mattblackwell / gov2002-book Goto Github PK

This aforementioned second last sentence is:

If any variable in $\mathbf{X}_i$ is continuous, we must estimate an infinite number of possible values of $\mathbf{x}$.

This should instead be:

If any variable in $\mathbf{X}_i$ is continuous, we must estimate an infinite number of possible values of $\mu(\mathbf{x})$.

since the thing of interest (that we want to estimate) is $\mu(\mathbf{x})$ and not the values $\mathbf{x}$.

The first (left) display equation after Section 5.2.2 is:

$m(X_i,X_i^2)=\beta_0+\beta_1 X_i\beta_2 X_i^2$.

This should be replaced with:

$m(X_i,X_i^2)=\beta_0+\beta_1 X_i+\beta_2 X_i^2$.

(added a missing plus sign)

The first sentence of the second paragraph of Section 6.6 starts with:

Let $\mathcal{C}(\mathbb{X})={ \mathbb{X}\mathbf{b}\colon \mathbb{b}\in \mathbb{R}^2} be the column space...

This should be:

Let $\mathcal{C}(\mathbb{X})={ \mathbb{X}\mathbf{b}\colon \mathbb{b}\in \mathbb{R}^{k+1}} be the column space...

(replaced $2$ with $k+1$ in the dimension of the Euclidean space since outside of examples, it is assumed that $\mathbb{X}$ is a $n\times(k+1)$ matrix)

In the paragraph before the "Chi-squared critical values" box, there is a sentence that starts with:

After recentering ad rescaling by the covariance matrix, ...

This should be

After recentering and rescaling by the covariance matrix, ...

(replaced "ad" with "and")

The middle sentence in Theorem 7.3 is:

and its conditional sampling variance issue

This should be:

and its conditional sampling variance is

(changed "issue" to "is")

The third point (3.) in the "Residual regression approach" box is:

Use OLS to regression $\tilde{\mathbf{e}}_2$ on $\tilde{\mathbb{X}}_1$.

This should be:

Use OLS to regress $\tilde{\mathbf{e}}_2$ on $\tilde{\mathbb{X}}_1$.

(replaced "regression" with "regress")

There is a sentence in Chapter 7 which starts with:

which also typically uses the the homoskedastic variance estimator ...

This should be replaced with:

which also typically uses the homoskedastic variance estimator ...

(removed an extra "the")

The second equation of Example 3.3 is:

$\sum_{i=1}^n (X_i - \bar{X}n)^2 = \sum{i=1}^n X_i^2 + n\bar{X}_n$

but should be

$\sum_{i=1}^n (X_i - \bar{X}n)^2 = \sum{i=1}^n X_i^2 + n\bar{X}_n^2$

which is consistent with the later derivation.

The first sentence of the third last paragraph before Section 5.4 is:

Thus, we can write the CEF with two binary covariates as linear when the linear specification includes and multiplicative interaction between them $(x_1,x_2)$.

This should instead be:

Thus, we can write the CEF with two binary covariates as linear when the linear specification includes a multiplicative interaction between them $(x_1,x_2)$.

(replaced "and" with "a" after "includes")

Theorem 3.3 states:

Let $X_1,\dots,X_n$ be a an i.i.d. draws ...

which should probably be replaced by

Let $X_1,\dots,X_n$ be iid draws ...

(removing the extra "a" and keeping the iid shorthand consistent with the rest of the book)

The last line of Example 4.3 ends with:

... and if we want an asymptotically level of 0.05, we can reject when $|T|>1.96$.

This should instead be:

... and if we want an asymptotic level of 0.05, we can reject when $|T|>1.96$.

(replace "asymptotically" with "asymptotic")

Before Section 2.4.2, there is the sentence:

Maximum likelihood estimators have very nice properties, especially in large samples. Unfortunately, it also requires the correct knowledge of the parametric model, which is often difficult to justify.

The second sentence should probably refer to MLEs in the plural as so:

Maximum likelihood estimators have very nice properties, especially in large samples. Unfortunately, they also require the correct knowledge of the parametric model, which is often difficult to justify.

The third sentence of Chapter 5 is:

For example, we may want to know how wait voting poll wait times vary as a function of some socioeconomic features of the precinct, like income and racial composition.

