Experimental and quasi-experimental designs can be
difficult to implement because randomization to the
intervention—critical to avoid selection bias—and even
recruitment of participants, is extremely difficult. Dansky,
Thompson, and Sanner (2006) note that mailings to providers, community organizations, and other gatekeepers
or radio or other traditional media are often used to meet
recruitment goals. However, because by design many social
media tools allow consumers anonymous access and use,
follow-up and linkage are often not viable.
Identification of effects of specific interventions also
presents problems because social media is often used
as one of several health promotion tools. For example,
consumers often use health information websites as an
adjunct to visits with their provider, sharing information with friends and family or using other social or
traditional media outlets, making it difficult to determine the relative influence of each.