Leveraging Social Media in the Pharmaceutical Industry - Innovation and Marketing in the Pharmaceutical Industry - page 487

Table 16.2 (continued)

Article/work Model type Data Key indings/guidelines Key limitations

Iyengar et al.
( 2011a )

Hazard model 185 doctors in three cities with

individual level data for
adoption, calls, demo-
graphic, self-reported
leadership, network data on
discussion, and patient
referral ties among
physicians supplemented
with co-location measures

The pattern of network ties is only weakly
associated with co-location of
physicians. Contagion operates through
social networks, likely driven by social
learning mitigating functional and
physical risks, as well as through
co-location, likely driven by social-
normative pressures

Unable to distinguish co-location contagion from
heterogeneity in time-invariant workplace
characteristics

Iyengar et al.
( 2011b )

Hazard model 185 doctors in three cities with

individual-level data for
adoption, calls, demo-
graphic, self-reported
leadership, network data on
discussion, and patient
referral ties among
physicians

Social contagion exists in new product
adoption after controlling for marketing
efforts and system-wide changes.
Opinion leadership and sources' volume
of product usage moderates social
contagion effects on new product
adoption

Low response rates from surveyed doctors. Small
sample size

Van den Bulte and
Iyengar ( 2011 )

Hazard model Simulated data, medical

innovation data, and zip
code-level adoption data
from Netgrocer.com

Hazard models built on right-truncated data
can induce spurious positive duration
dependence that suggests social
contagion exists when it doesn't

Assume positive duration dependence implies
social contagion, but it could relect other
constructs