LATEST ALERTS
E-cigarette activity on Twitter has grown less positive over time and is heavily dominated by automated content, according to new academic research.
The study, from scholars at the University of Vermont led by Eric M. Clark, used statistical models and machine-learning programs to analyse around 850,000 e-cigarette-related tweets posted between January 2012 and December 2014.
It found that content generated by people – as opposed to software – showed an increase in negative sentiment as time went on. This does not necessarily mean users became more negative about e-cigs as a product, but there was an increase in negative expression about a range of issues, according to the researchers.
In particular, the impact of research on youth vaping as well as proposed legislation including e-cigs in public smoking bans drove down the level of happiness around e-cigs on Twitter.
“This decrease in the average happiness score is due to a relative increase in the negative words ‘ban’, ‘tobacco’, ‘doesn’t’, ‘drug’, ‘against’, ‘poison’, ‘tax’ [and] a relative decrease in the positive words ‘haha’, ‘good’, ‘cool’,” the academics said. “Notably, there is also relatively less usage of the words ‘quit’, ‘addicted’, and an increase in ‘health’, ‘kids’, ‘juice’.”
Rise of the machines
However, the research also found that the vast majority (80%) of Twitter traffic relating to e-cigs was automatically-generated commercial content. A third of this offered discounts or free samples, with a large amount referencing e-cigarettes as a smoking cessation product.
Unlike the organically generated material, the automatic content was so positive in outlook and sentiment it skewed the overall results before being removed. Brands such as V2, Apollo and Blu featured heavily in the automated marketing, the researchers added.
The study also calculated that, although the total number of automated tweets went down over the course of the research period, the total number of potential views of each tweet went up.
“The total number of impressions from the commercial category increases from 195.25 million to 951.03 million between 2013 to 2014, even though the total count has dropped from 283,000 to 149,00,” the paper said. “This implies that promotional accounts that are successful in deceiving Twitter’s SPAM detector may be gaining many more social links to broadcast their commercial context.”
The research team noted a growing trend around the use of automated tweet-bots that effectively mimic human activity. “These messages are very intentionally structured and tend to swap a few words to appear organic. These messages also target specific individuals as a more personal form of marketing,” the researchers said.
For example, a tweet might be programmed as: “@USER {I, We} {tried, pursued} to {give up, quit} smoking. Discovered BRAND electronic cigarettes and quit in {#} weeks. {Marvelous, Amazing, Terrific}! URL.”
This should be an area of priority research in the future in order to better understand the effectiveness of this marketing technique, the researchers said.
Their study, “Vaporous marketing: uncovering pervasive electronic cigarette advertisements on Twitter”, is published in PLOS ONE.
What This Means: It is clear that much of the content on Twitter related to e-cigs is auto-generated material that is little more than spam, but this is probably true for any fast-moving consumer good – auto-tweets are a significantly larger proportion of Twitter than marketers would like known.
Equally, it is no surprise that even e-cig proponents on Twitter would grow increasingly negative as more bans are instituted and more questionable studies are released.
So while there are no great surprises here, the tools and statistical models developed by these researchers could prove to be useful in the future. The ability to automatically categorise and sort tweets could have a number of academic and commercial applications, not least in telling us how things have developed on this front over the past 18 months.
– Freddie Dawson ECigIntelligence staff
Photo: Johnny Silvercloud
