A review of compositional data analysis and recent advances

Alenazi, Abdulaziz

doi:10.1080/03610926.2021.2014890

Cited by 7 publications

(1 citation statement)

References 142 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Compositional data analysis (CoDA) is a well-established statistical method that is widely used in research where multiple “components” (variables) are involved [ 39 , 40 ]. The method considers that all “components” of a whole have mutuality where the scale of the component data are constant, i.e., the relative ratio between components is constant, and they are naturally subject to a unit-sum constraint [ 18 , 39 ].…”

Section: Methodsmentioning

confidence: 99%

The Associations between 24-Hour Movement Behaviours and Quality of Life in Preschoolers: A Compositional Analysis of Cross-Sectional Data from 2018–2021

Chen

Chua

Shen

et al. 2022

IJERPH

View full text Add to dashboard Cite

Studies show that the quality of life (QoL) of preschoolers is closely related to physical activity (PA), sedentary behaviour (SB), and sleep (SL). Yet many researchers looked at these 24-h movement behaviours as behaviours that are independent of one another when examining the association of QoL with these behaviours. The main purpose of the present study was to describe the temporal trends in PA, SB, and SL in preschoolers and the concomitant association with QoL of children. Annual cross-sectional data on QoL and 24-h movement behaviours of 8045 Singaporean preschoolers were collected from 2018 to 2021. Compositional analysis, linear regression, and isochronous replacement approaches were used to examine changes in PA, SB, and SL from 2018–2021 and how these changes were associated with QoL. Temporal trends in movement behaviours showed that PA and SL decreased after 2020. During 2018–2021, the association of PA and QoL in preschoolers was characterised by a ‘U’ curve (βPA-2018 = 3.06, p < 0.001; βPA-2019 = 1.43, p < 0.05; βPA-2020 = −0.43, p > 0.05; βPA-2021 = 2.82, p < 0.001), while SL and QoL were characterised by an inverted ‘U’ curve (βSL-2018 = −2.39, p < 0.001; βSL-2019 = −0.27, p > 0.05; βSL-2020 = 2.00, p < 0.01; βSL-2021 = −0.21, p > 0.05). SB was significantly and negatively associated with QoL after 2020, with 2020 identified as the inflection point for the change in SB (βSB-2018 = 0.67, p > 0.05; βSB-2019 = −1.16, p > 0.05; βSB-2020 = −1.56, p < 0.01; βSB-2021 = −2.61, p < 0.01). Using a time re-allocation technique to treat the 2021 data, reallocating time from SB to PA or to SL predicted improvements in QoL of preschoolers (Pall < 0.05). The study provided useful information on the temporal trends in PA, SB, SL, and QoL of preschoolers over four years. Additionally, these data provided insights into how changes in QoL are predicted by changes in duration in the 24-h movement behaviours.

show abstract