INTRODUCTION: As media exposure can influence people's opinions and perceptions about vaping and smoking, analyzing the valence of media content about tobacco products (ie, overall attitude toward tobacco, cigars, electronic cigarettes, etc.) is an important issue. This study advances the field by analyzing a large amount of media content about multiple tobacco products across six different media sources. AIMS AND METHODS: From May 2014 to December 2017, we collected all English-language media items about tobacco products that U.S. young people might see from mass media and websites (long-form) and social media (Twitter and YouTube). We used supervised machine learning to develop validated algorithms to label the valence of these media items. Using the labeled results, we examined the impact of product type (e-cigarettes vs. other tobacco products), source (long-form vs. social media), and time (by month) on the valence of coverage. RESULTS: We obtained 152 886 long-form media texts (20% with more than a passing mention), nearly 86 million tweets, and 12 262 YouTube videos about tobacco products. Most long-form media content opposed, while most social media coverage supported, the use of e-cigarettes and other tobacco products. Over time, within-source valence proportions were stable, though in aggregate, the amount of media coverage against the use of tobacco products decreased. CONCLUSIONS: This study describes the U.S. public communication environment about vaping and smoking for young people and offers a novel big data approach to analyzing media content. Results suggest that content has gradually become less negative toward the use of e-cigarettes and other tobacco products. IMPLICATIONS: This study is the first to examine how the valence of media coverage differs for e-cigarettes versus other tobacco products, across several media sources, and over time using a large corpus of media items. Unlike prior studies, these data allow us to draw conclusions about relative support and opposition for these two categories of products in a variety of media coverage because the same coding scheme was used across products and media sources.
INTRODUCTION: As media exposure can influence people's opinions and perceptions about vaping and smoking, analyzing the valence of media content about tobacco products (ie, overall attitude toward tobacco, cigars, electronic cigarettes, etc.) is an important issue. This study advances the field by analyzing a large amount of media content about multiple tobacco products across six different media sources. AIMS AND METHODS: From May 2014 to December 2017, we collected all English-language media items about tobacco products that U.S. young people might see from mass media and websites (long-form) and social media (Twitter and YouTube). We used supervised machine learning to develop validated algorithms to label the valence of these media items. Using the labeled results, we examined the impact of product type (e-cigarettes vs. other tobacco products), source (long-form vs. social media), and time (by month) on the valence of coverage. RESULTS: We obtained 152 886 long-form media texts (20% with more than a passing mention), nearly 86 million tweets, and 12 262 YouTube videos about tobacco products. Most long-form media content opposed, while most social media coverage supported, the use of e-cigarettes and other tobacco products. Over time, within-source valence proportions were stable, though in aggregate, the amount of media coverage against the use of tobacco products decreased. CONCLUSIONS: This study describes the U.S. public communication environment about vaping and smoking for young people and offers a novel big data approach to analyzing media content. Results suggest that content has gradually become less negative toward the use of e-cigarettes and other tobacco products. IMPLICATIONS: This study is the first to examine how the valence of media coverage differs for e-cigarettes versus other tobacco products, across several media sources, and over time using a large corpus of media items. Unlike prior studies, these data allow us to draw conclusions about relative support and opposition for these two categories of products in a variety of media coverage because the same coding scheme was used across products and media sources.
Authors: K Clegg Smith; M A Wakefield; Y Terry-McElrath; F J Chaloupka; B Flay; L Johnston; A Saba; C Siebel Journal: Tob Control Date: 2008-02 Impact factor: 7.552
Authors: Allison J Lazard; Gary B Wilcox; Hannah M Tuttle; Elizabeth M Glowacki; Jessica Pikowski Journal: Tob Control Date: 2017-03-24 Impact factor: 7.552
Authors: Allison E Myers; Brian G Southwell; Kurt M Ribisl; Sarah Moreland-Russell; Leslie A Lytle Journal: Tob Control Date: 2016-07-13 Impact factor: 7.552
Authors: Olivia A Wackowski; Daniel P Giovenco; Binu Singh; M Jane Lewis; Michael B Steinberg; Cristine D Delnevo Journal: Nicotine Tob Res Date: 2018-07-09 Impact factor: 4.244
Authors: Jan van der Tempel; Aliya Noormohamed; Robert Schwartz; Cameron Norman; Muhannad Malas; Laurie Zawertailo Journal: Int J Public Health Date: 2016-02-03 Impact factor: 3.380
Authors: Shyanika W Rose; Catherine L Jo; Steven Binns; Melissa Buenger; Sherry Emery; Kurt M Ribisl Journal: J Med Internet Res Date: 2017-02-27 Impact factor: 5.428
Authors: Robert Hornik; Steven Binns; Sherry Emery; Veronica Maidel Epstein; Michelle Jeong; Kwanho Kim; Yoonsang Kim; Elissa C Kranzler; Emma Jesch; Stella Juhyun Lee; Allyson V Levin; Jiaying Liu; Matthew B O'Donnell; Leeann Siegel; Hy Tran; Sharon Williams; Qinghua Yang; Laura A Gibson Journal: J Commun Date: 2022-01-13