Conference Papers 2019

Seminar
Day 2 (10 Oct 2019), Session 5, E-Commerce Logistics, 11:30 - 13:00

Status
Accepted, documents submitted

Submitted by / Abstract owner
Orlando Belcore

Authors
Orlando Marco Belcore
Massimo DI Gangi
Luigi Dell'Olio

Short abstract
A web-based survey, that combines an efficient design experiment (three steps Logit-model) and rating based conjoint tasks, has been developed in order to evaluate factors that affect e-grocery consumers’ attitude and to develop a provisional model.

Abstract
E-grocery, as a branch of e-commerce, is getting more attention among consumers. Even though sectorial studies report about a growth attractive market, penetration rates result substantially different even within any one country as a consequence of different policies and variety of supplies. Discrete Choice Models are widely used in transport economics; nevertheless, experimental design approach results not common in e-grocery studies. Even though stated preference experiments are commonly used to describe decisions, their main drawback is represented by low reliability in providing information about real behaviours. Moreover, users’ attitude towards a service represents itself an unobservable concept whose evaluation can be obtained, through variables that can be representative of the effect. So, the attitude it’s measurable by means of “latent variables” whose judgment can be expressed by consumers by means an ordinal scale task. Then, the purpose of this work is finding out if consumers’ choices, when they deal with foodstuff and hygiene products purchases, can be described simulating a complete cognitive process, through a survey, developing by such way an effective tool that intercepts consumers feeling and calibrate on its outputs a reliable provisional model.
Using the findings obtained from pilot stages, a web-based survey has been developed combining three fundamental stages; a revealed preference (RP) section; a stated preference (SP) tasks, developed by means an efficient experimental design generated on Ngene Software; and a rating based conjoint task on factors influencing consumers’ attitude, to be carried out using a 5° grade Likert scale.
The choice set developed, through the SP task, can be considered exhaustive since lets the interviewed to choose the strategy that best fits needs among offline, online and a multi-channel option (different mix of the previous ones). Moreover, in order to let people deal with a wider range of variables, solving simple tasks, a single scenario has been divided into three different steps, each one focused on a specific decision level here resumed as “cart composition and its value”; “accessibility conditions”; “delivery/collection options”. Each time it is asked to define the best option valuing which are the main variables that influence decision towards a specific strategy. As mentioned before, the SP task is followed by the conjoint rating one, where it is asked to the interviewees how different set of factors - such as convenience, risk related to the products and money, advantages offered by digital market and services offered by providers - affect their intention for online grocery shopping.
The parallel use of a stated preference and a rating based task allow us to capture real behaviours and observe preferences by means of latent class models. The choices from each SP step let us evaluate which variables affect more consumers' decisions towards a strategy and their statistical significance within the decisional process. Furthermore, the in-depth final analysis on attitude represents an opportunity to cluster consumers’ that belongs to heterogeneous populations. However, a disadvantage for this experiment lies in the heavy structure of the survey; indeed more than 15 minutes are necessary to complete all tasks for the longest tree.

Programme committee
Transport Models