How marketing vocabulary was evolving from 2005 to 2014? An illustrative application of statistical methods on text mining

📝 Abstract

Here a collection of 1169 abstracts, which corresponds to articles that the Journal of Marketing Research has published from 2005 to 2014, are analysed under a novel approach. We apply several statistical methods, such as Principal Components Analysis and Correspondence Analysis to identify the way Marketing vocabulary is evolving. Similarly those articles that introduce new vocabulary are identified and the preferred words by authors are also detected. In order to provide an easy-to-understand explanation, we provide our results graphically. A word-cloud with the most frequent words is given first. Secondly abstracts-words are represented on the factorial plane. Finally one representation of word-years allows us to detect changes on the vocabulary through the passing of time.

💡 Analysis

Here a collection of 1169 abstracts, which corresponds to articles that the Journal of Marketing Research has published from 2005 to 2014, are analysed under a novel approach. We apply several statistical methods, such as Principal Components Analysis and Correspondence Analysis to identify the way Marketing vocabulary is evolving. Similarly those articles that introduce new vocabulary are identified and the preferred words by authors are also detected. In order to provide an easy-to-understand explanation, we provide our results graphically. A word-cloud with the most frequent words is given first. Secondly abstracts-words are represented on the factorial plane. Finally one representation of word-years allows us to detect changes on the vocabulary through the passing of time.

📄 Content

• Corresponding author address: Av. Insurgentes Sur 1582, Col. Crédito Constructor . Delegación Benito Juárez C.P.: 03940, México, D.F. Telephone +52 55 5322-7700. email: jihbarahonato@conacyt.mx How marketing vocabulary was evolving from 2005 to 2014? An illustrative application of statistical methods on text mining

Igor Barahonaa*, Daría Micaela Hernándezb, Héctor Hugo Pérez-Villarrealc
a*Research Fellow - Cátedras CONACYT, México bTechnical University of Catalonia. Barcelona - Tech cPopular Autonomous University of Puebla State. Puebla, México.

Abstract Here a collection of 1169 abstracts, which corresponds to articles that the Journal of Marketing Research has published from 2005 to 2014, are analysed under a novel approach. We apply several statistical methods, such as Principal Components Analysis and Correspondence Analysis to identify the way Marketing vocabulary is evolving. Similarly those articles that introduce new vocabulary are identified and the preferred words by authors are also detected. In order to provide an easy-to-understand explanation, we provide our results graphically. A word-cloud with the most frequent words is given first. Secondly abstracts-words are represented on the factorial plane. Finally one representation of word-years allows us to detect changes on the vocabulary through the passing of time.

Keywords: Marketing, Textual Statistics, Vocabulary evolving, Influential articles, Correspondence analysis.

1. Introduction The increasing complexity on worldwide business environment, which is basically characterized by the globalization of the markets; the emergence of more powerful computers, accumulated data, proliferation of real time communication channels and social networks, are radically transformed the way organizations make decisions. According with Tuner et. al (2014) the digital data generated worldwide is growing 40% each year into the next decade. It will be expanding to include not only the increasing number of persons connected to Internet, but also the number of connected devices, as smart phones, tables, TV’s and vehicles. For instance, from 2013 to 2020, the accumulated data will grow by a factor of 10. This is from 4.4 trillion gigabytes to 44 trillion. Tuner et. al (2014) also estimated that in 2013, around the 22% of accumulated data is suitable for analysis through statistical methods. This makes contrast with the 35% that is expected to be reached on 2020, mostly due to the growth of data generated from embedded systems and real time applications. It draws our attention that the owners of nearly 80% of this “digital universe” will be private companies as Google, Twitter and Facebook. In this respect, countries governments will be required to introduce regulations, in order to successfully cope with issues as privacy, security and copyrights. JSM2015 - Section on Statistics in Marketing 169

Given this massive growth of digital data, it is clear that effective methods are needed in order to successfully extract knowledge from data. In the case of academic environment, methods are required to effectively perform literature reviews. According with Cooper (1998), the process of literature review is understood as identifying, assessing and synthesizing the best-available empirical evidence to answer a set of research questions. The foregoing implies that an effective literature review should identify all relevant articles and exclude irrelevant ones. Traditionally, researchers used to conduct literature reviews manually, by analysing and classifying one-by-one each document. Now, traditional methods are becoming inoperative as the available amount of articles increases. New approaches, based on information technologies and statistics are need. Given this scenario, it is clear that organizations must act promptly, if they are willing to ensure the leadership. Competitive advantages will be given to such organizations, which are able to extract useful information from big-data and make more accurate decisions. The evidence based management (EBMgt), introduced by Rousseau (2006), makes clear the importance of analysing data to improve organization’s performance. According with this author, EBMgt is the discipline which applies the scientific method principles to make better decisions. Similarly, data mining (DM) is another discipline that emerges as response of data accumulated. According with Witten & Frank (2005) the main purpose of DM is to extract useful knowledge from raw data. Text mining (TM) is a branch of the DM, which is focused on collecting, cleaning and processing text documents. Delen & Crossland (2008) introduced the application of TM on scientific texts, with the purpose of supporting researchers to conduct literature reviews, and consequently identify trends and topics on a given research subject. Although their research is f

This content is AI-processed based on ArXiv data.