Leveraging Social Media in the Pharmaceutical Industry - Innovation and Marketing in the Pharmaceutical Industry - page 485

Table 16.2 (continued)

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

Berger et al.
( 2010 )

Time-series model
and ANOVA

Sales and New York Times
reviews of 244 titles and two
experiments

Negative publicity may increase the sales
of products with low existing aware-
ness, by increasing awareness or
accessibility

Does not control for extremes of reviews

Van der Lans et al.
( 2010 )

Viral branching model A viral campaign to promote

inancial services to 228,351
participants

The reach of eWOM can be accurately
predicted by a viral branching model
that also accounts for marketing
activities

Reach of WOM may not be a signiicant predictor
of sales. Does not include quality of reach

Nam et al. ( 2010 ) Hazard model 3,650 adopters of Video-on-

Demand (VOD) movie rental
service. Individual-level
activation, usage, distance to
retail and rental stores data
and aggregate demographic
data

The effect of negative word of mouth on
the adoption of VOD service is more
than twice as large as the effect of
positive word of mouth

The truncated hazard model is based on only
adopters, potentially leading to spurious
evidence of contagion

Schmitt et al.
( 2011 )

Linear panel data
model with ixed
effects. Hazard
model.

Panel data on 9,814 customers
acquired by a leading
German bank through its
referral program and other
methods

Customers generated through referral
programs have a higher contribution
margin and a higher retention rate, and
are more valuable in both the short and
the long run than those acquired from
non-referral methods

Customer lifetime value and effectiveness of
referral programs could be a function of
product complexity and incentives

Rui et al. ( 2011 ) Multivariate VAR
model and
Dynamic panel
data model

Movie revenues and Twitter
tweets on 63 movies

WOM on Twitter is a signiicant determi-
nant of box ofice revenues

Omits movie characteristics in the models

Sonnier et al.
( 2011 )

Bayesian dynamic
linear model
(DLM)

Daily sales revenue data and
daily counts of positive,
negative, and neutral online
comments for the irm

Positive, negative, and neutral communica-
tions have signiicant effects on daily
sales performance after controlling for
dynamics and endogeneity

Omits marketing efforts and seasonality