Premium
ISOMORPHIC PERMUTED EXPERIMENTAL DESIGNS AND THEIR APPLICATION IN CONJOINT ANALYSIS
Author(s) -
GOFMAN ALEX,
MOSKOWITZ HOWARD
Publication year - 2010
Publication title -
journal of sensory studies
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.61
H-Index - 53
eISSN - 1745-459X
pISSN - 0887-8250
DOI - 10.1111/j.1745-459x.2009.00258.x
Subject(s) - conjoint analysis , computer science , respondent , identification (biology) , reliability (semiconductor) , design of experiments , product (mathematics) , segmentation , data mining , machine learning , artificial intelligence , statistics , mathematics , preference , power (physics) , botany , physics , geometry , quantum mechanics , political science , law , biology
This paper deals with experimental designs used in conjoint analysis. The described approach permutes the structure of the underlying fractional experimental design to make different sets of combinations. The resulting experimental designs, suggested to be called Isomorphic Permuted Experimental Designs (IPED), are statistically equivalent to each other while combining diverse sets of the variables and levels into different designs. By facilitating distinctive individual designs (for each respondent), IPEDs reduce a bias caused by some possibly unusually strong performing combinations, and allow detection and estimation of interactions among variables, as well as identification of pattern‐based segments emerging from individual models of utilities. This paper examines the theoretical foundation of the approach, formalizes the methodology for algorithmic implementation and shows a practical example of utilization.PRACTICAL APPLICATIONS Isomorphic Permuted Experimental Design (IPED) allows for overcoming multiple interlinked statistical problems that affect the traditional conjoint analysis approaches, thus leading to more reliable and targeted results in practice. IPED facilitates individual respondents' models based on unique designs, thus allowing for pattern‐based segmentation. The approach also allows for the detection of any and all interactions between the elements (features) of the experiments, thus increasing the reliability of conjoint analysis results. It has been utilized in many practical applications, such as for message optimization, early stage new product development, advertising, package and website optimization.