Bayesian Statistics

Typos in the first printing, corrected in second printing

- In the R code of Figure 1.1, a log(3) on the rhs column should be a log(10) to coincide with the lhs column. Similarly, in Figure 1.2, the first item produces a vector of 20 terms. At last, in Figure 1.3, the first command produces a vector with five, not three, values (Thanks to Soleiman).
- Similarly, the example on the syntax of while at the top of page 13 should be while (expres1) where expres1 is a boolean expression (Thanks to Petri Koistinen).
- In Section 2.2.1, at the bottom of page 19, the indicator on θ should be an indicator on μ (Thanks to Fazlollah Lak).
- In the remark page 34, the reference on pseudo-Bayes
factors should be Robert (2001, Chapter 5) (Thanks to Fazlollah Lak).

- In Section 3.1.2, in the formula right after Exercise 3.3, the summation term is superfluous, given the matricial notation. But the X matrix is missing in front of the β (Thanks to Dominik Hangartner).
- In Section 3.2.1.,
in the data analysis on page 57, the variance involves a (2.1-1)
^{2}rather than a (2.1-2)^{2 }but the value is correct (Thanks to Dominik Hangartner). - In Exercise
4.12, y
_{i}is defined as the indicator that z_{i}is negative and in Exercise 4.13, the conditions for properness cannot be expressed only in terms of both sums Σ_{i}y_{i}and Σ_{i}(1-y_{i}). - In Section 5.2.3, at the bottom of page 129, π(N,P) should be π(N,p).
- On page 162, the permutation τ(θ,p) is missing the
p
_{i}'s (Thanks to Jarrett Barber) - On top of page 163, there are two s
_{1}^{2}(z), instead of s_{1}^{2}(z) and s_{2}^{2}(z) (Thanks to Jarrett Barber) - On Figure 6.7, the lower rhs graph is for the pair (σ
_{i},μ_{i}) not (σ_{i},p_{i}) (Thanks to Jarrett Barber) - On page 173, the squares [...]
^{2}within the exponentials are missing the norm ||...||^{2}symbol (Thanks to Jarrett Barber) - Right after Exercise
6.18, there is a k
_{1}in the denominator of the birth acceptance probability that should be a k+1 (Thanks to Hossein Gholami.) - In Exercise
7.4, the Student's t
distribution is
with T-1
degrees of freedom and with a different value of ν
_{t}than the one proposed. - In formula (7.4), Section
7.2.1, the sum involves x
_{t-i}not x_{t+1-i}(Thanks to Hossein Gholami and to Jarrett Barber.) - In Exercise
7.7, the prior 1/σ
must be understood as a prior
on σ,
not σ
^{2} - In Exercise 7.8, the final sentence "Deduce that...." should be removed.
- In Exercise
7.9, the lognormal proposal should have log(σ
_{t-1}^{2}) as its mean. - Formula (7.10) is missing a minus sign after
ε
_{t}. - In Exercise
7.17, the definition of φ
_{r}involves x_{1:r-1}, not x_{1:t-1}. - On page 219, the g is missing in the polyprobit model P(y=g|y) (Thanks to Jarrett Barber)
- In Chapter 8, the conditional distributions
(8.1) do not
lead to a joint distribution if the N
_{k}(i) are different. Rewrite (8.1) with N_{k}(i) being the average number of neighbours over all x_{i}'s. Same thing for (8.2). (Then Exercise 8.5 can be solved!) - On page 222, the p
_{g}on the last line should be a p_{G}, but this is unrelated with parental guidance! (Thanks to Jarrett Barber) - On page 224, the uniform proposal is on the 2r neighbours of k, not the r neighbours (Thanks to Jarrett Barber)
- On page 225, for the vision dataset, the Metropolis-Hastings chain concentrates on k=6, not k=5 (Thanks to Jarrett Barber)
- In Exercise
8.6, the
condition on
the support of π
is unclear. It should be phrased as "If a joint density π(y
_{1},...,y_{n}) is such that the conditionals π(y_{-i}|y_{i}) never cancel on the supports of the marginals m_{-i}(y_{-i}), show that the support of π is equal to the cartesian product of the supports of the marginals." - In Exercise 8.10, the
indices are not right: it should read as x
_{ij}with (i+j) = 0(2) conditional on x_{ij}with (i+j) = 1(2) - On pages 237 and 239, including (8.8), the double summations should reduce to a single summation on the (i,j)'s such that i~j.