Finding good, generalizable samples has long been an Achilles heel for
those interested in entrepreneurship-related phenomena. Practitioners
, policy makers, and researchers frequently desire to take the pulse o
f entrepreuneurial activity but struggle in doing so. For example, Fri
edman (1995) reported that California overestimated job loss by 2.4 ti
mes because of failure to account for newer, more robust firms. With t
he goal of being able to better identify new businesses, this research
addresses the following questions: (I) Which sample sources best iden
tify new businesses? (2) Are the characteristics of a sample drawn fro
m one source similar to those drawn from a different source? (3) How a
ccurate are the different sample sources in the information they repor
t? and (4) How practical and cost effective are the different sample s
ources? We pursued answers to these questions by examining the state s
ales tax file and comparing it with the previously examined sources of
the state unemployment insurance file (ES202) and Dun & Bradstreet's
Dun's Market Indicator file (DMI) (Birley 1984; Aldrich et al. 1989).
The phonebook was also used to examine its overlap with the three prim
ary sources. In making these comparisons, we evaluate generalizability
issues, accuracy of founding date, cost, practicality, and other auxi
liary information.