Feelings of loneliness in children have been a major topic of interest in clinical and developmental psychology [11]. The reason is that the experience of loneliness is a sign of maladaptive functioning [5]; considerable research has revealed that loneliness is related to numerous problems of clinical relevance and risks including depression, anxiety, social withdrawal and suicide ideation [e.g., 4;9;13;28]. Additionally, lonely individuals tend to suffer from a cognitive bias toward threatening social stimuli, and easy access to the experience of loneliness [33]. According to van Dulmen and Goossens [36], 3-14% of children or adolescents were in a stable high or chronically high trajectory. Therefore, research involving its predictors is significant in terms of shedding light on how to prevent loneliness in children. The primary interest of the present paper was to find interpersonal predictors of loneliness in late elementary school-aged children through a combination of variable- and person- centered approaches.
Predictors of Loneliness in Children
Loneliness is a subjective experience and an emotional tendency described as sadness or pain, which is caused by an absence of connection with others [26]. As abundant research to date has focused on interpersonal experiences as predictors of loneliness, the primary trigger in children of loneliness is thought to be a lack of or difficulties in peer relations [1;2]. Under this assumption, which provides a rational for research regarding predictors of loneliness, positive peer relations (such as peer acceptance and social competence) and peer victimization are valid predictors of high loneliness. For example, Ladd, Kochenderfer, and Coleman [18] reported that the proportion of variance accounted by positive peer relations was of an above medium effect size in both cross-sectional and longitudinal studies and that victimization also predicted loneliness after half a year and changes in loneliness. Ladd and Troop-Gordon [19] conducted a five-year longitudinal study and found that chronic friendlessness and rejection accounted for a proportion of the variance in loneliness. Further, other studies using sociometric methods to obtain group members’ ratings or nominations of liked or disliked others found that victimized children (i.e., those rejected) reported high loneliness compared to accepted or liked children [e.g., 6;25].
As a potential interpersonal predictor of loneliness, the predictive role of aggression remains unclear. Prinstein, Boergers, and Vernberg’s [27] cross-sectional study showed that while victims of relational aggression are at risk of experiencing loneliness, there was no overall relation between aggression and loneliness. As recent research has suggested that aggression consists of two types [7]: overt aggression (e.g., punching, kicking, threatening) and relational aggression (e.g., excluding, manipulating others’ relationships, spreading rumors), we focused on the relation between two types of aggression and loneliness.
In sum, as reviewed, loneliness in children is deeply related to interpersonal experience. Although many of other potential antecedents have been explored such as friendship quality, social network, and individual characteristics [8], the present study focused on the role of interpersonal predictors that have been a focus in considerable research as reviewed above: positive peer relations, victimization, and the two types of aggression (overt and relational aggression).
Person-centered Approach
While variable-centered approaches have dominated this research area, incorporation of person-centered approaches for interpretation has been recommended in the developmental psychopathology field [33]. The person-centered approaches focus on the identification of groups that share specific attributes or individual differences in developmental patterns [21]. As evident in Kochenderfer-Ladd and Wardrop’s [15] study on victimization trajectories, one of the strengths of person-centered approaches was to provide more detailed insights into a relation. Besides adopting a variable-centered view, whereby they found a clear connection between victimization and loneliness, they took a person-centered view by focusing on trajectory patterns. This latter view illustrated that while children moving toward victimization status tended to experience greater loneliness, they did not necessarily become less lonely when moving away from victimization status.
In the first place, previous research on children’s loneliness focusing on predictors seems to have relied on a variable-centered approach; furthermore, researchers have not given much attention to the potential over-interpretation of covariates. For example, when a negative longitudinal relation is found between an independent variable (i.e., potential predictors) and loneliness, researchers can interpret the relation as indicating both that a lower score on the independent variable is predictive of increased feelings of loneliness and that a higher score on the independent variable is predictive of decreased feelings of loneliness. This interpretation, however, is not always applicable because a mere correlation does not necessarily mean that both conditions are simultaneously satisfied; that is, only one condition is well established and then the negative relation emerges. Therefore, in addressing the issue of potential over-interpretation of covariates, person-centered views are useful.
To investigate the predictors of loneliness from person-centered approach, mover-stayer latent trajectory analysis (mover-stayer LTA) is applicable. The mover-stayer LTA explores transitional patterns between latent profiles across time; it differs from typical LTA in being able to identify participants whose latent profiles change (i.e., movers) or remain the same (i.e., stayers) during the research period. Additionally, we can confirm whether the results of a variable-centered approach are in the same direction by applying mover-stayer LTA via comparison of the scores of potential predictors at specific time points. For example, if positive peer relations are strongly related to reduced loneliness (variable-centered view), individuals with latent profiles characterized by decreasing loneliness should have higher scores for positive peer relations during the period of the assessments (person-centered view).
The other advantage of mover-stayer LTA is that this method widely covers potential patterns of trajectories. Here, we reviewed seven empirical research that identified loneliness trajectories in children [3;10;12;17;29;32;37], revealing specific subpopulation of trajectory patterns such as low levels of loneliness, decreasing, increasing, and chronically high patterns. All seven studies reported a trajectory of individuals with decreasing loneliness. However, sub-groups of loneliness trajectories were not consistent among the previous studies; for example, although five of them reported increasing trajectory, Harris et al. [10] and Ladd and Ettekal. [17] did not report the increasing pattern.
Overview of the Present Study
The purpose of the present study was to identify predictors of loneliness through variable- and person-centered approaches. First, using hierarchical linear modeling (HLM) as the variable-centered approach, we investigated the effect of positive peer relations, victimization, and the two types of aggression on loneliness. Second, using mover-stayer LTA as the person-centered approach, we tested the replication of the results by HLM. The hypothesis regarding predictors of loneliness in HLM was that victimization and relational aggression would be positively correlated, and positive peer relations would be negatively correlated with loneliness. In addition, we expected that the results of the mover-stayer LTA would support the direction of results of the HLM.