The usefulness of strictly quantitative research is - due especially to deficiencies in methodological training - still questioned in educational science. Accordingly, empirical educational research often reveals considerable imperfections. This is illustrated by an analysis of a recent study on violence published by HEITMEYER et al. in 1995. The author criticizes the following aspects: (1) the theoretical quality, (2) the quality of the research design, (3) the quality of the operationalizations and measurements, and (4) the quality of the statistical evaluation. An empirical study is only really `'good'' if each one of these aspects is treated adequately. The description and discussion is restricted to the methodology of the testing of hypotheses, considering the problems of field research, in particular.