The present work is an introduction to Latent Class Growth Modelling (LCGM). LCGM is a semi-parametric statistical technique used to analyze longitudinal data. It is used when the data follows a pattern of change in which both the strength and the direction of the relationship between the independent and dependent variables differ across cases. The analysis identifies distinct subgroups of individuals following a distinct pattern of change over age or time on a variable of interest. The aim of the present tutorial is to introduce readers to LCGM and provide a concrete example of how the analysis can be performed using a real-world data set and the SAS software package with accompanying PROC TRAJ application. The advantages and limitations of this technique are also discussed. Longitudinal data is at the core of research exploring change in various outcomes across a wide range of disciplines. A number of statistical techniques are available for analyzing longitudinal data (see Singer & Willet, 2003).

No abstract

No abstract

No abstract

According to the two-factor theory of perfectionism (Stoeber & Otto, 2006), perfectionism comprises two superordinate dimensions-perfectionistic strivings (PS) and perfectionistic concerns (PC)-that show different, and often opposite, relations with psychological adjustment and maladjustment, particularly when their overlap is partialled out. Recently, Hill (2014) raised concerns about the interpretation of the relations that PS show after partialling. The present article aims to alleviate these concerns. First, we address the concern that partialling changes the conceptual meaning of PS. Second, we explain how the relations of residual PS (i.e., PS with PC partialled out) differ from those of PS, and how to interpret these differences. In this, we also discuss suppressor effects and how mutual suppression affects the relations of both PS and PC with outcomes. Furthermore, we provide recommendations of how to report and interpret findings of analyses partialling out the effects of PS and PC. We conclude that, if properly understood and reported, there is nothing to be concerned about when partialling PS and PC. On the contrary, partialling is essential if we want to understand the shared, unique, combined, and interactive relations of the different dimensions of perfectionism.

The aim of the present study was to verify, during a stressful sport competition, the associations between motivational antecedents and consequences of the coping process. Using a two-wave design, we tested a model that incorporates motivational orientations, coping dimensions, goal attainment, and affective states among athletes (N = 122). Path analyses using EQS revealed that selfdetermination toward sport positively predicted the use of task-oriented coping strategies during a stressful sport competition, while non-self-determined motivation predicted the use of disengagement-oriented coping strategies. Task-oriented coping, in turn, was positively associated with the level of goal attainment experienced in the competition, whereas disengagement-oriented coping was negatively associated with goal attainment. Finally, level of goal attainment was positively linked to an increase in positive emotional states from pre-to postcompetition, and negatively associated with an increase in negative emotional states. Findings are discussed in light of coping frameworks, self-determination theory, and the consequences of motivational and coping processes on psychological functioning.

The objective of the present study was to compare alternative factorial structures of the French-Canadian version of the Positive and Negative Affect Schedule (PANAS; Watson, Clark, & Tellegen, 1988 ) across samples of athletes at different stages of a sport competition. The first sample (N = 305) was used to assess, compare, and improve the measurement model of the PANAS. The second sample (N = 217) was used to cross-validate the model that provided the best fit with the calibration sample. Results of confirmatory factor analyses suggested that a modified three-factor model with cross-loadings provided a better fit to the data than either the hypothesized or the modified two-factor models. This model was partially replicated on the second sample. Results of a multiple-group confirmatory factor analysis have shown that the model was partially invariant across the two samples.

No abstract