Logistic regression analyses for associations between home environment characteristics and watching TV by the child (n = 3067).

<p>For details on the measures used, see <a href="http://www.plosone.org/article/info:doi/10.1371/journal.pone.0088486#pone.0088486.s001" target="_blank">Appendix S1</a>.</p><p>The frequencies (n (%)) and means presented are frequencies and means of the original dataset. Missing values were 49 (1.6%) for family rules about watching TV, 47 (1.5%) for parental monitoring, 55 (1.8%) for parental urging to turn off the TV, 73 (2.4%) for child autonomy concerning watching TV, 7 (0.2%) for number of TVs in the household, and 12 (0.4%) for whether the child has a TV in the bedroom.</p><p>The ORs are the pooled results of analysis of the five imputed datasets.</p><p>The ORs are adjusted for sociodemographic characteristics (sex and age of the child, ethnic background of the child, educational level of the parent, employment status and family structure).</p>a<p>The reference category (OR = 1.00) is ‘no rules’.</p>b<p>The reference category (OR = 1.00) is ‘never, seldom, or sometimes’.</p>c<p>An increase on child autonomy indicates higher autonomy of the child concerning screen time.</p>d<p>Households without a TV (n = 147, 4.8%) were excluded from analysis.</p>e<p>The reference category (OR = 1.00) is ‘no’.