The National Evaluation of School Nutrition Programs: data analysis methods.

An overview of the analytic methods used in the National Evaluation of School Nutrition Programs is provided. A path model is described that shows the hypothesized relationships between outcome and predictor variables. The outcome variables are participation in the School Nutrition Programs and the effects of participation on family food expenditures and student dietary intake and anthropometric measurements. Individual student participation in the programs is seen as the last of a series of participation decisions that include the school's decision to offer the program and the prices charged for meals. Selection of specific statistical and structural models to present these relationships are discussed. The decision to use ordinary least square estimates of impact rather than a modified two-stage least squares approach is explained.