{
  "id": "state-level-demographics-and-tobacco-control-correlates-of-smoking-cessation-behavioral",
  "type": "paper-conference",
  "title": "State–level demographics and tobacco–control correlates of smoking cessation behavioral change techniques on Twitter",
  "author": [
    {
      "family": "Toukhy",
      "given": "S."
    },
    {
      "family": "Vargo",
      "given": "C."
    },
    {
      "family": "Hopp",
      "given": "T."
    },
    {
      "family": "Choi",
      "given": "T."
    }
  ],
  "URL": "https://chrisjvargo.com/publications/state-level-demographics-and-tobacco-control-correlates-of-smoking-cessation-behavioral/",
  "issued": {
    "date-parts": [
      [
        2018
      ]
    ]
  },
  "container-title": "Paper presented at the 2018 Society for Research on Nicotine and Tobacco Annual Meeting, Baltimore, MD",
  "abstract": "Objective: Identify behavioral change techniques for smoking cessation reflected in discourse on Twitter and explore state-level demographic composition, smoking prevalence, quit rates, smoking-related health conditions, cigarette prices, and tobacco-control expenditures that are associated with smoking cessation tweets volume. Methods: Using Social Studio Radian 6 application programming interface (API), we retrieved tweets containing smoking-related keywords (eg, smoke) from 1/1/2009 to 12/7/2015. We developed a codebook based on Michie et al. (2011) taxonomy for smoking cessation behavioral change techniques. Two coders manually annotated 5715 random tweets, which were then used to build a machine learning algorithm using LightSide. We used geo-tags or self-report location to geocode tweets to the states using Google Maps Places API. We retrieved state-level data from Centers for Disease Control and Prevention and census data. A least absolute shrinkage and selection operator regression model was used to identify correlates between tweet volume and state-level variables. Results: A total of 1,431,790 tweets that included behavioral change techniques for smoking cessation were retained for analysis. We identified 10 techniques that fell under four categories: motivation, self- regulatory capacity, adjuvant activities, and information gathering. Variance explained in tweet vol"
}