This article extends the work of prior researchers by empirically investiga
ting the general hypothesis that biases in sample sources are sufficient to
significantly alter the results of empirical research. Data were gathered
from firms listed in one or more of the following three sample sources: Dun
and Bradstreet's Regional Directory-Houston, the Directory of Texas Manufa
cturers, and the Texas State Sales Tax Files. A model of venture performanc
e was adopted based largely on previous research. Results of the analysis i
ndicate a strong presence of sample source bias in the data, despite the un
iform manner in which the data were gathered and the variables constructed.
(C) 2001 Elsevier Science Inc. All rights reserved.