- 38 -
surveys: (1) The 1996/1997 Watson Wyatt Data Services survey of
over 1,700 companies throughout the nation in various industries
(Watson Wyatt survey) (which survey according to Packard’s report
included an unspecified number of companies in the construction
industry); (2) the 1995 Conference Board, Inc., survey of over
1,000 private and public companies throughout the nation in
various industries (Conference Board survey) (which survey
covered 12 companies in the construction industry--nine of which
had 1994 sales of $200 million or more and three of which had
1994 sales of $199 million or less); and (3) his own survey of
seven residential homebuilding companies. Mr. Packard then
applied regression analysis24 to the survey sample data (either by
the Watson Wyatt or Conference Board surveys themselves or by Mr.
Packard) to calculate the mathematically indicated relationship
between two chosen factors (e.g., annual sales of each company
24As explained in an excerpt from the Watson Wyatt survey
attached to Mr. Packard’s report, regression analysis is a
statistical analytical technique that examines the relationship
between two selected variables (e.g., compensation and sales
volume). The data from each surveyed organization is plotted on
a graph. For instance, a company’s sales volume is measured
along the horizontal (x) axis, and compensation is measured along
the vertical (y) axis. Assuming sufficient data is available to
provide a realistic picture of the relationship between
compensation and sales volume, a regression formula using the
“least squares method” is used to calculate the mathematical
equation that provides the “best fit” to the data. The resulting
equation (y = bx + a) can then be used to estimate the y for any
value of x within the range of the model. The appropriate range
of the model is determined by the data on which it is estimated.
Page: Previous 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 NextLast modified: May 25, 2011