The role of a program evaluation unit in the Illinois State Educational Agency (SEA) as it attempts to provide meaningful data for a variety of audiences about projects that have extremely short life is investigated. The focus is on two programs--migrant programs and neglected/delinquent programs--funded under Title I, Public Law 89-750. A common dilemma of these programs is determining the success of children whose attendance is of short duration (less than two months in the migrant programs and an average of seven months in the neglected/delinquent programs). Government guidelines had influenced the SEA and Local Educational Agency (LEA) staff to use standardized tests on a pre and post basis. The evaluation unit found such testing inappropriate because of time factors, student attendance patterns, and student language abilities. As soon as communication linkage developed, there was mutual agreement by the state evaluation personnel and field people that standardized tests provided meaningless data. It was concluded that the intervention of state data collection in local projects requires that the chief audience of the data should be the LEA's themselves. It is only in this way that relevant information can be amassed that will lead to improvement of projects. Data must be collected in a variety of ways and using a variety of tactics. (RC)