Abstract 694

Preference involves liking for specific items and/or choosing one thing over something else. Food preferences are thought to be determinants of consumption. Children's food consumption patterns and nutritional status may be effectively improved through school-based programs because more than 95% of children in the US are enrolled in school, and because children eat one or two meals a day at school. The National School Lunch Program is available to about 92% of all students in the US; on a typical day, about 56% of the students participate. This study determined the relationship between fourth-graders' consumption of and preferences for school lunch foods. Data for this study were collected during a series of studies concerning the accuracy of children's school lunch recalls. Students from 4 schools in one school district in Georgia were randomly selected, observed eating school lunch, and interviewed within 1.5 hours after lunch (n=89; 13 white m, 12 white f, 29 black m, 35 black f) or the following morning (n=148; 17 white m, 16 white f, 61 black m, 54 black f). Fifty-two to 75% of the students at the 4 schools were eligible to receive free or reduced price lunches. Amounts observed eaten were recorded as none, taste, little bit (or some), half, most, all, and >1 serving, and coded as .0, .1, .25, .5, .75, 1, and 2, respectively. School lunch items were categorized as either entree, bread, vegetable, fruit (including fruit juice), dessert (non-fruit), or milk. Preferences were coded as liked "not at all"=0, "a little"=1, and "a lot"=2. A General Linear Models Approach was used with consumption as the dependent variable after blocking on student (because each student had multiple food items and most students had varying preferences for the foods). The effects of sex, ethnicity, and food category on consumption were tested, and whether their interactions with preferences were significantly related to consumption. Statistical calculations were conducted using SPSS version 7.5. Results indicated a direct relationship between observed consumption and preferences (p<.001); consumption (least squares means or LSM) was .13, .50, and .92 servings for foods liked "not at all", "a little", or "a lot", respectively. As expected, a significant difference in consumption by sex (p<.01) was found; LSM were .73 servings for males and .64 for females. Vegetable consumption was significantly less than consumption of all other food categories (all p's<.001) even when controlling for preferences. All other main effects and interactions in the model failed to reach statistical significance (all p's >.20). Pediatricians need to be aware that children eat what they like "a lot" and don't eat what they don't like during school lunch; thus, food preferences need to be considered when implementing dietary interventions with children.

Research funded by NIH grant CA60806.