The effectiveness of multiple matrix sampling was examined in a study of 495 students' attitudes toward a computer education program. A post hoc analysis of variance was used, with the class as the unit of analysis. The Computer Attitude Scale, a 30-item rating scale, was used to measure attitudes toward learning about and working with computers. The amount of computer experience was also obtained: less than one month, one to six months, or over six months. Results indicated that: (1) increasing amounts of computer experience were associated with more positive attitudes toward computers; (2) means obtained by multiple matrix sampling generally estimated actual means satisfactorily; (3) the same conclusions were drawn from using the estimated means in a one way analysis of variance as were drawn from using population values; and (4) power of the analysis was affected by the class size. (Author/GDC)