7.11 Question bank
No answers are given for this section, by design.
Multiple choice
QB 7.1 The OLS slope estimator equals
- the correlation between \(x\) and \(y\)
- the sample covariance of \(x\) and \(y\) divided by the sample variance of \(y\)
- the sample covariance of \(x\) and \(y\) divided by the sample variance of \(x\)
- the ratio of the two sample means
QB 7.2 Which of the following hold by construction in every OLS regression? (select all that apply)
- \(\sum \hat{u}_i = 0\)
- \(\sum x_i \hat{u}_i = 0\)
- \(E(u \mid x) = 0\)
- The line passes through \((\bar{x}, \bar{y})\)
QB 7.3 In a simple regression, \(\hat{\sigma}^2\) divides the sum of squared residuals by
- \(n\)
- \(n - 1\)
- \(n - 2\)
- \(n - 3\)
QB 7.4 \(\mathrm{Var}(\hat{\beta}_1)\) decreases when
- \(\sigma^2\) increases
- the spread of \(x\) increases
- the spread of \(y\) increases
- \(R^2\) decreases
QB 7.5 A regression of wages on schooling gives \(R^2 = 0.11\). This means
- the model is misspecified
- schooling accounts for 11% of the variation in wages
- 11% of the coefficient is reliable
- schooling causes 11% of wage differences
QB 7.6 Omitted variable bias arises when the omitted variable
- is correlated with \(y\) only
- is correlated with \(x\) only
- is correlated with both \(x\) and \(y\)
- has a large variance
Short reasoning
QB 7.7 Explain why \(\sum x_i\hat{u}_i = 0\) is not evidence that \(E(u \mid x) = 0\).
QB 7.8 A student reports that their regression “explains 95% of the variation, so the model must be correct.” Evaluate.
QB 7.9 Why does the divisor in \(\hat{\sigma}^2\) change from \(n-1\) in Unit 3 to \(n-2\) here?
QB 7.10 In the fertiliser example the slope was 1.25 quintals per kilogram. Explain what would happen to that number, and to its standard error, if fertiliser were measured in grams instead.
QB 7.11 Explain why a regression can have a correctly computed standard error, a valid \(p\)-value, and a completely misleading interpretation.
QB 7.12 For six observations, \(\sum(x_i - \bar{x})(y_i - \bar{y}) = 84\), \(\sum(x_i - \bar{x})^2 = 56\), \(\bar{x} = 9\), \(\bar{y} = 21\) and \(\sum \hat{u}_i^2 = 12\).
- Compute \(\hat{\beta}_1\) and \(\hat{\beta}_0\).
- Compute \(\hat{\sigma}^2\) and \(\mathrm{se}(\hat{\beta}_1)\).
- Test \(H_0: \beta_1 = 0\) at the 5% level.
- Construct a 95% confidence interval for \(\beta_1\).
Numerical
QB 7.13 A regression on 30 observations gives \(\hat{\beta}_1 = 4.8\), \(\mathrm{se}(\hat{\beta}_1) = 1.5\), \(\text{SST} = 900\) and \(\text{SSR} = 400\).
- Compute \(R^2\).
- Test \(H_0: \beta_1 = 0\) at the 1% level.
- Test \(H_0: \beta_1 = 8\) at the 5% level.
- Compute \(\hat{\sigma}^2\).
QB 7.14 Yields and rainfall for five districts:
| Rainfall (cm) | 60 | 80 | 100 | 120 | 140 |
|---|---|---|---|---|---|
| Yield (quintals) | 18 | 23 | 24 | 30 | 35 |
- Compute the OLS estimates.
- Compute all five residuals and verify they sum to zero.
- Compute \(R^2\).
- Predict yield at 110 cm, and say why predicting at 300 cm would be unwise.