Several sampling procedures for collecting observational data on student activities were studied in an effort to determine their relative efficiency. The setting was a fifth grade Individually Prescribed Instruction (IPI) mathematics class of thirty-three pupils. A criterion measure was obtained by cumulating the measurements obtained on the thirty-three pupils every thirty seconds for thirty minutes over a period of twenty days. Various observational designs were created by manipulating the following four variables: (1) the length of the observation period; (2) the time interval between recording observations during the observation period; (3) the number of days (replications) of observing; and (4) the number of pupils observed each day. Through this procedure it was possible to build observational designs and compare results using each design with the criterion measure, thus determining the efficiency of the several sampling plans. Comprehensive tables are included showing Scott coefficients of agreement with the criterion of the various designs. (Author/LR)