Agronomy and Horticulture Department


Date of this Version



HortScience, Vol. 33(2), April 1998


Copyright 1998 American Society for Horticultural Science. Used by permission.


Researchers in ornamental horticulture often want to assess the effects of experimental treatments on plant quality. Producers often use the results of such experiments to establish the minimum level of a treatment, for instance, the amount of a growth regulator or a nutrient, such as nitrogen, needed to achieve desired plant quality. For edible plants, quality can be quantified objectively by using numeric response variable such as yield or nutritional content. However, for ornamental plants, quality depends on aesthetic appeal and consumer acceptance, traits which are subjective and qualitative.

Statistical models for the design and analysis of experiments involving numeric or quantitative responses are generally considered “standard” statistical methods. Often, however, the relationship between these variables and quality factors such as aesthetic appeal and consumer acceptance is not clear. For subjective, qualitative response variables, standard statistical methods may not be used without modification. The purpose of this paper is to present statistical methods useful for designing and analyzing experiments to assess plant quality. We focus particularly on the use of unreplicated designs and their analysis using half-normal plots.