1)Which is not the correct assumption regarding the error term?

a)Expected value of error term should be 0

b)Variance of the error term should be constant .

c)Covariance between error terms should be 0 .

d)Errors should all be very small .

2)Which of the following would indicate a perfect model fit and perfect correlation between the variables ?

a) R2=1 .

b)R2=0 .

c)Durbin bin - Watson = 0 0.5 .

d)Durbin bin - Watson = 4.0 .

3)The significance level reported in standard statistical packages for the estimated slope coefficient for the regression relates to which null hypothesis ?

a)HO : Beta = 0

b)HO : Beta not = 0

c) HO : Beta > 0

d)HO : Beta < 0 .

4)Which of the following is correct ?

a)R2 is a not used that much in selecting a model .

b)Forecasters should base model selection criteria on the minimization of R2 .

c)Adjusted R2 is always much greater than R2 .

d) Forecasters should base model selection criteria on the maximization of R2

5)Testing for the individual statistical significance of the relationship between the dependent and independent variable

a) is accomplished by application of the F - distribution for small samples .

b) is accomplished by application of the t -distribution for small samples .

c) is accomplished by application of the Chi - Squared - distribution for small samples .

d)can't be accomplished .

6)The Y -intercept of a regression line is -14 and the slope is 3.5 . Which of the following is correct ?

a)When X increases by one, increases by 3.5.

b)When Y increases by one , X increases by 3.5.

c)The intercept is 3.5 .

d)X and Y are negatively related .

7)The least squares procedure minimizes the sum of

a)the residuals

b)cubed maximum error .

c)absolute errors .

d)squared residuals .

8)What would be my forecast for 4th quarters , given the below equation ? Yhat=100-5Q1 dummy-2 Q2 dummy-1 Q3,di Q3_dummy .

Can't be determined.

a)98

b)92

c)95

d)100

9)What would be my forecast for 1st quarters , given the below equation ? Yhat = 100 - 5 Q1_dummy - 2 Q2_dummy - Q3_dummy .

Can't be determined

a)100

b)95

c)98

d) 92

10)Which one of the following is incorrect regarding nonlinearity within CLRM ?

a)The resulting error terms will be nonlinear .

b)The forecasts will be inaccurate .

c)We need to transform the X variable to change the functional form .

d)We can detect the problem using a scatter plot .

e)Nonlinearity is really not a huge problem , we can ignore it .

11)Given the below regression model ( assume everything is significant ), what is the forecasted MPG for a 10 year old car whose driver is a male ? MPG = Miles per gallon FI = Fem indicator (takes the value of 1 if the individual is a female , if male ) AGE = age of the car

MPG=30+2.0^ * FI-1.0^ * AGE

a)25

b) 20

c) 27

d) 22

12)How can we try fix a possible Autocorrelation problem ?

a)By adding the lags of error term .

b)By adding the lags of the explanatory variables .

c) By adding the lags of the y variable . By taking the log of the explanatory variables .

d) All of the above should work .

13Which one of the following is used to check the overall significance of the model ?

a)VIF

b) F - test

c)Correlation matrix

d)DW test

e)t - test

13)Which of the following would be the only reason to eliminate explanatory variables ? a)Nonlinearity

b) Multicollinearity

c)Sign switch

d)Autocorrelation T

e)here is never any good reason to eliminate information.

14)Cross-sectional regression models of personal income to consumption are likely to have the following problem

a)nonlinearity .

b)homoscedasticity .

c)autocorrelation .

d)Heteroskedasticity .

15)Which of the following is NOT an assumption of multiple regression models ?

a) All Xs should be measured should be continuous scale ; they can't be discrete .

b)A linear relationship exists between each X variable and Y.

c)The error terms should be independent of each other .

d)X variables should not be very highly correlated with each other .

16Which of the following is correct regarding the relation between the R -squared and the correlation coefficient between the X and Y variable ?

a)if x and Y are nonlinearly related , the correlation coefficient will be negative , but R -squared will always be positive .

b)If X and Y are negatively related, the correlation coefficient will be negative, so will the R-squared .

c) Since they both measure linear relations hip between X and Y the correlation coefficient and the R- squared of the regression should be related .

d)R - squared and the correlation coefficient have nothing to do with each other .

16)Which one is NOT correct regarding the error term?

a)ERROR = y - yhat

b)If ERRORS are random model is adequate .

c)ERROR is what is left over after we model .

d)Every error term should be 0 .

17)Given the following information : YHAT = 32 + 15X

Which of the following may NOT be true?

a)This is the least squares fitted line for a simple regression .

b) If x = 3 then yhat = 77 .

c) X is a significant variable .

d)We can forecast values of Y for any X.

18) Which of the following is NOT used in residual analysis ?

a)ACF of residuals Residuals over time .

b)Residuals against error term

c)Residuals against fitted values .

d)Histogram of residuals .

19)How do we calculate the t-value to test the significance of a variable in re analysis?

a)T - value = Coefficient / S * E of Coefficient

b)Take the square root of the R- squared .

c) Average across the X variable .

d)MINITAB gives it to us , otherwise we can't calculate it .

20)Which of the following does NOT have to be performed for the success of your regression model ?

a)R -squared .

b)F - test

c)Residual analysis t test

d) Stepwise regression

21)Why do we transform our variables ?

a)To eliminate the correlation between the X variables .

b) To make the relationship between X and Y linear .

c)We do not transform variables.

d)To get rid of autocorrelation .