ISYE 6414 REGRESSION ANALYSIS —
MIDTERM 1 PRACTICE BANK
QUESTIONS 1–100 WITH ANSWERS AND DETAILED
EXPLANATIONS
FALL –2027 STUDY EDITION
Note: These are original practice questions, not actual or leaked Georgia Tech
examination questions.
QUESTION 1
A researcher fits the simple linear regression model
Yi=β0+β1Xi+ϵi.Y_i=\beta_0+\beta_1X_i+\epsilon_i.
Which interpretation of β1\beta_1 is correct?
A) The expected value of YY when X=0X=0
B) The variance of YY when X=0X=0
C) The expected change in YY associated with a one-unit increase in XX
D) The correlation between XX and YY
Rationale: C is correct because the slope represents the expected change in the
response for a one-unit increase in the predictor. A describes the intercept. B is
unrelated to the slope, and D confuses a regression coefficient with correlation.
QUESTION 2
In a simple linear regression model, the estimated equation is
Y^=12+4.5X.\hat Y=12+4.5X.
What does the value 12 represent?
A) The slope
1
,B) The residual standard error
C) The predicted mean response when X=0X=0
D) The correlation coefficient
Rationale: C is correct because 12 is the estimated intercept. A slope is 4.5. B
measures unexplained variation, while D is a standardized measure of linear
association.
QUESTION 3
A fitted regression model has residuals that become increasingly spread out as fitted
values increase. Which assumption is most directly questionable?
A) Independence
B) Linearity
C) Normality of predictors
D) Constant variance
Rationale: D is correct because increasing residual spread indicates
heteroscedasticity. A concerns dependence among errors, B concerns systematic
curvature, and C is not a standard regression-error assumption.
QUESTION 4
A residual-versus-fitted plot displays a pronounced U-shaped pattern. What is the
most likely interpretation?
A) The errors are perfectly normal
B) The linear mean structure may be inadequate
C) The predictors are independent
D) The response has zero variance
Rationale: B is correct because systematic curvature in residuals suggests that a
straight-line relationship does not adequately represent the conditional mean.
Randomly scattered residuals would be preferable.
2
,QUESTION 5
Suppose a regression model produces R2=0.81R^2=0.81. Which statement is most
appropriate?
A) 81% of observations are predicted perfectly
B) The slope must be statistically significant
C) 81% of the sample variability in the response is explained by the fitted linear
model
D) The model's prediction error is 19%
Rationale: C is correct. R2R^2 measures the proportion of observed response
variability accounted for by the fitted regression model. It does not imply perfect
prediction or automatically establish statistical significance.
QUESTION 6
A statistician wants to determine whether the population slope differs from zero.
Which hypothesis test is appropriate?
A) H0:β0=0H_0:\beta_0=0
B) H0:R2=1H_0:R^2=1
C) H0:β1=0H_0:\beta_1=0
D) H0:σ2=0H_0:\sigma^2=0
Rationale: C is correct because testing whether the slope is zero evaluates whether
there is evidence of a linear association between the predictor and response. The
other hypotheses address different quantities.
QUESTION 7
A regression coefficient has an estimated value of 2.8 with standard error 0.7. What is
the corresponding t-statistic for testing H0:β=0H_0:\beta=0?
A) 0.25
B) 1.96
C) 4.00
3
, D) 4.90
Rationale: C is correct because t=(2.8−0)/0.7=4t=(2.8-0)/0.7=4. The t-statistic
measures how many estimated standard errors the coefficient is from the null value.
QUESTION 8
A 95% confidence interval for a regression slope is (−0.4,2.1)(-0.4,2.1). What
conclusion follows?
A) The slope is definitely positive
B) The slope is definitely negative
C) There is insufficient evidence at the 5% level to conclude that the population
slope differs from zero
D) The regression model is invalid
Rationale: C is correct because zero lies inside the 95% confidence interval.
Therefore, a two-sided test of H0:β1=0H_0:\beta_1=0 would not reject at the 5% level.
QUESTION 9
What is the primary purpose of a prediction interval for a future observation?
A) Estimate only the population mean response
B) Estimate the regression coefficient
C) Quantify uncertainty for an individual future response
D) Test whether the intercept equals zero
Rationale: C is correct. A prediction interval incorporates both uncertainty in
estimating the mean response and the additional individual-error variation.
Consequently, prediction intervals are wider than confidence intervals for the mean.
QUESTION 10
For a fixed predictor value x0x_0, which interval is generally wider?
