges in clinical prediction modelling. Although new models are developed on a regular basis, few are externally replicated or rigorously validated, and even when validation occurs, their clinical utility often remains unclear. We will discuss how decision-curve analysis—an intuitive yet underutilised method for evaluating the net benefit of prediction models across a range of risk thresholds—can help overcome current limitations and improve translation to clinical practice.
Second, we will address the ongoing debate surrounding the adjustment of cognitive norms for age and education. While such adjustments may reduce false-positive dementia diagnoses among older individuals and those with lower educational attainment—thereby limiting misdiagnosis and potential stigmatisation—they may also reduce the discriminative ability of cognitive tests and raise treatment thresholds for individuals already at increased risk of healthcare marginalisation.
Finally, the symposium will consider fundamental issues related to missing data. We will discuss the rationale for using multiple imputation and address unresolved methodological questions, including the selection of appropriate auxiliary variables and the choice of imputation algorithms.
Together, these presentations aim to foster critical reflection and methodological rigor in cognitive aging and dementia research.
