Unlocking Linear Regression: Advanced Concepts and Applications — Questions & Answers

This quiz contains 10 questions. Below is a complete reference of all questions, answer choices, and correct answers. You can use this section to review after taking the interactive quiz above.

1. Question 1: Interpretation of Slope in Linear Regression

   In a simple linear regression analysis relating students’ study hours (X) to their test scores (Y), what does the slope coefficient represent?

   • A) The expected change in test score for each additional hour studied
   • B) The minimum value that test scores can reach
   • C) The ratio between average test scores and study hours
   • D) The proportion of study hours explained by test scores
   • E) The sum of squared residuals divided by the number of students
   Show correct answer

   Correct answer: A) The expected change in test score for each additional hour studied

2. Question 2: Gauss-Markov Theorem Application

   Which assumption is necessary for the Ordinary Least Squares (OLS) estimator in linear regression to be the Best Linear Unbiased Estimator (BLUE)?

   • A) The variance of the errors is constant (homoscedasticity)
   • B) There must be at least twice as many predictors as observations
   • C) The dependent variable must be binary
   • D) Residuals must always be positive
   • E) Predictors are always measured in logarithmic scales
   Show correct answer

   Correct answer: A) The variance of the errors is constant (homoscedasticity)

3. Question 3: Interpreting the R-squared Value

   What does an R-squared value of 0.82 indicate in the context of a linear regression predicting house prices from square footage?

   • A) 82% of the variance in house prices is explained by square footage
   • B) 18% of the house prices are predictable from other variables
   • C) The residual standard error is 0.82
   • D) 82% of the coefficients are statistically significant
   • E) The correlation between square footage and price is 0.82
   Show correct answer

   Correct answer: A) 82% of the variance in house prices is explained by square footage

4. Question 4: Dealing with Multicollinearity

   If two independent variables in a multiple linear regression are highly correlated, what is the main risk introduced to the model?

   • A) Coefficient estimates may become unstable and difficult to interpret
   • B) The regression line will always pass through the origin
   • C) The model will automatically switch to non-linear regression
   • D) Prediction residuals will necessarily become negative
   • E) The number of observations will have to double
   Show correct answer

   Correct answer: A) Coefficient estimates may become unstable and difficult to interpret

5. Question 5: Detecting Non-Linearity

   When examining residual plots after fitting a linear regression model, which pattern suggests that the linearity assumption may have been violated?

   • A) Residuals forming a distinct curve or pattern rather than being randomly scattered
   • B) Residuals are all near zero
   • C) Residuals exactly match the predicted values
   • D) Residuals show a perfectly vertical line
   • E) Residuals have only positive values
   Show correct answer

   Correct answer: A) Residuals forming a distinct curve or pattern rather than being randomly scattered

6. Question 6: Assumption Checking: Independence

   Why is it problematic if the errors in a linear regression model are autocorrelated, such as in time series data?

   • A) Standard error estimates become unreliable, leading to invalid hypothesis tests
   • B) The slope will always be zero
   • C) R-squared value will exceed 1
   • D) The residual plot shows only negative values
   • E) The response variable must be categorical
   Show correct answer

   Correct answer: A) Standard error estimates become unreliable, leading to invalid hypothesis tests

7. Question 7: Interpreting Regression Output

   If a 95% confidence interval for a regression coefficient includes zero, what can you conclude about that predictor?

   • A) The predictor may not be statistically significant at the 5% level
   • B) The predictor explains all the variability in the response
   • C) The predictor must be removed from the dataset
   • D) The predictor causes perfect multicollinearity
   • E) The regression coefficients cannot be interpreted
   Show correct answer

   Correct answer: A) The predictor may not be statistically significant at the 5% level

8. Question 8: Effect of Outliers

   How do influential outliers typically affect the fitted regression line?

   • A) They can disproportionately shift the regression line and bias parameter estimates
   • B) They improve the generalizability of the model
   • C) They automatically reduce the residual standard deviation to zero
   • D) They have no impact on the line due to normalization
   • E) They always make the R-squared value larger
   Show correct answer

   Correct answer: A) They can disproportionately shift the regression line and bias parameter estimates

9. Question 9: Application of Dummy Variables

   When including categorical independent variables, such as gender or region, in a linear regression model, which technique is commonly used?

   • A) Creating dummy variables for the categories
   • B) Calculating the square root of their means
   • C) Ignoring the categorical variables entirely
   • D) Applying the chi-square test directly
   • E) Dividing all predictors by the median value
   Show correct answer

   Correct answer: A) Creating dummy variables for the categories

10. Question 10: Addressing Heteroscedasticity

    If a plot of residuals against fitted values shows a fan or cone shape, what statistical issue might be present and what is a common remedy?

    • A) Heteroscedasticity; try transforming the dependent variable or using robust standard errors
    • B) Multicollinearity; remove one of the variables
    • C) Non-stationarity; use lagged variables
    • D) Homoscedasticity; proceed as usual
    • E) Autocorrelation; shuffle the observations
    Show correct answer

    Correct answer: A) Heteroscedasticity; try transforming the dependent variable or using robust standard errors