I would find it hard to recommend this book. On several occasions, I'm left wondering whether it's me or the author who has misunderstood.
For example, the author notes that always rounding numbers like 2.45 up (to 2.5) introduces statistical bias, and prefers a variant of the "round to even" method, which in this case means rounding down to 2.4. Fair enough, but he also rounds down in cases like 2.459, which he would round to 2.4, even though it's clearly closer to 2.5. I'm left wondering what the real justification for this is.
Another example: the author gives two methods of calculating percentiles; the first of these seems to assume that the distribution is perfectly flat, yet this isn't explicitly stated. Again, I'm left wondering whether it's me or the author who has misunderstood.