"The researchers also note that Twitter users may not be representative of the general U.S. population."

Not only are Twitter users not representative of the entire US, but the Twitter users who consider vapes tweeting about are not representative of e-cig users as a whole. Sentiment analysis of these tweets tells us nothing about vaping in the large.

For example: consider someone who uses an e-cig to quit smoking, but has no particular affinity for vaping otherwise. Is this person likely to go on Twitter and post "I feel indifferent about vapes"? Obviously not.

If you design your study methods to select for hardcore vape fans, it should come as no surprise when you "discover" that these people tend to enjoy vaping and want to keep doing it.