The current study examined child and parent correlates of energy balance-related parenting practices, as well as the association between these practices and diet and activity behavior at age 5, and BMI development from age 5 to 7 years. Parents were found to be more restrictive regarding their daughters' diet than their sons', which is in line with previous research . However, the current study also showed that girls were less restricted than boys when it came to sedentary time. Parents may have different priorities for boys and girls when it comes to restricting unhealthy behaviors; perhaps inactivity is of greater concern to parents where their sons are concerned, while overconsumption is of greater concern to parents where their daughters are concerned. Parental restriction of unhealthy intake was also positively associated with child BMI, in agreement with previous studies . Child BMI was also positively associated with parental stimulation of healthy intake. Both the increased restriction of unhealthy intake and the increased stimulation of healthy intake in heavier children might reflect reactions of parents to their child's weight, trying to get heavier children to eat a healthier diet so as to decrease their weight. A similar mechanism might be operative for children with a hungry or picky eating style, who were shown to be more restricted by their parents. Parents might feel that these children need more external control over their eating to compensate for their deviant eating style. In view of the cross-sectional nature of our data, however, we cannot exclude the possibility that these children's eating style actually became more deviant in reaction to the strict control their parents exercised over their diet. In line with the latter explanation, various studies have reported that high parental control over child eating interferes with children's self-control over their intake [e.g., ], thus leading to a deviant eating style.
There were also several parental characteristics that predicted which practices parents would apply. Maternal BMI was found to be inversely associated with dietary restriction and stimulation, which confirms previous findings [26–28]. Maternal educational level was positively associated with stimulation of both healthy intake and PA. This adds to previous research showing that parental education is positively associated with restriction and other controlling practices [10, 25, 26]. The number of hours that the mothers worked was negatively associated with monitoring their children's diet and activity behavior and stimulation to be physically active. A similar association was previously reported by Brown and colleagues , showing that parents who stayed at home to take care of their children exercised stricter control over their children's diet. As working parents leave part of the child rearing to others, such as child-care staff [e.g., ], they may be inclined to be less strict during the limited time they can spend with their children.
With regard to the associations between parenting practices and their children's behavior and BMI development, we found that monitoring a child's diet and stimulating healthy intake were both associated with the child having a healthy diet. Stimulation of healthy intake even had a desirable effect on the child's BMI development up to the follow-up at age 7. By contrast, dietary restriction was not associated with any of the dietary outcomes, nor was it associated with BMI development. Previous studies have shown conflicting results with regard to all three of the above parenting practices (monitoring [e.g., [17–19]]; stimulation [e.g., [12–15]]; restriction [e.g., [7–11]]), with some studies supporting our findings and some contradicting them. We believe that the key to resolving these conflicting findings might lie in the interaction between children and parents. In line with our hypotheses based on previous studies [e.g., [10, 18]], the current study showed that the associations between parenting practices and child behavior and weight development depended on the children's characteristics. Dietary restriction was associated with undesirable dietary intake behaviors by children with a deviant eating style (i.e., children who were relatively hungry compared to peers). In line with this, previous research showed that the associations between restriction and desirable dietary intake behavior at a very young age (2 years) were partly lacking in children with deviant eating styles . Analogous to our findings with regard to restriction, the associations between monitoring and a desirable child diet were not found for relatively hungry children or picky eaters. By contrast, stimulation to eat healthy was found to be specifically beneficial for picky eaters, as well as for children with a high BMI. In line with previously raised hypotheses , this indicates that although restriction and monitoring might be less suitable for children with certain unfavorable characteristics (e.g., deviant eating style, high BMI), stimulating these children to eat a healthy diet seems all the more effective for them. It is worrying, however, that picky eating also correlated with less parental stimulation to eat healthy. Educating parents might therefore be an important step toward improving children's diet, perhaps especially for children with a deviant eating style.
