The value-added model is presented as a method of assessing the impact of treatments in pretest-posttest quasi-experimental designs. The model yields a projection of treatment group performance at posttest time with no special treatment. The difference between the expected result and the actual outcome is "the value added," a measure of treatment effect. No-comparison-group design cases were investigated, and problems and procedures associated with linear and non-linear growth models are discussed. Ages, pretest scores and posttest scores with known means and covariances were generated, and expected growth was calculated from the age and pretest score. The estimated value-added was compared with the actual value-added, and the power of the significance test for the treatment effect was observed. An adequately-specified growth function, sufficient sample size and a reliable, valid test, well-correlated with age and free from ceiling and floor effects, provided an effective model when the measured variable was a function of age or time. Since the model allows for flexibility in the pretest-posttest interval, and does not always require a control group, it was found particularly suitable for early childhood program evaluation. (Author/CM)