Comparing Twitter and Online Panels for Survey Recruitment of E-Cigarette Users and Smokers
Guillory, J., Kim, A., Murphy, J., Bradfield, B., Nonnemaker, J., & Hsieh, Y. (2016). Comparing Twitter and Online Panels for Survey Recruitment of E-Cigarette Users and Smokers. Journal of Medical Internet Research, 18(11), . DOI: 10.2196/jmir.6326
Background: E-cigarettes have rapidly increased in popularity in recent years, driven, at least in part, by marketing and word-of-mouth discussion on Twitter. Given the rapid proliferation of e-cigarettes, researchers need timely quantitative data from e-cigarette users and smokers who may see e-cigarettes as a cessation tool. Twitter provides an ideal platform for recruiting e-cigarette users and smokers who use Twitter. Online panels offer a second method of accessing this population, but they have been criticized for recruiting too few young adults, among whom e-cigarette use rates are highest.
Objective: This study compares effectiveness of recruiting Twitter users who are e-cigarette users and smokers who have never used e-cigarettes via Twitter to online panelists provided by Qualtrics and explores how users recruited differ by demographics, e-cigarette use, and social media use.
Methods: Participants were adults who had ever used e-cigarettes (n=278; male: 57.6%, 160/278; age: mean 34.26, SD 14.16 years) and smokers (n=102; male: 38.2%, 39/102; age: mean 42.80, SD 14.16 years) with public Twitter profiles. Participants were recruited via online panel (n=190) or promoted tweets using keyword targeting for e-cigarette users (n=190). Predictor variables were demographics (age, gender, education, race/ethnicity), e-cigarette use (eg, past 30-day e-cigarette use, e-cigarette puffs per day), social media use behaviors (eg, Twitter use frequency), and days to final survey completion from survey launch for Twitter versus panel. Recruitment method (Twitter, panel) was the dependent variable.
Results: Across the total sample, participants were recruited more quickly via Twitter (incidence rate ratio=1.30, P=.02) than panel. Compared with young adult e-cigarette users (age 18-24 years), e-cigarette users aged 25 to 34 years (OR 0.01, 95% CI 0.00-0.60, P=.03) and 35 to 44 years (OR 0.01, 95% CI 0.00-0.51, P=.02) were more likely to be recruited via Twitter than panel. Smokers aged 35 to 44 years were less likely than those aged 18 to 24 years to be recruited via Twitter than panel (35-44: OR 0.03, 95% CI 0.00-0.49, P=.01). E-cigarette users who reported a greater number of e-cigarette puffs per day were more likely to be recruited via Twitter than panel compared to those who reported fewer puffs per day (OR 1.12, 95% CI 1.05-1.20, P=.001). With each one-unit increase in Twitter usage, e-cigarette users were 9.55 times (95% CI 2.28-40.00, P=.002) and smokers were 4.91 times (95% CI 1.90-12.74, P=.001) as likely to be recruited via Twitter than panel.
Conclusions: Twitter ads were more time efficient than an online panel in recruiting e-cigarette users and smokers. In addition, Twitter provided access to younger adults, who were heavier users of e-cigarettes and Twitter. Recruiting via social media and online panel in combination offered access to a more diverse population of participants.