Many Swedish parents report that they tend to employ moderate control as well as frequent use of democratic parenting practices (Trost, 2012). This approach to parenting combined with policy and institutional supports for child and family welfare make the Swedish parenting context in some ways unique and in need of further investigation (e.g., Trost, El-Khouri, & Sundell, 2015). From a measurement perspective, when parents and children in the same family report on the same construct, one is faced with several questions such as are the child and parent reports reflective of the same construct? If yes, then child and parent reports can be modelled or conceptualized as indicators of the same construct. Another possibility is that adolescents’ and parents’ perceptions of a given construct are fundamentally different from one another. A number of options exist in terms of how to conceptualize measurement when parents and children report on the same construct (Bank, Dishion, Skinner, & Patterson, 1990).

The results presented in this poster are from a longitudinal study of school climate and mental health (study name and citation blinded for peer review). The sample consists of a cohort of Swedish 7th graders (N=3,932, aged 13-14 years old, 51.8% =female, from 101 schools in eight Swedish regions). At grade 7, using structural equation modelling (SEM), different measurement models of family-related factors were tested, namely democratic parenting (Wray-Lake & Flanagan, 2012), parental warmth (Trost, et al, 2007; Tilton-Weaver et al., 2010), and child-parent communication (Kerr & Statin, 2000). Adolescents in the cohort and one of their parents reported on these constructs at grades 7, 8, and 9. The results presented here are for grade 7. Analyses were conducted with Mplus 7.4 (using Mplus’s robust maximum likelihood estimator, TYPE = COMPLEX in Mplus, to account for the nested nature of responses by school building, and FIML estimation for missing data). Three SEM models were tested, including (1) a three-factor model in which parallel child and parent reports were indicators of the same family factors, (2) a six-factor model in which parallel child and parent reports were indicators of separate family factors, and (3) a multi-trait multi-method model with three family factors and two method factors (parent and child report; Model 3). Two models showed good fit to the data: the six-factor model, χ2(89) = 164.873; CFI = .997; RMSEA = .015 [.011, .018], and the multi-trait multi-method model, χ2(84) = 533.735; CFI = .982. RMSEA = .037 [.034, .040]. While both models evidenced good fit, the six-factor model provided the more parsimonious solution. This result indicates that adolescents and parents in this sample may be viewing the family-related factors in different ways. In the poster, other analyses will be reported on the utility of the models identified in grade 7 at the other study time points, such as at grades 8 and 9. Study findings will also be discussed in terms of culture and parenting.