1. To what extent do you agree with the following statements? a. The best test of the performance of two different regression equations is their respective values of R2. b. Time-series regressions should be run using as many years of data as possible; more data means more reliable coefficient estimates. c. Including additional variables (even if they lack individual significance) does no harm and might raise R2. d. Equations that perform well in explaining past data are likely to generate accurate forecasts.

2. ) A water expert was asked whether increased water consumption in a California community was lowering its water table. To answer this question, the expert estimated a linear regression equation of the form:

W = a + bt

where W = height of the water table and t = time measured from the start of the study period. (He used 10 years of water-table measurements.) The estimate for b was b = - 0.4 with a t-value of - 1.4. a. From this evidence, would you conclude that the water table was falling? b. A second expert suggests yearly rainfall also may affect the water table. The first expert agrees but argues that total rainfall fluctuates randomly from year to year. Rainy years would cancel out dry years and would not affect the results of the regression. Do you agree?