"[The] only [book] known to me in which "please" is common wording in exercises."
"[T]he author is a kind of magpie fascinated by much of what he reads. He is especially fond of quotations, digressions, and allusions, which here are numerous, but not always apposite or presented concisely."
Improvements
Update subtitle to mention quantitative social scientists.
Change ideal reader in preface.
Check for "a repeated mantra that open source is good"
Make case "for using the tidyverse or ggplot2". Note in the end (when mentioning Python) that "some smart R users regard them as between over-rated and best avoided."
Add some more explanations of the code. Maybe not in comments, but instead in text?
"Note that the use of proprietary statistical software is not mentioned at all, as if self-evidently beyond the pale."
Fix graphs:
clumsy text labels;
using alphabetical order for categories (often a default that is easily improved on);
histograms with unrevealing bin choices (the two ECDF plots given show better practice, and to my taste quantile plots would be better still);
redundant or avoidable legends;
poor color choices.
Typos
Some typos are accidentally entertaining, as when "denotated" has morphed into "detonated" (p.279); "denoted" would have served as well or better.
As a conservative on English style, I suggest that (usually) it is better to write based on, centered on, compared with, composed not comprised, different from, and distinct, not unique.
A notable tic is that Alexander writes "a lot" -- a lot, often repeatedly in the same paragraph, or on the same page. (What's wrong with "many" or "much"?)
The author says (p.188) Euler but means Gauss.
He repeats (p.135) the unfounded but oddly persistent idea that M.E. Spear invented the box plot in 1952. Not so; similar diagrams were standard in geography from the 1930s on and they have yet longer 19th century roots.
More importantly, the explanation of Tukey's 1.5 IQR rule is incorrect.
The author mixes N(mean, SD) and N(mean, variance) conventions for denoting normal distributions, but then again that is true of the literature and much software.
The formula for the normal density is wrong on p.388. An
exercise about growing seedlings is set on p.15 and there is a promise to return to analysing the data, but beyond an allusion on p.163 I didn't see that this was ever done.
Funny quotes
Improvements
Typos