[Discrimination of errors in statistical analysis of quantitative data with common designs].