Several authors have pointed out that a strictly variable- or person-centred approach, which is most often used in parenting research, is unable to capture the full complexity of child and family psychology. In parenting research, a variable-centred approach has a focus on parenting dimensions; these are the key concepts and most often studied using (continuous) variables. Alternatively, the personcentred approach regards parenting as a complex system that is best understood in terms of whole-system properties and categorises parents into parenting styles. The current longitudinal study combined a person-centred and a variable-centred approach using subspace K-means clustering.
In this study four objectives were addressed. First, the study tried to identify meaningful groups of parents in longitudinal adolescent reports on parenting behaviour. Second, the dimensional structure of every cluster was inspected to uncover differences in parenting between and within clusters. Third, the parenting styles were compared on several adolescent characteristics. Fourth, to examine the impact of change in parenting style over time, we looked at the cluster membership over time.
Our study used subspace k-means cluster analysis, to distinguish clusters based on the five parenting dimensions, in each wave separately. This analysis aims at finding mutually exclusive clusters from subspaces of data instead of the entire data space. This type of analysis simultaneously models the centroids and the within-cluster residuals in subspaces. The location of each centroid is identified via scores on a few between-components, capturing the main differences between the clusters. The within-cluster variability in the observed variables is represented into a few cluster-specific within-cluster components.
We identified two clusters (authoritative and authoritarian parenting) in which parenting dimensions were interrelated differently. Authoritative parenting seemed to be beneficial for adolescent development. Longitudinal data revealed several parenting group trajectories which showed differential relations with adolescent outcomes. Change in membership from the authoritative cluster to the authoritarian cluster was associated with a decrease in self-concept and an increase in externalising problem behaviour, whereas changes from the authoritarian cluster to the authoritative cluster were associated with an increase in self-concept and a decrease in externalising problem behaviour.
The results of our study suggest that it is possible to capture the complex interplay of group membership and dimensional information by combining the information from multiple dimensions into non-arbitrary groups.
Our findings highlight the importance of identifying parenting styles based on adolescents’ perspectives on parenting. Identifcation of parenting styles in adolescence may prevent the risk to develop social and emotional problems such as low self-esteem and externalising problem behaviour. In this study we identifed a group of parents that was categorised as authoritarian in all waves (7.75%). These adolescents and their families might beneft from professional guidance on parent-adolescent relationships (e.g., in family therapy) that infuence the patterns of family functioning. Diferentiated parenting advice (e.g., psycho-education) depending on cluster trajectories might increase intervention effectiveness.Contact: Filip.Calders@kuleuven.be
Link to paper > Contact > Website >