Dispersion measures in biomedical research on ageing: nuances in the meaning of variability

gerontology. In the present paper w e report the level of attention paid to dispersion measures in recently published geriatric studies. We also studied whether sufficient data w ere reported to test the generally accepted age-related increase in variability between subjects in the biomedical outcom e parameters rt w* tion and spread. 'Heterogeneity', 'diversity', 'variation1 and 'spread1 will not be used in the present report because they are often used in everyday language, and in other contexts. In this study 'dispersion' refers to the distribution width of a parameter; 'dispersion measures' (DMs) are defined as all possible measures that can be used to quantify the width of a distribution. For numerical data, the most important DMs are: standard deviations (SDs)— and variances; coefficients of variation (CVs); ranges; and percentiles. The term 'variability' can be used unequivocally in statistics only as a part of a compound. It can be defined as differences betw een measurements carried out at the same time or change in measurements with time. In biomedical research, variability can generally be divided into a biological, an analytical (measurement error) and a temporal com ponent, Intra-individual variability, betweetvgroup variability and within-group variability (WGV) are essentially different. In this study,