Evaluating the Number of Replication Effect on the Validity of Agricultural Experiments Results

پژوهشهای زراعی ایران

Article 19, Volume 10, Issue 1 - Serial Number 25, April 2012, Pages 163-169   

Document Type: Research Article

DOI: 10.22067/gsc.v10i1.14502

Author

M Khazaei^*

Abstract

One of the basic principles of an experiment is the repetition of its treatments so that treatments cannot be statistically compared without repetition. Using enough replications has a great impact on scientific validity, but in most cases, not enough attention is paid. Generally, selecting the number of replications is done according to previous experiments at similar conditions, regarding the number of replications is specifically depended on the experiment conditions. In this study effort has been made to investigate the number of treatment replications effect on the validity of experimental results by considering agricultural experiments. The results have shown that by increasing test error fluctuations and reducing the difference between the treatments that must be found to achieve a minimum level of significance and an acceptable test power, a minimum number of replications will be required. And the minimum number of required replications declines by controlling test error fluctuations. Therefore, under the same conditions, fewer replications are required for experiments under controlled conditions such as growth chamber experiments. Fewer replications will reduce test power and only through increasing the number of replications, significance level and test power can be simultaneously maintained at high level.

Keywords

Replication; Test power; Significant level