ABSTRACT

Quality in a statistical sense means meeting a specied standard. It is always desirable to aim at perfection, but that is quite often elusive. For example, it would be wonderful to have a hospital that discharges fully cured patients all the time without any death or disability, but that is impossible to achieve. Thus, a lower standard of, say, a 90% cure rate in casualty cases and 98% in routine admissions may have to be xed. Limitations of knowledge and cost considerations often determine this feasibility level. This applies to statistical methods also. Once a standard-high or not so high-is xed, everything possible should be done to achieve it. This is best done by controlling errors.