Overall, the Home Environment Survey showed consistency with previous research on childhood nutrition and physical activity and their relationship to the home environment. Parental role modelling of physical activity, parental policies to support physical activity in children and the availability of physical activity toys all showed correlations with child physical activity which supports previous research. Child nutrition findings were also consistent with the previous research that links child nutrition to family eating policies, parental role modelling and the availability and accessibility of foods in the home. In addition, the concepts of availability, accessibility, parental role modelling and parental policies correlate with each other.
In comparison with other surveys measuring child eating patterns, home environment or physical activity environment, the Home Environment Survey showed comparable internal consistency and reliability. The physical activity home and neighborhood environment questionnaire developed by Hume had similar test-retest correlations and internal consistency (e.g., α = .43–.77) for scaled responses. In Golan and Weisman's Family Eating and Activity Habits Questionnaire, internal consistency and test-retest reliability were high. The diet related psychosocial questionnaire developed by Hume showed similar Cronbach's Alpha scores for food availability scales and also showed slightly smaller correlations between the food availability scales and reported food intake.
Although most scales on the Home Environment Survey demonstrated good internal consistency and reliability, scores were lower than expected for Fruit/Vegetable Accessibility and Fat/Sweet Accessibility. These questions were very broad (i.e. "How often do you store high-calorie foods in a place that was known but not seen?") and developed so that there would be consistency across behaviours. From participant feedback while taking the survey, many thought it would be better to make these questions very specific such as "How often are fruits kept in a fruit bowl on the kitchen counter or table?"
For the inter-rater reliability, the vast majority of "primary" caregivers were mothers who also said that they were the primary person who prepared food, planned activities and had the most knowledge of their child's eating and physical activity. With this in mind, inter-rater reliability was still quite high on most sections of the Home Environment Survey. Fat/Sweet Accessibility had a low inter-rater reliability most likely because of the broad questions that were open to interpretation. Physical Activity Role Modelling and Physical Activity Parental Policies both had much lower inter-rater reliability; however, this should be expected because physical activity is often an individual choice and it is likely that two parents could be very different in their amount of physical activity and in their parental policies to encourage their children to be physically active. Healthy Eating Role Modelling and Healthy Eating Parental Policies did have reasonably high inter-rater reliability indicating that nutrition is likely a "family affair." Given that the same foods are available to the entire family and that families often eat together it is reasonable to think that parents would have more similar nutrition role modelling and parental policies around food. Although not within the scope or data available for this study, an interesting future area of research could consider the impact of having more homogeneous or heterogeneous parent perceptions of role modelling and parental policies on child behaviour.
Most scales in the Home Environment Survey did show significant correlations with the appropriate child physical activity/nutrition or parental physical activity/healthy eating variables. However, the availability of physical activity equipment/spaces did not significantly correlate with a child's physical activity. This may be a case where availability is necessary but not sufficient by itself without accessible equipment or play spaces to encourage physical activity. Similarly, accessibility of vegetables was not significantly correlated to child vegetable consumption.
For child percentage of kilocalories from sweets, the fact that availability of fats/sweets was not significant and that accessibility was correlated significantly but in the opposite direction of what was hypothesized, may indicate that other environments play a larger role in fat/sweet consumption for children. Many parents commented that children were often getting sweet/high fat foods at school or activities away from the home. In a study of European children, children reported that fruits/vegetables were not as available when they were outside of the home and that high calorie snacks were more available outside the home lending support to this hypothesis [29
Parental physical activity was correlated to physical activity role modelling and parental policies; however, the correlation between role modelling and parental physical activity was only moderate. This is likely because the role modelling items on the Home Environment Survey referred to physical activity that was directly observed by the child. Many parents who were regular exercisers reported that they exercised early in the morning or at work when their children could not directly observe their physical activity.
Three main limitations apply to this study. First, all children participating in this study had a Body Mass Index at or above the 85th percentile placing them at risk of overweight or overweight status. This validation study was not able to compare home environments of families whose children were at a healthy Body Mass Index. If children with healthy Body Mass Indices had been included, the ability to detect differences in the home environment between the two groups would have added to the validity testing of the Home Environment Survey survey. The second limitation of this study was that it cannot be generalized to all populations since all study participants were drawn from the Kaiser Permanente membership in one geographic region of the United States. Third, this study is a cross-sectional study using baseline data and so cannot assess the ability of the survey instrument to detect change over time.
Future investigations are necessary to test the survey in a larger sample to allow for the use of confirmatory factor or latent modelling to determine the strength of the factor structure proposed by the Home Environment Survey. Many parents reported that they felt that older children (11 and 12) were less influenced by the home environment as they spent more time at school and out with friends. Testing this survey in several different age populations would be helpful to determine if this survey is equally valid in all age groups. Finally, we had originally intended to include sedentary behaviour as reverse scored items within the physical activity sections; however, it became clear in attempting to score these items and in preliminary analysis that sedentary behaviour and physical activity are related but separate domains. The questions we had included within the physical activity availability, accessibility and parental role modelling sections did not correlate well with the other questions in their scales. In retrospect, we believe it would have been better to have separate scales for sedentary activity. We had also originally included a parental policies section that asked about parental limits for sedentary behaviours. We had an insufficient response rate for the section on parental limits for sedentary behaviours perhaps because parents had a difficult time quantifying the amount of time children are allowed to spend on sedentary activities.