and classify the standing movement, potentially facilitating rapid analysis of motion capture data with minimal visual inspection. We propose the use of hierarchical clustering, to categorize sit-to-walk strategies. Cluster analysis is a statistical technique used to identify structure in a series of
Dimitrios-Sokratis Komaris, Cheral Govind, Andrew Murphy, Alistair Ewen and Philip Riches
Aaron England, Timothy Brusseau, Ryan Burns, Dirk Koester, Maria Newton, Matthew Thiese and Benjamin Chase
computed for each decision tree, Z transformed, and combined into a Z matrix. To form the basis of a hierarchical cluster analysis, the Z matrix was transferred into a Euclidean distance matrix. The resulting Euclidean distance matrix resulted in an individual cluster solution on the 12 concepts and formed
Carolina F. Wilke, Felipe Augusto P. Fernandes, Flávio Vinícius C. Martins, Anísio M. Lacerda, Fabio Y. Nakamura, Samuel P. Wanner and Rob Duffield
hierarchical cluster analysis based on Euclidean distance and average linkage criteria was performed (Python 2.7; Python Software Foundation, https://www.python.org/ ). Briefly, each subject’s data for each measure are plotted in a multidimensional plan, and the Euclidean distance between subjects is
Bo Shen and Ang Chen
Using the model of domain learning as a theoretical framework, the study was designed to examine the extent to which learners’ initial learning profiles based on previously acquired knowledge, learning strategy application, and interest-based motivation were distinctive in learning softball. Participants were 177 sixth-graders from three middle schools. A hierarchical cluster analysis was conducted to determine what kinds of learning profiles would result from the interactions among prior knowledge, learning strategies, and interest. The results revealed that individual learners could be classified into subgroups with distinctive learning characteristics. It is supported that learning in physical education is a progressive process that involves both cognitive and affective dimensions. An effective physical education curriculum should address both knowledge and skill acquisition and motivation simultaneously.
Bo Shen, Nate McCaughtry, Jeffrey J. Martin and Mariane Fahlman
With the belief that theoretical integration in motivation may help us better understand motivational behavior, we designed this study to explore adolescents’ motivational profiles and their associations with knowledge acquisition, leisure-time exercise behaviors, and cardiorespiratory fitness. Middle school students from a large urban inner-city school district (N = 603, ages 12–14) completed questionnaires assessing motivational constructs and leisure-time exercise behavior. Knowledge and cardiorespiratory fitness were also assessed with a knowledge test and the Progressive Aerobic Cardiovascular Endurance Run (PACER) test, respectively. Using hierarchical cluster analysis, we found that students’ motivation in physical education could be explained from a multi-theoretical perspective. The interactive patterns among different motivation constructs were homogeneous overall and associated with in-class effort, knowledge, and leisure-time exercise behavior. These findings suggest that students’ development in physical education may depend upon a collective impact of changes in knowledge, physical activity ability, and sources of motivation.
Kieran Dowd, Deidre Harrington, Ailish Hannigan, Helen Purtill, Sarah M. Kelly, Alan P. Macken, Niall Moyna, Clodagh S. O’Gorman and Alan E. Donnelly
This study aims to (1) use the objective activPAL activity monitor to assess physical activity behaviors, including sitting/lying, standing, and both light (LIPA) and moderate-to-vigorous physical activity (MVPA); (2) to develop distinct activity profiles based on time spent in each behavior in a sample of adolescent females; and (3) examine whether levels of adiposity differ across these activity profiles.
Female adolescents (n = 195; 14–18 y) had body mass index (median = 21.7 [IQR = 5.2] kg/m2) and 4-site skinfold thickness (median 62.0 mm; IQR = 37.1) measured. Physical activity behaviors were measured using the activPAL. Hierarchical cluster analysis grouped participants into activity profiles based on similar physical activity characteristics. Linear mixed models explored differences in body composition across activity profiles.
Three activity profiles were identified, a low (n = 35), moderate (n = 110), and a high activity profile (n = 50). Significant differences across activity profiles were observed for skinfold thickness (p = .046), with higher values observed in the low activity profile compared with the high activity profile.
Profiling free-living activity using behaviors from across the activity intensity continuum may account for more of the variability in energy expenditure then examining specific activity intensities, such as MVPA alone. The use of activity profiles may enable the identification of individuals with unhealthy activity behaviors, leading to the development and implementation of more targeted interventions.
Stephen Zwolinsky, James McKenna, Andy Pringle, Paul Widdop, Claire Griffiths, Michelle Mellis, Zoe Rutherford and Peter Collins
Increasingly the health impacts of physical inactivity are being distinguished from those of sedentary behavior. Nevertheless, deleterious health prognoses occur when these behaviors combine, making it a Public Health priority to establish the numbers and salient identifying factors of people who live with this injurious combination.
Using an observational between-subjects design, a nonprobability sample of 22,836 participants provided data on total daily activity. A 2-step hierarchical cluster analysis identified the optimal number of clusters and the subset of distinguishing variables. Univariate analyses assessed significant cluster differences.
High levels of sitting clustered with low physical activity. The Ambulatory & Active cluster (n = 6254) sat for 2.5 to 5 h·d−1 and were highly active. They were significantly younger, included a greater proportion of males and reported low Indices of Multiple Deprivation compared with other clusters. Conversely, the Sedentary & Low Active cluster (n = 6286) achieved ≤60 MET·min·wk−1 of physical activity and sat for ≥8 h·d−1. They were the oldest cluster, housed the largest proportion of females and reported moderate Indices of Multiple Deprivation.
Public Health systems may benefit from developing policy and interventions that do more to limit sedentary behavior and encourage light intensity activity in its place.
Alexandre Magalhães, Milton Severo, Roseanne Autran, Joana Araújo, Paula Santos, Maria Fátima Pina and Elisabete Ramos
We aimed to assess the validity of a single question to evaluate leisure-time physical activity (PA) in adolescents. We included 209 participants (57.4% girls) aged 14–18 years from Porto, Portugal, evaluated as part of the SALTA project. A self-reported question with four answer options, designed for the EPITeen study, was used to classify the intensity level of usual leisure-time activities. Actigraph accelerometers were used to objectively measure total PA during 7 consecutive days. Since the accelerometers measured PA as a continuous variable, hierarchical cluster analysis was used to identify clusters of individuals with similar level of objectively measured PA. Correlations between self-reported and objective measures were evaluated through polychoric correlations. In girls, we found higher mean time on sedentary activities among those describing their leisure-time PA as “sitting”, and an increase on the time spent on light and moderate activities with increasing intensity of PA on self-reported classification. A similar trend was found in boys, but not reaching statistical significance. The correlation between the two measures of PA was 0.42 for girls and 0.46 for boys. We found an acceptable correlation between our single question and the objectively measured PA, showing that, although the single question is not adequate to quantify the intensity of the physical activity, it allows to rank adolescents according to leisure-time physical activity.
Jing Dong Liu and Pak-Kwong Chung
students who displayed similar motivational profiles ( Hair et al., 2010 ). Prior to cluster analysis, all clustering variables were transformed into z-scores. In Step 1, hierarchical cluster analysis using Ward’s method ( Ward, 1963 ) with squared Euclidean distance was conducted to explore the number of
Rafael Burgueño, José Macarro-Moreno, Isabel Sánchez-Gallardo, María-Jesús Lirola and Jesús Medina-Casaubón
proposed by Hodge and Petlichkoff ( 2000 ), z scores below −.50 are classified as low, z scores between −.50 and .50 as moderate, and z scores above .50 as high. In the first step, a hierarchical cluster analysis was conducted to identify the number of motivational profiles of the students. For this