An Approach to Developing Accountability Measures for the Public Schools.
A general strategy for evaluating pupils' progress so that each educator is held responsible for those outcomes--and only those outcomes--that he can affect, to the extent that he can affect them. The basic statistical technique is multiple regression analysis. Several stages are needed (1) to take account of variation due to the pupils' backgrounds and characteristics, (2) to estimate the remaining interclassroom variation, and (3) to attribute it to particular teachers, other classroom variables, and other school characteristics. In principle, the method can be extended to evaluate the contributions of administrators, provided the district is large enough for adequate comparisons. The interclassroom variation remaining after nonteacher effects have been accounted for is probably the most widely useful measure. Complex econometric models may be required to cope with interaction effects--the effect of problem children on their classmates, or of last year's teacher on this year's performance. 30 pp.