A) Confidence interval for the mean response
4
MIDTERM 1 PRACTICE BANK
QUESTIONS 1–100 WITH ANSWERS AND DETAILED
EXPLANATIONS
FALL –2027 STUDY EDITION
Note: These are original practice questions, not actual or leaked Georgia Tech
examination questions.
QUESTION 1
A researcher fits the simple linear regression model
Yi=β0+β1Xi+ϵi.Y_i=\beta_0+\beta_1X_i+\epsilon_i.
Which interpretation of β1\beta_1 is correct?
A) The expected value of YY when X=0X=0
B) The variance of YY when X=0X=0
C) The expected change in YY associated with a one-unit increase in XX
D) The correlation between XX and YY
Rationale: C is correct because the slope represents the expected change in the
response for a one-unit increase in the predictor. A describes the intercept. B is
unrelated to the slope, and D confuses a regression coefficient with correlation.
QUESTION 2
In a simple linear regression model, the estimated equation is
Y^=12+4.5X.\hat Y=12+4.5X.
What does the value 12 represent?
A) The slope
1
,B) The residual standard error
C) The predicted mean response when X=0X=0
D) The correlation coefficient
Rationale: C is correct because 12 is the estimated intercept. A slope is 4.5. B
measures unexplained variation, while D is a standardized measure of linear
association.
QUESTION 3
A fitted regression model has residuals that become increasingly spread out as fitted
values increase. Which assumption is most directly questionable?
A) Independence
B) Linearity
C) Normality of predictors
D) Constant variance
Rationale: D is correct because increasing residual spread indicates
heteroscedasticity. A concerns dependence among errors, B concerns systematic
curvature, and C is not a standard regression-error assumption.
QUESTION 4
A residual-versus-fitted plot displays a pronounced U-shaped pattern. What is the
most likely interpretation?
A) The errors are perfectly normal
B) The linear mean structure may be inadequate
C) The predictors are independent
D) The response has zero variance
Rationale: B is correct because systematic curvature in residuals suggests that a
straight-line relationship does not adequately represent the conditional mean.
Randomly scattered residuals would be preferable.
2
,QUESTION 5
Suppose a regression model produces R2=0.81R^2=0.81. Which statement is most
appropriate?
A) 81% of observations are predicted perfectly
B) The slope must be statistically significant
C) 81% of the sample variability in the response is explained by the fitted linear
model
D) The model's prediction error is 19%
Rationale: C is correct. R2R^2 measures the proportion of observed response
variability accounted for by the fitted regression model. It does not imply perfect
prediction or automatically establish statistical significance.
QUESTION 6
A statistician wants to determine whether the population slope differs from zero.
Which hypothesis test is appropriate?
A) H0:β0=0H_0:\beta_0=0
B) H0:R2=1H_0:R^2=1
C) H0:β1=0H_0:\beta_1=0
D) H0:σ2=0H_0:\sigma^2=0
Rationale: C is correct because testing whether the slope is zero evaluates whether
there is evidence of a linear association between the predictor and response. The
other hypotheses address different quantities.
QUESTION 7
A regression coefficient has an estimated value of 2.8 with standard error 0.7. What is
the corresponding t-statistic for testing H0:β=0H_0:\beta=0?
A) 0.25
B) 1.96
C) 4.00
3
, D) 4.90
Rationale: C is correct because t=(2.8−0)/0.7=4t=(2.8-0)/0.7=4. The t-statistic
measures how many estimated standard errors the coefficient is from the null value.
QUESTION 8
A 95% confidence interval for a regression slope is (−0.4,2.1)(-0.4,2.1). What
conclusion follows?
A) The slope is definitely positive
B) The slope is definitely negative
C) There is insufficient evidence at the 5% level to conclude that the population
slope differs from zero
D) The regression model is invalid
Rationale: C is correct because zero lies inside the 95% confidence interval.
Therefore, a two-sided test of H0:β1=0H_0:\beta_1=0 would not reject at the 5% level.
QUESTION 9
What is the primary purpose of a prediction interval for a future observation?
A) Estimate only the population mean response
B) Estimate the regression coefficient
C) Quantify uncertainty for an individual future response
D) Test whether the intercept equals zero
Rationale: C is correct. A prediction interval incorporates both uncertainty in
estimating the mean response and the additional individual-error variation.
Consequently, prediction intervals are wider than confidence intervals for the mean.
QUESTION 10
For a fixed predictor value x0x_0, which interval is generally wider?
A) Confidence interval for the mean response
4
