In collaboration with Bertiana Balliu, Raffaele Marchesi and Irene Parolin.
Full report in the related pdf file.
In microeconomics, measurement of consumers’ preferences is one of the most important elements of marketing research. It helps to explain the reasons of consumers’ decisions and can lead to better managerial decision making, as a higher situational awareness is present. Using some statistical methods it is possible to quantify preferences and answer the question: what product or service will consumer choose? To answer this question in a specific use case the present work aims at illustrating an application of Conjoint Analysis methods. The use case considers the product Christmas box and the type of specific Conjoint Analysis used is the choice-based.
The methods of conjoint analysis are based on the premise that individuals consider various aspects of a choice alternative. The methods then permit a decomposition of an individual’s overall preference judgments about a set of choice alternatives into separate and compatible utility values corresponding to each attribute. These separate functions are called attribute-specific partworth functions. The choice-based conjoint methods (for stated choices) are based on the behavioral theory of random utility maximization. This approach decomposes an individual’s random utility for an object into two parts: deterministic utility and a random component. Depending on the distributional assumptions for the random component, a number of alternative models are developed to describe the probability of choice of an object. The most popular one is the multinomial logit model that uses the extreme value distribution for the random term. These methods belong to the family of discrete choice analysis methods.
The business settings of this work consist in a company operating in the market of a specific product - Christmas box. The effort goes to the application and description of a path/heuristic on using the choice-based conjoint analysis (CBCA) on data collected through a survey in order to have indications on what product the respondents choose and how to construct a better product in a given simulated market. In the following sections the steps of the work are described and developed. Firstly, a survey was designed with the aim of collecting the needed data. Some data cleaning and preparation has been performed.
Following, the analysis has consisted in considering two versions of the CBCA, the fixed effects multinomial (MNL) model and the mixed MNL one. In the next step, through these models, partworths/utilities of individual preferences has been estimated. Based on these estimates of partworths calculation of willingness to pay has been performed to have a more interpretable view. in addition, estimates of parthworth has served also the purpose of simulating preference share (through a market simulator/optimizer) and how a better product can be designed. Successively, some experiments have been conducted to explore effects of individual levels on the heterogeneity of respondents. Lastly, some conclusions and some ideas about future development have been drawn.