Qualitative Evaluation of Statistical Methods Used in Some Soil Science Research

In order to evaluate quality of the statistical methods used in some agricultural research, a total of 100 final research reports, graduate theses, and published papers in Iranian agricultural journals were selected by simple random sampling. Then, using a designed check-list, each sample was reviewed for the factors such as design of the experimental methods, method of treatments randomization, number of investigated treatments, sampling methods, number of measured traits or variables, and the statistical tests and methods applied in the investigation and analyses of the results, methods of mean comparison, frequency distribution of statistical tests and design of experiments, the software used, quality of statistical reference, the way statistical tables and graphs were used in presentation of the results, and whether or not a statistical expert participated or was consulted in the research. After reviewing the Check-Lists, the most common problems and errors in the samples were divided in two groups: first, errors related to design of the experiment and, secondly, errors related to the analysis of the results. Afterwards, the importance of correct decision with respect to each of the abovementioned factors and the impact of negligence in each case were evaluated. Finally, the items whose application in the experimental design or result analyses were not essential, such as describing mathematical model of the design or experimental layout, but could affect experiment implementation and results presentation, were mentioned. According to the results, adequate training in statistical method, design of experiments, and use of the related soft wares would enhance the reliability and efficiency of agricultural research.