The effects of rules about television viewing on activity behaviors have previously been found to depend on the child's gender, with desirable effects on girls, but undesirable effects on boys . We did not find indications of such a difference in the current study, but we did find undesirable correlations between restriction of sedentary time and behavior and BMI development, for both boys and girls; restriction was associated with increased sedentary time, decreased PA, and an increased BMI development up to age 7. This contradicts previous studies that showed that explicit rules restricting children's television watching were associated with less viewing time [22–24]. An explanation for these contradictory findings might lie in the assessment of restriction of sedentary time in the current study, which not only included explicit rules limiting television and computer use, as in the previous studies, but took a broader view of restrictive parenting. For example, the measure of restriction of sedentary behavior in the current study included items assessing what parents thought would happen if they did not restrict their child's sedentary behavior (see Table 1). The inclusion of such broader items was based on the diet-related restriction scale of the CFQ . Stimulation to be active was positively associated with children's PA and negatively with sedentary time in our study, which is in line with a review showing that encouragement and support are important predictors of increased PA .
The findings of the current study have implications for both research and practice. With regard to research, studies into the effects of parenting practices that do not incorporate the possibility of moderation by child characteristics will tend to produce conclusions on the effects of parenting practices that strongly depend on their study population. In addition to the moderators identified in the current study, previous research has revealed several additional child factors that moderate the effects of parenting practices, including the child's personality, temperament [10, 11] and gender . These interactions might also contribute to the many contradictions in the current evidence base on diet-related parenting practices. Therefore, we believe that research into the effects of parenting practices cannot be limited to the direct association between practices and outcomes, but should always incorporate a theory-based examination of possible moderation effects [35, 47]. The practical implications of the current findings are that overweight prevention interventions targeted at parenting practices should be tailored to individual child characteristics, since specific parenting practices might be beneficial for one child, but useless (or even potentially disadvantageous) for another.
The current study had several strengths and limitations. One of the strengths is that it included a longitudinal follow-up to assess the effects on BMI development. However, behavioral outcomes were only assessed cross-sectionally. Thus, we cannot establish whether these behaviors are the consequence of certain parenting practices, or that they perhaps evoke these parenting practices. The same goes for eating style and activity style, which we regarded as relatively stable child characteristics, and therefore included as predictors of parenting practices. They could, however, also be influenced by parenting practices. Many of the previous studies in this research area have limited themselves to cross-sectional explorations, and there is a need for prospective research to establish causality [e.g., [7, 21]]. It is reassuring, though, that the associations between two of the parenting practices (i.e., stimulation of healthy intake and restriction of sedentary time) and behavior were supported by the associations with later BMI development, pointing in the same direction. An additional strength is that the data in the current study were assessed prospectively, limiting the risk of recall bias and other problems inherent in retrospective research.
A major limitation of the current study is that all data, including dietary intake, activity behavior and anthropometrics, were self-reported by the parents, which may have led to bias. However, previous research has shown that parental reports of weight and height differed little from measured data . An additional limitation is that the Cronbach's α values of some of our scales were relatively low. Although a Cronbach's α ≥.6 is generally considered acceptable , some authors advocate different cut-off points. Furthermore, caution is warranted when generalizing our results to the broader population of young Dutch children. Parents with an 'alternative', relatively healthy lifestyle were overrepresented in our sample, due to the choice of recruitment methods, i.e., recruiting some of the women from 'alternative lifestyle' circles . The relatively healthy average lifestyle of our study sample is reflected in the children's relatively low mean BMI z-scores. However, secondary analyses showed that excluding the children who where underweight at age 5 did not change our findings. Moreover, all analyses were adjusted for recruitment channel. Finally, it may be noted that the reported effect sizes are small, indicating that the amount of variance in behavior and weight status explained by the parenting practices is limited. This may be partly attributable to the fact that parenting behavior is a concept that is hard to assess, and there is no consensus about the proper way to measure it. There are dozens of questionnaires assessing diet-related parenting practices, activity-related parenting practices, or both [e.g., [37, 50–53]]. We feel quite confident, though, about the instruments adapted from the CFQ  for the current study, although the diet-related 'stimulation of healthy intake' scale and the 'Activity-related Parenting Questionnaire' were not previously validated. The fact that our adapted scales for 'stimulation of healthy intake' and 'restriction of sedentary time' predicted BMI change from age 5 to age 7 may be considered reassuring in this respect. Future research would benefit from a consensus about feasible and valid measurement methods.