Intervention-related bias in reporting of food intake by fifth-grade children participating in an obesity prevention study.