This should instead be:

For example, we may want to know how voting poll wait times vary as a function of some socioeconomic features of the precinct, like income and racial composition.

(removed an extra "wait" before "voting poll")

In point #2 on this page, the sentence:

So many methods are either use regression estimators like ordinary least squares or extend it in some way.

should be modified to

So many methods either use regression estimators like ordinary least squares or extend it in some way.

i.e. there's an extra "are" term.

The first part of the second display equation in Section 6.9.1 is:

$\hat{\mathbf{Y}}=\mathbf{P}\mathbf{Y}$

This should be replaced with:

$\hat{\mathbf{Y}}=\mathbf{P}_{\mathbb{X}}\mathbf{Y}$

to keep the notation of the projection matrix consistent with what is used elsewhere in the chapter

The sentence after Figure 6.6 is:

One measure of influence is called DFBETA$_i$ measures how much $i$ changes the estimated coefficient vector ...

This should be:

One measure of influence, called DFBETA$_i$, measures how much $i$ changes the estimated coefficient vector ...

A sentence in the paragraph before Section 6.9.1 starts with:

Thus, by definition, This means that when an observation ...

This should be replaced by:

Thus, by definition, this means that when an observation ...

The second sentence of the second paragraph of the Warning box before Section 4.9 is:

Of course, this doesn’t make sense from our definition because the p-values conditions on the null hypothesis—it cannot tell us anything about the probability of that null hypothesis.

This should instead be:

Of course, this doesn’t make sense from our definition because the p-value conditions on the null hypothesis—it cannot tell us anything about the probability of that null hypothesis.

(replace "p-values" with "p-value")

One sentence after the Section 7.1 header reads:

Remember that since $\hat{\boldsymbol{\beta}}$ is a vector, then the variance of that estimator will actually be a variance-covariance matrix.

This should be:

Remember that since $\hat{\boldsymbol{\beta}}$ is a vector, the variance of that estimator will actually be a variance-covariance matrix.

(removed the "then")

A sentence right before Section 7.4.1 starts with:

Recall the heteroskedastic-consistent variance estimator is ...

This should be:

Recall the heteroskedastic-consistent variance estimator ...

(removed the "is" because right after the display equation the sentence starts with "is")

Near the end of the page:

Wooldridge, Jeffrey. Econometric Analysis of Cross Section and Panel Data

should be replaced by

Wooldridge, Jeffrey. Econometric Analysis of Cross Section and Panel Data. The MIT Press.

for consistency with earlier bullet points?

The end of the second last sentence of the first paragraph in Section 6.4 is:

... linear independence means that if $\mathbb{X} \mathbf{b} =0$ if and only if $\mathbf{b}$ is a column vector of 0s.

This should be:

... linear independence means that $\mathbb{X} \mathbf{b} =0$ if and only if $\mathbf{b}$ is a column vector of 0s.

(removed the "if" before the "if and only if" statement)

The first display equation after Section 5.3 of Chapter 5 is (using the LaTeX shorcuts you defined in the source code):

$$ \mu(\X) = \E[Y_{i} \mid \X_{i}] = \argmin_{g(\X_i) \in L_2}; \E\left[(Y_{i} - f(\X_{i}))^{2}\right], $$

This should instead be:

$$ \mu(\X) = \E[Y_{i} \mid \X_{i}] = \argmin_{g(\X_i) \in L_2}; \E\left[(Y_{i} - g(\X_{i}))^{2}\right], $$

(replaced "f" in the argmin expression with "g" to match the "g" the argmin is being taken over)

The statement of Definition 4.4 is:

The p-value of a test is the probability of observing a test statistic is at least as extreme as the observed test statistic in the direction of the alternative hypothesis.

This should instead be:

The p-value of a test is the probability of observing a test statistic at least as extreme as the observed test statistic in the direction of the alternative hypothesis.

(the "is" before "at least" should be removed)

The last sentence in the paragraph after Section 5.3 of Chapter 5 starts with:

In particular, if we label $L_2$ be the set of all functions of the covariates $g()$ that have finite squared expectation,...

This should instead be:

In particular, if we label $L_2$ to be the set of all functions of the covariates $g()$ that have finite squared expectation,...

