Every multiple regression analysis is influenced by the sample of data—timeseries,
cross-sectional, or panel—that is used. The analyst wants to estimate behavioral
relationships that can be generalized beyond the sample of observations
included in the analysis. Yet large-scale data collection can be very expensive and
time-consuming. Thus, the analyst must be concerned that the estimated relationships
may hold only for the sample of data analyzed, and not the larger population.
As we discussed above, the analyst engages in hypothesis testing to determine how
much confidence can be placed in the results of a particular analysis and whether
these outcomes can be generalized to a larger population.