Pharmaceutical Detailing Elasticities: A Meta-Analysis - Innovation and Marketing in the Pharmaceutical Industry - page 526

18.2.1 Database Description

Our database spans the last 5 decades, and the USA and Europe (Belgium, France,
Germany, Italy, Netherlands, Portugal, United Kingdom) regions, encompassing a
wide range of sales-environments. The database (see Table 18.1 ) includes relevant
papers from multiple disciplines (marketing, management, operations research,
economics, and health economics) as well as industry- and government-based
studies. A paper is a distinct document offering some original analysis fi nding/s
by its author/s—this rules out duplications or redundant papers in the database
(see, e.g., Wood 2008 ). Collectively, these papers provide analyses of many
distinct datasets , each containing information about sales response to detailing
effort in some specifi c market setting. If a different estimation technique / model is
applied to the same dataset in either the same paper (e.g., Berndt et al. 1995 ) or
different papers (e.g., Berndt et al. 2003 ), we treat the resulting elasticity observa-
tions as multiple distinct measurements from one dataset . Conversely, one paper
may provide analyses of multiple distinct datasets , contributing one (or more)
distinct detailing elasticity estimates from each dataset (e.g., Narayanan et al.
2004 ). Applying these defi nitions, our database includes 48 research papers that
use 44 distinct datasets , providing a total of 373 detailing elasticity measurements
(see Table 18.1 ).

18.3 Methodology

18.3.1 Database Compilation

We compiled our database using a variety of sources as in AMS ( 2010 ). These
include: (1) All relevant sales force models review articles , e.g., Albers and
Mantrala ( 2008 ), Vandenbosch and Weinberg ( 1993 ) and references they cite;
(2) a number of computerized publication search services such as Proquest from
ABI / Inform , Business Source Premier from EBSCO , Kluwer Online ); (3) all rel-
evant working papers posted on the Web, e.g., those on Social Science Research
Network ; (4) reports of relevant consulting engagements from prominent schol-
ars; (5) archives of technical reports and/or working papers of the Marketing
Science Institute , the Institute for the Study of Business Markets , and leading
business schools; and (6) responses to a request for unpublished works posted on
the marketing network [e-mail]. Inclusion of unpublished works
helps avoid publication bias that would reduce measurement variability in the
meta-analysis (e.g., Assmus et al. 1984 , p. 66; Andrews and Franke 1991 , p. 83;
Tellis 1988 , p. 340; Rust et al. 1990 ).