These findings must be interpreted in light of several limitations. First, the cross-sectional nature of these data precludes statements about causality or temporality in the relationship between social environmental factors and latent class membership. Second, all data were assessed by adolescent self-report and are thus subject to social desirability bias, although the use of computer-assisted self-interview in many of the countries may have helped to mitigate this issue . Third, with the exception of childhood adversity and neighborhood cohesion, the measurement of risk and protective factors relied on single dichotomized items rather than validated scales. It is possible that this measurement limitation led to spurious conclusions about the role of certain factors. For instance, while we found that peer socialization acted as a risk rather than a protective factor in several countries, given its operationalization as time spent with friends outside of school, it is plausible that this covariate captured unsupervised social activities rather than peer connectedness. Indeed, prior research among youth in the United States has found correlations between increased peer activity in the evening and a range of behavioral problems . Finally, it has been suggested that multiple imputation may be inappropriate in a latent class analytic framework due to the theoretical incompatibility between multiple imputation, which assumes a single underlying population, and LCA, which assumes multiple latent subgroups within a population .
These limitations notwithstanding, the current study has several important strengths, such as its inclusion of early adolescents from four LMICs, its use of a person-centered analytic approach to examine co-occurring psychosocial challenges, and its simultaneous examination of risk and protective factors across multiple social environmental domains. Across countries, we found a number of factors which were consistently and robustly associated with emotional and behavioral problems, including childhood adversity, peer bullying behaviors, and a perceived lack of school safety. This consistency is suggestive of the generalizability of risk factors in early adolescence, and indicates that interventions targeting psychosocial adjustment among this age group may have applicability in diverse cross-national settings. In addition, the patterns of association across latent classes point to especially heightened vulnerability among a subgroup of adolescents with co-occurring emotional and behavioral problems. Given resource limitations in many LMICs, this information can be used to guide decision-making around which adolescent populations to prioritize through interventions.
Shoshanna L. Fine, Rashelle J. Musci, Judith K. Bass, Effie Chipeta, Eric M. Mafuta, Anggriyani W. Pinandari, Siswanto A. Wilopo, Xiayun Zuo, Robert W. Blum,
A Multi-Country Study of Risk and Protective Factors for Emotional and Behavioral Problems Among Early Adolescents, Journal of Adolescent Health,2022,ISSN 1054-139X, https://doi.org/10.1016/j.jadohealth.2022.05.002.