-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathapp01_errata.qmd
185 lines (107 loc) · 4.57 KB
/
app01_errata.qmd
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
# Errata {#errata}
## Preface
page xi, last word of first paragraph is **standaridzation**, s/b *standardization*
## Chapter 1
### Section 1.2.3.2, p. 11
The sentence of the 6th line on top of the page is
> We simulated the data according to the **hyothetical**
Should be *hypothetical*.
## Chapter 2
### section 2.4 Compute the standard error
The standard error formula used is with the standard deviation of the population and equivalent to
```{r }
#| eval: false
x <- whatifdat$Y[whatifdat$A == 1 & whatifdat$`T` == 1 & whatifdat$H ==1]
n <- length(x)
sd(x) * sqrt(n-1) / n
```
when the correct definition is with the standard deviation of the sample
```{r }
#| eval: false
x <- whatifdat$Y[whatifdat$A == 1 & whatifdat$`T` == 1 & whatifdat$H ==1]
n <- length(x)
sd(x) / sqrt(n)
```
### Figure 2.1, p. 30
This is really a small detail. The caption of the bottom plot is $\hat{E_{np}}(Y \mid A= 1, H =1, T = 1)$, s/b $\hat{E}_{np}$
## Chapter 3
### Page 37
The p-value found using a chi-square test is 3.207e-11. Using a t-test of the mean difference gives a p-value of 2.269e-9. It is not explained how the p-value of 0.032 is arrived at in the book.
### Typography: section 3.2 p. 40, equation 3.1
The current latex expression of conditional independence used seems to be `(Y(0), Y(1)) \ \text{II} \ T` with the output
$$
(Y(0), Y(1)) \ \text{II} \ T
$$
a better typography would be `\perp\!\!\!\perp` for the symbol $\perp\!\!\!\perp$. When used for equation 3.1 as `(Y(0), Y(1)) \perp\!\!\!\perp T` we obtain
$$
(Y(0), Y(1)) \perp\!\!\!\perp T
$$
In the case when we want to show dependence, that is *no independence* then the latex expression is `\not\!\perp\!\!\!\perp` for the symbol $\not\!\perp\!\!\!\perp$. For example equation 3.1 would become
$$
(Y(0), Y(1)) \not\!\perp\!\!\!\perp T
$$
## Chapter 4
### Section 4.1 p. 67 (on top)
The line is
> which is statistically **signficant**
should be *significant*
## Chapter 5
```{r echo=FALSE}
message("nothing found")
```
## Chapter 6
### Section 6.1 p. 100, first paragraph
The second sentence says
> Mistakingly equation $E_H E(Y \mid T=t \mid H)$ with \[...\]
Should it be $E_H (E(Y \mid T=t) \mid H)$? See extra $)$ before the last $\mid$.
### Section 6.3 p. 126, the script of `simdr`
The last paragraph of p. 126 says
> We simulated $T$ \[...\] such that approximaly 600 individuals had $T=1$
The `simdr` gives an incorrect result of 540 with the *constant 0.13*. That constant *should be 0.15* to obtain 600. See the mathematical proof and proof by simulation in the appendix *Doubly Robust Simulation* at [Analyse $T$](#errata6a).
## Chapter 7
### Section 7.2, equation (7.11), p. 139
> $$
> \begin{align*}
> E(Y_1 \mid A=1) - (E(Y_1\mid A=0, Y_0=1) - E(Y_1\mid A=0, Y_0=0)) - E(Y_0 \mid A=0) - (E(Y_1\mid A=0, Y_0=0) - E(Y_1\mid A=0, Y_0=0))
> \end{align*}
> $$
### Exercise 2
In the last paragraph of the exercise
> In addition, use `exsim.r` to simulate \[...\]
It should be `ex2sim.r`
## Chapter 8
### Section 8.2, p. 150
The very first sentence of section 8.2 says
> \[...\] the front-door **theorm** of Pearl \[...\]
It should be **theorem**
## Chapter 9
### Beginning of chapter, p. 158
Missing parentheses in the equation
$$
ITT = E(Y(1, A(1)) - E(Y(0, A(0))
$$ but is missing parentheses and should be
$$
ITT = E(Y(1, A(1))) - E(Y(0, A(0)))
$$
Based on the notation of mentioned in the second paragraph of p. 158, that we let $Y(t,a)$ be the potential outcome of $Y$ assuming we set $T=t$ and then $A=a$. Then the equation could be written more simply as
$$
ITT = E(Y(1, 1)) - E(Y(0, 0))
$$
### Section 9.3, p. 165
In the code for the example the problem is caused by the fact that `IV <- ITT / denom` does not work when `denom` is too small. What about setting the result to `NA` when `denom < tolerance` so the bootstrap will skip it and increase the number of bootstraps?
### Section 9.3, p. 169, 170, table 9.1
My results seem to be more consistent than the textbook's. Is this a mistake, how to test these results which can possibly be too good to be true.
## Chapter 10
### Section 10.3, code for equartiles.r
The following coding line is superfluous
`quartiles <- quantile(eb, c(0, .25, .5, .75, 1))`
## Chapter 11
### Section 11.2, p. 190
The data set **i17dat** is not in the material provided.
## Chapter 12
### section 12.1, p. 198
At the bottom of the page, th first sentence of the paragraph says
"by substituting parametric or **nonparmetric**. s/b **nonparametric**.
### section 12.3, p. 206
Just before the start of the exercise, beneath table 12.3
"is helpful of terms of **teasing** apart ...", s/b **tearing**