We conducted a secondary data analysis of the 2005 Styles mail panel survey (ConsumerStyles, HealthStyles, and YouthStyles) because it included a household module. Styles is a proprietary consumer and health behavior database developed by Porter Novelli with data collected by Synovate, Inc. In 2005, the Centers for Disease Control and Prevention and the National Cancer Institute purchased the rights to analyze the de-identified data. Because Porter Novelli/Synovate, Inc, funds and conducts data collection, approval of the Office of Management and Budget and the institutional review boards is not necessary. Porter Novelli/Synovate, Inc, adheres to all professional standards and codes of conduct set by the Council of American Survey Research Organizations (www.casro.org/pdfs/10CodeOfStandards.pdf
Panel members were first surveyed with the ConsumerStyles survey. The ConsumerStyles survey, conducted May through June 2005, was sent to a stratified random sample of 20,000 adult panel members (N = approximately 450,000 adults aged 18 or older), which was balanced to help create a nationally representative sample. Low income and minority subgroups (blacks and Hispanics) and households with children were oversampled. The response rates were 65% for the main sample, 59% for the low-income and minority sample, and 62% for households with children.
From July through August 2005, approximately half (n = 6,209) of the panel members who completed the ConsumerStyles survey were followed up with the HealthStyles survey for households without children (n = 3,692) or with both the HealthStyles and YouthStyles surveys (n = 2,517) for households with children. Of the 6,209 panel members who received the follow-up surveys, 4,943 (80%) completed them. The HealthStyles survey was completed by an adult or parent in the household. The YouthStyles survey, mailed in conjunction with the HealthStyles survey, was completed by 1 child in the household. The YouthStyles survey targeted children aged 9 to 18 years and provided linked data between parents and their children. Of the 2,517 households with children that received both the HealthStyles and YouthStyles surveys, 1,685 (67%) completed both surveys. Participants who responded to these surveys were nominally compensated for their participation ($1-5).
Of the 1,685 participants who completed both the HealthStyles and YouthStyles surveys, we excluded those who had missing demographic data (n = 30) and those with missing responses to the questions that asked whether children have a television in their bedroom (n = 210), they are active as a family (n = 18), they allow junk food consumption (n = 5), or they go to fast-food restaurants as a family (n = 10). For the analyses that examined whether the household practices were associated with child or adolescent television watching, physical activity, and dietary behaviors, we excluded participants with missing responses to these questions (n = 232), resulting in an effective sample size of 1,190. The analyses that included BMI percentiles had an effective sample size of 1,050 (missing BMI percentiles, n = 140). The panel members included in the analyses, compared with those excluded for the reasons cited above, had a significantly higher weighted proportion of children aged 9 to 12 (46.8% vs 38.8%), whites (71.7% vs 59.5%), families with higher education (34.6% vs 23.7%), higher income families (39.0% vs 28.0%), more parents aged 35 to 44 (46.2% vs 41.6%), and a higher proportion of families with married parents (84.5% vs 76.3%).
Parents self-reported their age, sex, education, race/ethnicity, marital status, and annual household income. Household practices were assessed with 4 questions. Parents were asked, “Does your child have a television in his/her room?” (yes/no). Using a 5-point response format (strongly agree, agree, neutral, disagree, and strongly disagree), we asked parents to what extent they agreed with the following statements: 1) “I don’t allow my children to eat junk food,” and 2) “We are active as a family.” Responses were recoded into 3 categories (agree/neutral/disagree) for the analyses. Family visits to fast-food restaurants were assessed with the question, “How many days per week do you take your children out to eat at fast-food restaurants, such as . . . ?” Answers ranged from 0 to 7 days.
Age and sex of children and adolescents were reported by their parents. Children self-reported their height and weight, which were used to calculate BMI percentile categories; children with a BMI at or above the 85th percentile and less than the 95th percentile were classified as overweight, and those with a BMI at or above the 95th percentile were classified as obese (19
). Television-watching time was assessed with 2 questions: “After school, on an average weekday (Monday to Friday) during the school year, how many hours a day do you spend watching television?” and “On an average weekend (both Saturday and Sunday) during the school year, how many hours a weekend do you spend watching television?” For both questions, the responses were 0 hours, less than 1, 1, 2, 3, 4, and 5 or more hours. Eating behavior at school was assessed with the following question: “How many days in a typical week did you buy sodas or other snack foods such as chips, chocolate bars, or cookies from vending machines at school?” Answers ranged from 0 to 5 days. Vigorous physical activity (VPA) was assessed with 1 question: “On how many days of the past 7 days did you exercise or participate in physical activity for at least 20 minutes that made you sweat or breathe hard?” Examples were provided and answers ranged from 0 to 7 days. For the analyses, those who were active at least 3 days were classified as meeting the VPA recommendation, those who were active 1 to 2 days were classified as doing some VPA, and those who reported 0 days were classified as doing no VPA. The television watching and VPA questions were selected from the Youth Risk Behavior Surveillance System (YRBSS) survey; the television questions were modified to ask about television watching behaviors on weekday and weekend days instead of just focusing on an average school day (23
). The validity of these questions has been assessed previously with accelerometers (r
= 0.36 for the VPA question, r
= 0.37 for television watching on weekday, and r
= 0.47 for television watching on the weekend) (24
Stata version 11.0 (StataCorp LP, College Station, Texas) was used for all the analyses. All analyses were weighted so that the results would be reflective of US families in terms of race/ethnicity, sex, household size, and income matched to the US Census 2004.
Three multivariate polytomous logistic regressions were performed to examine the associations between household practices and the behaviors of children and adolescents (ie, television watching, participation in physical activity, and purchasing sodas and snacks at school). All the household practices were entered as independent variables because parental practices may be inconsistent across areas (eg, families that provide their child with a television in his or her bedroom may or may not have rules regarding access to less healthful foods). A multivariate polytomous logistic regression was performed to examine associations with self-reported BMI percentile categories; household practices and the behaviors of children and adolescents were entered as independent variables.
For all the analyses, sex and age of the child, race/ethnicity of family, and parent education were entered as covariates. Income was also considered as a covariate; however, education had a stronger association and when both variables were entered, income did not remain significant. A Bonferroni correction was used to control for family-wise error rate. Significance was set at P < .01 (.05/4 hypotheses = .013).