STAT 202 FINAL ASSESSMENT: STATISTICAL MODELS AND ANALYSIS

STAT 202 FINAL ASSESSMENT: STATISTICAL MODELS
AND ANALYSIS




After solving a linear regression problem, I have calculated all of my residuals and plotted them.
I have discovered that for low values of the independent variable, I have little dispersion in
residuals but for high values of the independent variable, I have a lot of dispersion in residuals.
Which assumption of linear regression have I violated?



Normally distributed errors



Constant variance of errors



Independence of errors (no autocorrelation)



None of the above - answers✔✔Constant variance of errors



The default hypothesis test in linear regression is two tailed.



True



False - answers✔✔True



When new variables are added in multiple linear regression, R2 (R squared) can never decrease.

, True



False - answers✔✔True



In linear regression, we can build confidence intervals for many of the outputs. The confidence
interval for the point estimate is always smaller than the confidence interval for the estimate of
the conditional mean.



True



False - answers✔✔False



Linear regression can only take continuous variables as independent (predictor variables).



True



False - answers✔✔False



In linear regression, the sum of the residuals will always sum to 0.



True



False - answers✔✔True



How large must my sample size be in linear regression in order to ignore the assumption of
normally distributed errors?