Applying Content Analysis Methodology to Social Media Efforts

Theoretical Background and research questions/hypothesis: Communities Putting Prevention to Work (CPPW) is a locally driven initiative that supported 50 communities to tackle obesity and tobacco use. Five infographics were created to disseminate core CPPW messages and support the efforts of local CPPW communities. Additionally, the infographics aimed to drive traffic to the “Making Health Easier” website. One of these infographics, “The New (Ab)Normal,” was launched in May 2012 and is estimated to have reached over 34 million unique individuals. The New (Ab)Normal focused on the increase in fast food portion sizes over the past several decades. The spread of the infographic was spurred by a targeted dissemination strategy and further viral sharing of the infographic on social media platforms including Twitter, Facebook, and Pinterest. This presentation will describe the application of content analysis methodology to examine online comments about The New (Ab)Normal. Content analysis was proposed as a supplement to standard sentiment analysis analytics (e.g., Radian6), and to more closely examine public reaction and online discourse generated through a mainstream outlet (Yahoo!). Specific objectives were to:

• Describe the content and sentiment of comments regarding the infographic’s overall message.
• Explore dialogue or “conversations” within comments.

Methods: During the two-week period following the infographic’s launch, a total of 1,533 comments were posted on Yahoo!, of which 216 were sampled for analysis. Comments were first coded regarding whether they supported the infographic’s main message. For items that were neither supportive or negative regarding the main message, coders noted whether the reaction was “tangential” (i.e., another overweight/obesity topic), or irrelevant. In addition, social media elements of each posting were noted, such as the number of “likes” and “dislikes.” Characteristics of subsequent responses to each posting were also captured.

Results: About half of the comments were coded as being either irrelevant (n=59, 27%), or only tangential to the main message of the infographic (n=51, 24%); the remainder were coded as being supportive of the infographic (n=29, 13%), oppositional (n=69, 32%), or mixed (n=8, 4%). Of the 216 comments, 35% (n=75) generated at least one reply; additionally, a total of 1,068 “likes” and 143 “dislikes” were generated by the comments.

Conclusions: The infographic was successful at generating conversation about obesity on a mainstream web outlet. Further, the large volume of comments tangential to overweight/obesity (but not about the infographic) suggests that, regardless of sentiment about the infographic itself, the presence of the infographic prompted conversation about related topics. While many comments were coded as irrelevant or “nuisance,” this is not unexpected considering that comments were not moderated, and users were not required to disclose their identities; this environment of low accountability and low barriers to participation may foster irrelevant conversation. 

Implications for research and/or practice: Content analysis is a useful tool to derive insights about social media efforts. Public health practitioners should consider the context of “negative” responses to social media efforts, particularly when generated via mainstream websites. Further, this reinforces the utility in consulting a variety of metrics—rather than only a sentiment score—in considering how social media items are perceived.