Given the public health risks associated with obesity and the strong demonstrated influence that parents can have on their children’s energy balance behaviors – one of the most proximal determinants of obesity – understanding and improving obesity-related parenting practices is warranted [13–19]. Advancing current and future obesity-prevention efforts necessitates the reduction of gaps and barriers to effective measurement of such parenting practices.
Obesity-related parenting practices must be measured more consistently and validly [30, 43, 44], but researchers must also address the challenges of gathering parenting data such as cost, time intensiveness, and low response rates, particularly if populations are hardly reached [23, 31].
This project study aimed to assess the role of proxy reports from the recipients of obesity-related parenting practices (i.e., adolescent children) as prospective obesity parenting measurement targets. We found that adolescent responses were significantly correlated with parent responses on their obesity-related parenting practices across parenting domains, and each construct showed strong internal consistency. Also, we found statistically significant invariance in the screen time parenting measure across groups. Lastly, actor-partner effects showed that both parent- and adolescent-reported parenting practices for fruit and vegetable consumption were associated with parent- and adolescent-reported fruit and vegetable intake (3 out of 4 pathways), and only adolescent-reported physical activity parenting practices were associated with parental physical activity and their own physical activity.
Establishing the validity of these four obesity-related parenting domains addresses a significant gap in the literature on obesity-related parenting. Systematic reviews conducted by Vaughn et al. and Davison et al. on the measurement of diet-related and physical activity-related parenting practices, respectively, demonstrated need for obesity-related parenting practices measurement tools with established validity [30, 44]. An additional review study of physical activity parenting measurement found widespread use of non-validated tools [43]. With the validation approach taken in our study, we hope future research adopts well validated tools to measure obesity-related parenting.
Adolescents’ reports on their parents’ obesity-related parenting practices were also evaluated with regards to their similarity to their parents’ reports on the same obesity-related parenting practices. Our findings show that they are significantly associated. This is indicative of the relative congruity between adolescent proxy-reports and parent self-reports of their obesity-related parenting practices. While the literature is limited, our validation results support prior findings that adolescent proxy reports may be a valid source of information on obesity-related parenting practices. Previous research has found child proxy reports of parents’ behaviors are less accurate than self-report, but also showed that accuracy increased as children were older [31, 45, 46]. Additionally, much of this work was investigating reports on socioeconomic status, not parenting practices, and there is evidence to suggest that child proxy reports are more accurate in areas that are more salient to them [31]. Considering that obesity-related parenting practices are reasonably salient to adolescents, our findings support past research and suggest that adolescents are reliable sources of information on obesity-related parenting practices.
In addition to the factorial validity and reliability of adolescent proxy reports, the predictive validity of adolescent- and parent-reported obesity-related parenting practices on diet and physical activity outcomes was also measured and compared. This was an important additional step because obesity-related parenting practices are most pertinent to the public health issue of obesity in how they influence energy balance behaviors. By modeling actor-partner effects, we found that both adolescent and parent reports on fruit and vegetable parenting practices were predictive of adolescent fruit and vegetable intake, but adolescent reports yielded a higher slope coefficient at a higher significance level than parent reports. With physical activity, only adolescent reports of parenting practices predicted adolescent outcomes. Only one small study has assessed parent-practices in parent-child dyads, but it found that child reports of obesity-related parenting practices were associated with child diet and physical activity, but parent reports were not [47]. There are two major hypothesized reasons why adolescent proxy reports on parenting could be more predictive of behavioral outcomes: parent reports may be subject to social desirability bias, and the effects parenting behaviors have on adolescent behaviors likely be contingent on adolescent perceptions of them [48]. As such, our findings suggest that parent- and adolescent-reported parenting practices correlated very well, thus parent-reporter bias is unlikely to have had an impact in the study outcomes.
This finding that adolescent proxy reports could potentially be more predictive of adolescent energy balance behaviors has substantial implications for future research on obesity-related parenting practices. The primary justification for studying obesity-related parenting practices is the strong influence they may have on energy balance behaviors. Our findings suggest that adolescent-reported parenting practices are more predictive of pertinent behavioral outcomes than those that are parent-reported, and as such, might be a relatively more accurate source of parental influence, rather than simply a valid alternative proxy. Future work should be conducted to further clarify and confirm this disparity of predictivity between adolescent and parent reports and the reasons for this disparity, as well why neither adolescent nor parent reported screen time and junk food parenting practices were predictive of relevant behavioral outcomes.
There have long been issues with measurement of parenting practices, including but not limited to cost, time, response rate, and bias [23]. Adolescent proxy reports could potentially address some of these barriers [31]. In studies that are gathering data from adolescents that should be analyzed in comparison with parenting practices, the ability to gather that information from only one study subject, rather than bringing parents in would greatly lift the logistical burden of such research. Furthermore, the existence of schools and other community settings as centralized places to partner with, recruit, and follow up with adolescents may make them a study population that is feasibly approachable. In these ways and others, using adolescent proxy reports could offer an avenue that makes research on parenting practices more feasibly, and could open doors to research on otherwise hard to reach populations. These findings are also promising for future family-based obesity prevention strategies, as this indicates that adolescents could be trustworthy stakeholders in obesity prevention research and interventions.
There are several strengths to this study. First, data was taken from a large sample that was selected to be demographically representative of the U.S. population. This allows for generalizability of the findings in this study at the population level. Furthermore, having dyadic data from parents and their adolescent children allowed for direct, ensuring that the parent-adolescent reports were on the same parenting practices. Furthermore, including four obesity-related parenting domains allowed our team to get a more complete picture of this concept and how adolescents report on them, rather than focusing in on behavioral domains independently. Another important strength of this study is the variety of validation steps that were taken to ensure the parenting practices measures included in our study were rigorously validated; this is contrasted with past studies that have often used unvalidated tools [43].
There are a few limitations to this study that are important to mention. The main outcome measures rely on self-report data, which can be vulnerable to social desirability bias, as respondents may feel they should answer to reflect what they should be doing rather than what they actually do. Past research has shown that on self-reported questionnaires, social desirability bias can lead to underreporting of eating and overreporting of physical activity [49, 50]. Our reports on energy balance behaviors can also be subject to recall bias; past research has shown that asking for self-reports of physical activity in the past can be flawed when compared to data collection of more recent physical activity, such as in the form of a daily diary [51]. Lastly, our study did not assess gender concordance between parents and adolescents which some evidence suggests it is relevant to the influence of parenting practices [52, 53], which future studies should account for this contribution.