Comparison of analysis of variance and multiple linear regression techniques as applied to an experiment of mango storage

Abstract

The purpose of this study was to compare two techniques of data analysis: Analysis of variance and linear multiple regression, as used to evaluate the effect of flexible packaging films, during storage, on sweetness and development of off-flavors in mango var. Tommy Atkins. The adjusted analysis of the variance model allowed to observe differences between levels and the general mean, but did not allow observing differences among the levels of the different factors. It also allows determining differences between the slopes of each level and the general slope, but the level used as a base to run the model remains unknown. The linear regression model allows determining differences between each level and the level used as a base. It also showed differences in slopes at each level compared to the slope of the level used as a base, as well as information regarding all the levels of one factor. After evaluating the effect of the four different packaging films on the quality of mango, it can be concluded that the micro- perforated and the thermo-shrinkable films are the most adequate. The thermo-shrinkable film results in the least production of off-flavors when compared to the control and, along with the micro-perforated film, shows the same sweetness level as the control. This was not true with the two polyethylene films, where sweetness increased with time, indicating a higher ripening rate compared to the control.