versión On-line ISSN 1678-4596
BERTOLDO, Juliano Garcia et al. Use or abuse in mean comparison tests: scientific or empiric knowledge?. Cienc. Rural [online]. 2008, vol.38, n.4, pp.1145-1148. ISSN 1678-4596. http://dx.doi.org/10.1590/S0103-84782008000400039.
The objective of this research was to verify which are the main mistakes and the successes in the application of mean comparison tests in scientific studies, demonstrating viable alternatives in the sense of increasing the impartiality of the results obtained by researchers. One of the researcher's largest challenges is the interpretation of the obtained results in a trustworthy way. In spite of the concern of most researchers with the data analysis, many times certain disregard is observed in the interpretation of the results. Thus, the incorrect application of statistical tests leads the researchers to publish information not completely reliable. One hundred and forty-eight papers dealing with one or more than one factor were evaluated. All of them are related to the crop production major area, published from 2002 to 2006, in the 'Ciência Rural', a Brazilian scientific journal. Most of the studies (72%) were classified as incorrect due to the abuse of means comparison tests. In addition, only 4% and 24% were classified respectively as partially correct and correct.
Palabras clave : means comparison tests; qualitative and quantitative factors; contrasts.