Socialised promotions

Fairly recently a whole ‘new study’ saw the light of day in the Tobacco Control Journal but it didn’t seem to get the fanfare of scary headlines that most get, mostly because the “research” is way too picky. Not to mention it is “the first of its kind”. Natch.

So what was this “study” all about then?

Price-related promotions for tobacco products on Twitter

At first glance you could be forgiven if you thought it was all about traditional tobacco products, cigars, cigarettes and the like, but you would in fact be completely wrong.

From the abstract:

E-cigarettes were the most frequently mentioned product (90.1%),

Yes, release that sigh because they’ve included vapour products in with tobacco. Again. But let’s rewind a bit first.

The Family Smoking Prevention and Tobacco Control Act (FSPTCA) implemented price-related restrictions for regulated tobacco products, that is, cigarettes, cigarette tobacco, roll-your-own tobacco and smokeless tobacco

This Act bans mail-order coupons and free samples of cigarettes. It also prohibits companies from offering branded “non-tobacco” promo items in exchange for buying cigarettes. Mostly, this is relatively reasonable until you note that you can’t exchange “non-tobacco” items for smokeless tobacco either. Score one for the prohibitionists!

So what did these ‘researchers’ actually do to get their data? Social marketing at its finest here. Twitter has an interface they call the “Firehose” that allows entities to view ALL tweets in real-time regardless of whether or not you follow the user. Think about that for a second, Twitter has tens of millions of users worldwide. At any one time there are likely to be millions on-line and tweeting. That’s literally millions of tweets per minute. Twitter is huge.

These researchers looked at all tweets posted between the 6th December 2012 and 20th June 2013 based around a series of specific keywords and pulled 35,373,122 tweets. 35 million tweets all based around specific keywords, like smoking, cigarette, e-cigarette, and tobacco. In six months, that’s 5,895,520 tweets per month or 196,517 tweets per day using those keywords alone. Those figures are actually staggering, and that was two years ago can you imagine the daily tweet figures now? Double? Triple? No wonder Governments are ‘concerned’!

Utilising additional filtering methods, the researchers added promotional keywords to the ‘potentially tobacco related tweets’ giving them 155,249. Out of 35 million, less than 1% (0.44%) of the total tweets gathered were analysed for a six month period. Not exactly huge figures considering the time frame and the total number of tweets that matched the initial filter keywords.

The researchers went further, they created two distinct groupings based on whether the tweet contained a price-related keyword, including abbreviated and plural versions, and whether the tweet contained multiple price-related keywords. In true form, any tweet that wasn’t in English, unrelated to a tobacco or cessation price promotion were excluded. Out of the 155,249 tweets originally filtered, the researchers were left with 2847 using a random sampling method out of the two groups and applying the final filter of non-English and unrelated tweets.

To be fair, 2847 tweets is far more manageable for analysis than 155K and infinitely more manageable than 35 million, from an analysis point of view I certainly wouldn’t want to be going through 35 million tweets and applying a specific code to each one. Muchero mundano.

Measurements. What coding did they use to segment the tweets? Only three segments (domains):

  1. Product Type
  2. Promotion Type
  3. Tweet Source (commercial or personal user)

Interestingly, the researchers classified the product types:

  • Cessation
  • Cigarettes
  • Cigars
  • E-Cigarettes (Vapour products)
  • Smokeless Tobacco

I really did wonder why they included cessation products in this analysis but it did become clear. In the “tweet source” domain; where links to websites were available, if the link went straight to a vendor it was classed as “commercial”.

Tweets that directly or indirectly linked to tobacco or cessation vendor websites were coded as ‘commercial’. Commercial tweets linked to three categories of websites, which we tracked to capture the range of sites through which products are promoted: tobacco or cessation vendor sites, landing sites and cross-industry promotional sites.

Not only did the researchers bundle vapour products up as a tobacco product, they now bundled cessation up with tobacco retailers. Classy folks.

Interesting side-bar here, out of the 2847 tweets they looked at, there were only 2102 unique Twitter IDs with an average of 3951 followers; amusingly the “range” of followers was very wide – from 1 (yes they looked at tweets from users with only one follower) to 320,902 (who wants to be that popular seriously?!).

Unsurprisingly, e-cigarettes were the most commonly mentioned product (2564 of the 2847 tweets) with “device” coming in second place with 2274. Unusually, “nicotine free” e-cigarettes didn’t score. Cessation products only had 86 tweets, a lowly 3% which tells me that twitter is not the place to be if you are looking for help with cessation. Shocker.

What is kind of interesting is the number of cigarette related tweets, 153 of them to be exact which is in direct contravention of the FSPCTA Act. Even more interesting, they don’t classify any product as a “tobacco harm reduction” product, such as Smokeless or vapour products as two prime examples. But then, it wouldn’t fit with the study title would it?

As a minor-league geek, I found this table to be quite interesting as it shows the type of promotions that were going on ranging from a simple discount to free samples, including the typical buy one get one free (BOGOF) offers or getting a free gift (drip tip? ).

So what is the upshot of this “study”? Well, in the authors own words (emphasis mine):

This study is the first to characterise tobacco-related price promotions on Twitter. As in other research, the vast majority of tweets were commercial in nature. Our data indicate that price promotions for tobacco products regulated by the FDA (ie, cigarettes, cigarette tobacco, roll-your-own tobacco, smokeless tobacco) are far less common than price promotions for ecigarettes.
In our sample, we saw very few promotions for FDA-regulated tobacco products.

That one sentence, in my potentially misguided opinion is the main reason for undertaking such a study. How much are the FDA and other Governmental Agencies losing out on? The answer is pretty clear-cut, vapour products get far more attention on social media (promotional material not withstanding) than the traditional counterparts.

Pharmaceutical companies also appear to market coupons via direct tweets to users, particularly those who mention quitting smoking.

Nothing like a small nod to the paymaster is there? But what is this?

Our study has several limitations. First, location information was not available for tweets in our sample, so we cannot be certain that they originated from within the USA, which is covered by the FSPTCA, as opposed to another country.

The study authors admit to limitations, so the faff about FSPTCA may not even be applicable to some of these tweets. Well duh!

Second, our sample may be limited by the keywords used to collect tweets. Although we included a number of e-cigarette brand names in our list of keywords, we did not include brand names for other tobacco products

Freely admitting that their keywords were rubbish (reading between the lines) and instead specifically searching for e-cigarette brand names, most of which are related to products created by, dum dum daah, the tobacco industry.

Absolutely shocking, they weren’t getting the info they needed so they deliberately skew results by searching for known tobacco e-cigarette brand names.  All this “study” is telling me is these researchers haven’t got a clue what an “e-cigarette” actually is, which is hardly surprising.

Related Post