(added "to" before "be")

The paragraph after Definition 4.3 in Chapter 4 is:

A test with a significance level of $\alpha = 0.05$ will have a false positive/type I error rate no larger than 0.05. This level is widespread in the social sciences, though you also will $\alpha = 0.01$ or $\alpha = 0.1$. Frequentists justify this by saying this means that with $\alpha = 0.05$, there will only be 5% of studies that will produce false discoveries.

This should instead be (with differences in brackets []):

A test with a significance level of $\alpha = 0.05$ will have a false positive/type I error rate no larger than 0.05. This level is widespread in the social sciences, though you also will [see] $\alpha = 0.01$ or $\alpha = 0.1$. Frequentists justify this by saying this means that with $\alpha = 0.05$, there will only be [at most] 5% of studies that will produce false discoveries.

(a missing "see" word and an addition of "at most" to make the last statement more precise)

The first two points (1. and 2.) of the "Linear projection assumptions" box is:

1. ${(Y_i,\mathbf{X}i)}{i=1}^n are iid random vectors.
2. $\mathbb{E}[Y_{i^2}]<\infty$ (finite outcome variance)

These should be replaced by:

1. ${(Y_i,\mathbf{X}i)}{i=1}^n are iid random vectors
2. $\mathbb{E}[Y_{i}^2]<\infty$ (finite outcome variance)

(removed the period in (1.) and fixed the subscript typo in (2.))

Here a few typos I've collected while consulting the textbook (the corrections are in bold):

1. Here: "Implicit in this analysis [...]"
2. Here: "Remember that the interpretation of confidence [...]"
3. Here: "@fig-ci-sim shows 100 iterations of these steps. Here we see that, as expected, the large majority of calculated CIs contain the true value. [...] The guarantee of the 95% confidence intervals [...]"
4. Here: "we view $\mb{Y}$ as an $n$-dimensional vector in $\mathbb{R}^n$."

Thank you @mattblackwell for writing this helpful resource and making it publicly available for free!

In Example 2.8, the random variable $X_i$ initially introduced has support on ${0,1,2,3,4}$, which has size 5. However, later on this is modeled as a Binomial with support ${0,1,2,3}$ with the choice size = 4, which doesn't match the original rv. It would probably be more appropriate to model using size = 5 for all the subsequent simulations in the example.

In the paragraph preceding the large Jacobian matrix display equation, there is a sentence which ends like:

... and be continuously differentiable (we make the function bold since it ).

The remark in the parenthesis should probably be completed.

In the first paragraph after Section 2.2, it is said that:

They are independent in that the random vectors $X_i$ and $X_j$ are independent for all $i\neq j$,

However, this is pairwise independence, which is strictly weaker than mutual independence (which is required in the iid assumption). I checked this here:

https://en.wikipedia.org/wiki/Independent_and_identically_distributed_random_variables#Definition_for_more_than_two_random_variables

In the first warning box, there is the sentence:

Then the statistic is a random variable that has a distribution over the numbers from {2, , 12} that describes our uncertainty over what the sum will be before we roll the dice.

However, the {2, , 12} seems like an odd choice to format the set, and could possibly be replaced with nicer LaTeX formatting, i.e. {2,\dots,12} (formatted in LaTeX)?

The sentence right before the Jacobian matrix is:

It will help us use more compact matrix notation if we introduce a Jacobian matrix of all partial derivatives

This should be replaced with

It will help us to use more compact matrix notation if we introduce a Jacobian matrix of all partial derivatives

(adding a "to")

The second sentence in Chapter 5 is:

In particular, these tools show how the conditional mean of $Y_i$ varies as a function $\mathbf{X}_i$.

This should be:

In particular, these tools show how the conditional mean of $Y_i$ varies as a function of $\mathbf{X}_i$.

(added an "of" near the end)

Example 2.10 starts as:

Example 2.10 (Sampling variance of the sample mean:)

but this should probably be replaced with

Example 2.10 (Sampling variance of the sample mean)

to be consistent with the other examples before this

The sentence right before Section 2.6.3 is:

Given the above derivation, the standard error of the sample mean under iid sampling is $\sigma/\sqrt{n}$

This is missing a period at the end of the sentence, i.e. should be replaced with:

Given the above derivation, the standard error of the sample mean under iid sampling is $\sigma/\sqrt{n}$.

