Modeling controversies in the press: the case of the abnormal bees' death

TL;DR

This paper analyzes the evolution of media opinions on the causes of bee colony deaths in France from 1998 to 2010, using textual analysis and opinion dynamics modeling to reveal underlying pressures and influences.

Contribution

It introduces a novel combination of textual data analysis and Galam's opinion dynamic model to study media controversy evolution over time.

Findings

Identified three distinct states of opinion regarding bee deaths.

Modeled opinion dynamics with inflexible and flexible journalists, matching observed data.

Revealed influence of lobbying and early whistleblowers on opinion shifts.

Abstract

The controversy about the cause(s) of abnormal death of bee colonies in France is investigated through an extensive analysis of the french speaking press. A statistical analysis of textual data is first performed on the lexicon used by journalists to describe the facts and to present associated informations during the period 1998-2010. Three states are identified to explain the phenomenon. The first state asserts a unique cause, the second one focuses on multifactor causes and the third one states the absence of current proof. Assigning each article to one of the three states, we are able to follow the associated opinion dynamics among the journalists over 13 years. Then, we apply the Galam sequential probabilistic model of opinion dynamic to those data. Assuming journalists are either open mind or inflexible about their respective opinions, the results are reproduced precisely provided we account for a series of annual changes in the proportions of respective inflexibles. The results shed a new counter intuitive light on the various pressure supposed to apply on the journalists by either chemical industries or beekeepers and experts or politicians. The obtained dynamics of respective inflexibles shows the possible effect of lobbying, the inertia of the debate and the net advantage gained by the first whistleblowers.