Whose Post Is It? Predicting E-cigarette Brand from Social Media Posts.
Tob Regul Sci
; 4(2): 30-43, 2018 Mar.
Article
en En
| MEDLINE
| ID: mdl-30662930
ABSTRACT
OBJECTIVES:
E-cigarette advertisers know that 76% of youth use social media, yet little is known about the nature of e-cigarette advertising on social media most favored by youth. We utilized text-mining to characterize e-cigarette advertising and marketing messages from image-focused social media brand sites, and to construct and test an algorithm for predicting brand from brand-generated social media posts.METHODS:
Data comprised 5022 unique posts accompanied by an image from Facebook, Instagram or Pinterest e-cigarette brand pages for Blu, Logic, Metro, and NJOY from February 2012 to April 2015. Text-tokenization was used to quantify text for use as predictors in analyses.RESULTS:
Blu had the largest social media presence (65%), followed by Logic (16%), NJOY (12%) and Metro (7%). Blu's average post length was significantly shorter than all other brands. Words most commonly used in posts differed by brand. Regression analyses successfully differentiated Blu and NJOY brands from other brands.CONCLUSIONS:
Analyses revealed e-cigarette brands used different types of messages to appeal to social media users. Whereas words used by Blu and NJOY sold a "lifestyle," words used by Logic and Metro relied on device and product identification.
Texto completo:
1
Banco de datos:
MEDLINE
Tipo de estudio:
Prognostic_studies
/
Risk_factors_studies
Idioma:
En
Revista:
Tob Regul Sci
Año:
2018
Tipo del documento:
Article