A correspondent writes:
I just want to spend a few words to point you to this book I have just found on Amazon: “Understanding The New Statistics: Effect Sizes, Confidence Intervals, and Meta-Analysis” by G. Cumming. I have been attracted by the rather unusual and ‘sexy’ title but it seems to be nothing more than an attempt at alerting the psychology community on considering point estimation procedures and confidence intervals, in place of hypothesis testing, the latter being ‘a terrible idea!’ in the author’s own words.
Some more quotes here. Then he says: “‘These are hardly new techniques, but I label them ‘The New Statistics’ because using them would for many researchers be quite new, as well as a highly beneficial change!’”
Of course the latter is not stated on the book cover.
That’s about as bad as writing a book with subtitle, “Why Americans vote the way they do,” but not actually telling the reader why Americans vote the way they do.
I guess what I’m saying is: Not everybody can write a good book title. Even the great John Updike and Gore Vidal couldn’t manage it, most of the time.
The author was interviewed on Australian radio’s “Ockham’s Razor”:
http://www.abc.net.au/rn/ockhamsrazor/stories/2011/3333636.htm
The clip is about 14 minutes long.
Best titles:
Hemingway: In Our Time, Men Without Women, To Have And Have Not, The Old Man And The Sea, The Sun Also Rises, Death In The Afternoon, A Farewell To Arms
Mickey Spillane: Kiss Me, Deadly, The Girl Hunters, The Body Lovers, My Gun is Quick, I, the Jury -
It must be getting towards the end of the year if we are doing ‘Top 10′ lists (or is that just a UK TV thing?).
So: The Big Nowhere, Neuromancer, Generation X, I, Robot and American Gods.
I’m guessing that views on a good title do depend to some degree on the enjoyment of the book.
Maybe we should judge a book by its title, after all?
There’s a very funny blog based on the notion that any book title can be “improved”:
http://betterbooktitles.com/
Updike was a reactionary hack. I predict your time in purgatory will be spent on an island with only a copy of “Marry Me” for company. After a thousand years of penance you’ll get a set of the Rabbit novels.
I would be interested in that book.
The “career reformers” are guilty of many of the same foibles as the “career witch hunters” (errorstatistics.blogspot.com, Sept. 26)—notably, they’ve made a career out of espousing “taskforces” and “rules of thumb” that keep a level of hysteria (against significance tests) going and, at times, commit the same fallacies they were intended to combat. But during a statistical navel gazing episode in Santa Barbara, Cumming created a neat excel tool to compute severity along side of computations of confidence intervals and hypotheses tests. I still use it.
I think Cummings has a point about it being new for the _intended_ audience.
For instance with meta-analysis (or just some way of addressing the presence of more than one relevant study) very few students in an introductory stats course (and just as few in a full Phd stats program) will encounter much at all on that.
For those who studied with Don Rubin, recall that he was involved in early meta-analysis work. (So was William Cochrane).
But I have never understood the motivation for effect size measures (in meta-analysis), that seems to make the mistake (as Fisher would put it) mixing up parameters with estimates (of parameters) or functions of parameters with function of reported summary statistics.
The real challenge, I believe, should be with identifying parameter to pool or partially pool. (The wiki entry on meta-analysis seems to split 70:30 in favour of effect size measures.)
Read some of Rubin material on that and it all seems to take that task as given – identifying and pooling effect size measures.
By the way not a productive area for stats research and development as their is nothing _formally_ different between n studies, n levels of a strata or even n observations.
Darn put that _e_ at the end of Cochrane, the Cochrane collabouration (named after Archie Cochrane) is totally unrealted (and makes the same _mistake_ of focusing on estimates rather than parameters.