Whether initiated by law, regulation, or administrative direction, evaluations have political purposes. Improvement, the classic purpose of evaluation, is most faithfully observed when the importance, funding, and constituency of a program are small. If something is wrong, or if the program's existence is threatened, evaluation can become a weapon in policy disputes, hiding values from constituents in a mystique of scientific inquiry. Purposes may also be reflected in the biases of those who initiate and conduct evaluations, in the selection of a program and objectives of evaluation, and in the amount of evaluation funding relative to program funding. Similarly, problems inherent to evaluation methods are sources of bias. Traditional quantitative measures do not lend themselves to broad and sweeping social programs, such as bilingual education. Finally, evaluation results are often used selectively to further political ends. In conclusion, evaluation can be viewed as an historical enterprise which seeks to recreate the past with selected emphases or biases. (CP)