The Longitudinal Interplay of Adolescents' Self-Esteem and Body Image: A Conditional Autoregressive Latent Trajectory Analysis

Abstract

Self-esteem and body image are central to coping successfully with the developmental challenges of adolescence. However, the current knowledge surrounding self-esteem and body image is fraught with controversy. This study attempts to clarify some of them by addressing three questions: (1) Are the intraindividual developmental trajectories of self-esteem and body image stable across adolescence? (2) What is the direction of the relations between body image and self-esteem over time? (3) What is the role of gender, ethnicity, and pubertal development on those trajectories? This study relies on Autoregressive Latent Trajectory analyses based on data from a 4-year, 6-wave, prospective longitudinal study of 1,001 adolescents. Self-esteem and body image levels remained high and stable over time, although body image levels also tended to increase slightly. The results show that levels of self-esteem were positively influenced by levels of body image. However, these effects remained small and most of the observed associations were cross-sectional. Finally, the effects of pubertal development on body image and self-esteem levels were mostly limited to non-Caucasian females who appeared to benefit from more advanced pubertal development. Conversely, Caucasian females presented the lowest self-esteem and body image levels of all, although for them more advanced pubertal development levels were associated with a slight rise in body image over time.

Notes

¹This study uses data collected in Quebec (Canada). In Quebec, children start elementary school around the age of 6 and usually remain in the same school until Grade 6, after which they transition to secondary schools (close to the age of 12), where they remain 5 years (Grades 7 to 11). Quebec secondary schools thus combine North American middle, junior high, and high schools.

²Given the sample size dependency of Δχ², some suggest that changes in fit indices should also be considered in nested model comparisons (e.g., Cheung & Rensvold, 2002). In the present study, these additional verifications did not change the conclusions from the Δχ².

³Models were estimated with manifest variable indicators (mean scale scores) to avoid unnecessary complexity. Still, longitudinal models based on manifest indicators may present problems because they rely on an often untested assumption of measurement invariance and may confound unstable reliability with stability/instability of the construct (Marsh, Muthén et al., 2009). In addition, in ALTs, the autoregressions are estimated on the time-specific uniquenesses of each process, which in manifest variables combines measurement errors (that are partialled out in latent models) and state deviations. This could lead to an underestimation of the autoregressive parameters. Fortunately, our decision to rely on manifest indicators did not affect the results because (a) we found evidence of longitudinal measurement invariance for self-esteem and body image and (b) key fully latent models were estimated and yielded highly similar results, confirming the absence of bias in the reported results.

⁴All models were estimated while ignoring the clustering of students within schools. This did not affect the results because (a) the estimated intraclass correlations coefficients on the study variables were all very low (0.004 to 0.067; M = 0,023, SD = 0.018); (b) key models were estimated while considering this clustering with Mplus “Type = Complex” feature, a method that has been shown to be as effective as full multilevel models (Marsh, Lüdtke et al., 2009; Marsh & O'Mara, 2010) and converged on highly similar results, although these models were never fully proper (negative variance estimates, warnings, etc.) potentially due to the low number of Level 2 units (n = 5 schools).

⁵In LCMs and ALTs, only linear trajectories were estimated (intercepts and slopes). The exclusion of quadratic trends is based on substantive and statistical reasons. Substantively, three of the preceding LCM studies estimated quadratic trends and found that they did not significantly contribute to the models (Greene & Way, 2005), were too small to be meaningful (Greene, Way, & Pahl, 2006), or were significant only in specific subgroups (Moneta et al., 2001). In this study, model comparisons of preliminary quadratic and latent basis models (McArdle & Epstein, 1987; Meredith & Tisak, 1990; Ram & Grimm, 2007) with linear models revealed that the former did not provide a better representation of the data than linear models. For these reasons, and because adding nonlinearity in ALTs involves constraining meaningful parameters, linear ALTs were estimated.

⁶A reviewer noted that the group comparisons implicit in these interactions effects rely on assumption of strict measurement invariance of the body image and self-esteem constructs in gender, ethnicity, and gender X ethnicity groups. Upon verification, these assumptions were reasonably met in this study.

*p ≤ .05.

*p ≤ .01.

* = interaction; b = regression coefficient; s.e. = standard error of the coefficient; p = statistical significance.