The paragraph before the numbered statements in the Blue box after Section 5.2.3 is:

Without some assumptions on the joint distribution of the data, The following “regularity conditions” will ensure the existence of the BLP:

This should be replaced by:

Without some assumptions on the joint distribution of the data, the following “regularity conditions” will ensure the existence of the BLP:

(changed "T" to "t" in "the")

The sentence right before Figure 3.3 reads:

... establishes how the standard 95% confidence interval for the sample mean above asymptotically valid.

This should be:

... establishes how the standard 95% confidence interval for the sample mean above is asymptotically valid.

(adding an "is")

The first sentence after the first display equation in Section 5.4 is:

so that the change in the predicted outcome for increasing $X_{i1}$ by one unit isn’t

Following the logical structure of the argument, this should be:

so that the change in the predicted outcome for increasing $X_{i1}$ by one unit is

(replaced "isn't" with "is")

The start of a paragraph in Chapter 6 is:

Why does this rank condition matter for the OLS estimator? A key property of full column rank matrices is that $\mathbb{X}$ if of full column rank if and only if $\mathbb{X}'\mathbb{X}$ is non-singular and a matrix is invertible if and only if it is non-singular.

This should be replaced by:

Why does this rank condition matter for the OLS estimator? A key property of full column rank matrices is that $\mathbb{X}$ is of full column rank if and only if $\mathbb{X}'\mathbb{X}$ is non-singular and a matrix is invertible if and only if it is non-singular.

(replaced wrongly used "if" with "is")

The first sentence of Footnote #2 in Chapter 4 reads:

Different people and different textbooks describe what to do when do not reject the null hypothesis in different ways.

This should instead be:

Different people and different textbooks describe what to do when we do not reject the null hypothesis in different ways.

In the proof of Theorem 3.1, the notation $\mathbb{1}$ is used for the indicator random variable. But in the definition of the empirical cdf right before Section 2.5, the notation $\mathbb{I}$ is used for the indicator random variable. Not sure if this is an issue or typo, but just wanted to make a note of it in case it is of interest.

The display equation right before Theorem 7.1 is:

$$ \bhat \inprob \beta + \mb{Q}_{\X\X}^{-1}\E[\X_ie_i] = \beta, $$

This should instead be:

$$ \bhat \inprob \boldsymbol{\beta} + \mb{Q}_{\X\X}^{-1}\E[\X_ie_i] = \boldsymbol{\beta}, $$

since beta is a vector and not a scalar.

The first display equation after Section 5.2.3 is:

$\mathbf{x}'\boldsymbol{\beta}=x_1\beta_1+x_2\beta_2+\cdots+X_k\beta_k$.

This should instead be:

$\mathbf{x}'\boldsymbol{\beta}=x_1\beta_1+x_2\beta_2+\cdots+x_k\beta_k$.

(the last x should be lowercase)

The word after the $n\to \infty$ expression is "of" when it should probably be "or".

The last sentence of the second last paragraph before Section 6.6 is:

Then coefficient on the West dummy will be

This should be:

Then, the coefficient on the West dummy will be

(added a "the")

The sentence after the $T$ test statistic display equation after the Section 4.11 heading is:

As we discussed in the earlier, an $\alpha = 0.05$ test would reject this null when $|T|>1.96$, or when...

This should instead be:

As we discussed earlier, an $\alpha = 0.05$ test would reject this null when $|T|>1.96$, or when...

(removed the "in the" before "earlier")

The first sentence of the footnote reads:

This approach to inference is often called a model-based approach since we assuming a probability model in the cdf, $F$.

and should instead be:

This approach to inference is often called a model-based approach since we are assuming a probability model in the cdf, $F$.

(a missing "are" term)

In the last paragraph of Example 3.3, it is said that the bias shrinks as a function of the sample size, so as long as the sampling variance shrinks as a function of the sample size (which it does), the estimator is consistent by Theorem 3.5.

However, Theorem 3.5 only says that an estimator is consistent if it is unbiased and its sampling variance shrinks as a function of the sample size, and no mention was made about the slightly stronger version used in Example 3.3.