engleski

META-ANALYSIS AND THE COMBINED ANALYSIS OF VARIANCE OF THE STUDIES WITH BIOSTIMULATOR TREATMENT IN APPLE

Meta-analysis is novel statistical tool in agricultural sciences intended for summarizing treatment effects and the conclusions in scientific studies. Its application in agricultural sciences in general and especially in horticulture has been limited so far. This paper aims i) to introduce meta-analysis model ; ii) to compare the results with traditionally and widely recognized analysis of variance ; and iii) to discuss the application of those two models in five studies in apple. Data from five studies of Golden delicious apple cultivar treated with Sigma biostimulator were generated for yield. Dataset included normally distributed both treatment and control (non-treated) cases, with equal number of plots and trees in the plots, in order to match requirements for both combined analysis of variance and meta-analysis. Results and Five studies were analyzed using first the method of analysis of variance and provided full modeling with results. It was found that out of five studies three resulted with statistically significant treatment effect and two without. Combined analysis of variance of the studies set up as blocks was provided by highly significant result (F=9.22, p=0.002), in favor of biostimulator treated plants in comparison to the control plants. Meta-analysis of the same studies proved to provide similar results as by combined analysis of variance, but with much less requirements for the data, i.e. only metadata were used. In addition, straightforward meta-analysis provided average effect size with its confidence intervals and thus enabled easy comparison with individual studies. This research has demonstrated through practical example that meta-analysis yields results directly comparable to the combined analysis of variance. Moreover, meta-analysis requires only metadata and therefore it represents a good method of selection of the reliable decision making on the available research regarding the statistical significance of the effect and, more practical, treatment effect size in situations where several studies are compared on the basis of similar treatment. This is of particular importance for treatments providing ambiguous results through different studies, which is often the case in agricultural research.

meta-analysis, anova, horticulture, fruit science, treatment effect

nije evidentirano