The variables included in a multiple regression analysis may be influenced by
data availability, as well as by the underlying economic theory. Data for demand
estimation are often drawn from large-scale surveys undertaken by the federal
government, universities, nonprofit groups, industry or trade associations, and
company consumer research departments. In many cases, analysts would like to
include certain variables, but a consistent set of observations for all individuals in
the analysis may not be available. Some data sources may have better information
on economic variables, while others may have more data on personal characteristics
of the individuals included in the analysis. Analysts may also have to use other
variables as proxies for the variables of greatest interest.