I just finished reading The Aesthetics of Junk Fiction, by Thomas Roberts; and it’s the most thought-provoking book I’ve encountered since Taleb. (By “thought provoking,” I mean just that: These books provoked more thoughts from me than any other books I’ve read recently.)
It’s a book all about literary genres, and what people get out of reading detective stories, romances, science fiction, and westerns.
With genres on my mind, my reaction to receiving Kaiser’s new book, Numbers Rule Your World, was that this is the latest in the increasingly popular genre of pop-statistics books.
And then this got me thinking about different sorts of genres. Roberts discusses how a reader will go through detective stories, say, like potato chips–actually, he criticizes the food analogy, but you get the picture, with some people reading book after book by the same author or even the same series, others reading more broadly within a genre, and others dipping into a genre from time to time.
Books are different from T.V., where it’s so easy to just flip the channels and encounter something new. With books, it’s easier to stay within your preferred genre or genres.
Anyway, here’s the thing. People who love mysteries will read one after another. People who love science-fiction will read libraries of the stuff. But, even if you looove pop-statistics books, you probably won’t read more than one or two. Unlike mysteries, romances, westerns, etc., pop-statistics books are designed not for addicts but for people who aren’t already familiar with the area.
Because of my familiarity with applied statistics, I’m in some ways the worst possible reviewer for Kaiser’s book. It’s hard for me to judge it, because these ideas are already familiar to me, and I don’t really know what would work to make the point to readers who are less statistically aware. (Christian Robert had a similar reaction.)
The book has a blurb on the back from someone from SAS Institute, but I looked at it anyway. And I’m glad I did.
My favorite part was the bit about “How steroid tests miss ten dopers for each one caught.” I liked how he showed it with integers rather than probabilities–and I think there’s some research that this sort of presentation is helpful. And, more than that, I liked that Kaiser is sending the message that this all makes sense: rather than trying to portray probability as counterintuitive and puzzle-like, he’s saying that if you think about things in the right way, they will become clear. The story of the college admissions test questions was interesting too, in the same way.
There is an inherent tension in all these pop-statistics books, which send two messages:
1. The statistician as hero, doing clever things, solving problems, and explaining mysteries.
2. The method as hero, allowing ordinary people (just plain statisticians) to do amazing things.
Superman or Iron Man, if you will.
As a statistician myself, I prefer the Iron Man story: I like the idea of developing methods that can help ordinary people solve real problems. My impression is that Kaiser, professional statistician that he is, also prefers the Iron Man frame, although it can be hard to convey this, because stories work better when the heroes are humans, not methods. The next book to write, I guess, should be called, not Amazing Numberrunchers or Fabulous Stat-economists, but rather something like Statistics as Your Very Own Iron Man Suit.
P.S. I didn’t understand Kaiser’s description of how they handle the waiting lines at Disneyland. When I went there, you’d buy a packet of tickets, ranging from A (lame rides like It’s a Small World that nobody ever wanted to go on), through intermediate rides like the teacups, up to the E tickets for the always-crowded rides like Space Mountain. Apparently they changed the system at some point and now have something called a Fast Pass, which sounds like a take-a-number sort of system with beeper that tells you when it’s your turn to go on to your ride. Kaiser describes this as a brilliant innovation, which I guess it is–it seems like an obvious idea, but they certainly don’t do it in most doctor’s waiting rooms!–but he also describes it as more of a psychological trick in crowd management than an efficiency gain. That’s where he loses me. Sure, I accept the point that the rides have a finite capacity, so in that sense you can’t really shorten waiting times very much, but if you can wander around while waiting for your ride instead of standing on line, that’s a positive gain, no? Standing on line is generally pretty unpleasant.
P.P.S. Do youall like this kind of rambling blog that goes through several ideas, or would it be better for me to split this sort of thing into multiple entries (for example, a review of Kaiser’s book, a question about Disneyland, the discussion of genres, and the Superman/Iron Man issue)? I kinda feel that multiple entries would work better on the blog; on the other hand, the sort of single wide-ranging discussion you see here is more interesting in a published review. Maybe I can send this to the American Statistician or some other such publication.
I vote for the rambling blog with several ideas. But then I read this mostly for pleasure.
As long as you keep blogging I am fine either way.
A quick question. I teach intro statistics and I am always interested in how other people approach the subject. I do admire the way you present your material and thats why I keep coming back to this blog.
My question involves your comment about not making statistics counter-intuitive or puzzle solving like. I believe I have also seen this idea in one of your presentations. Now I know making statistics very counter intuitive – very challenging will (as I experienced) not lead to productive results. However wouldnt the student go "What am I learning, I know this stuff already" if we introduce these concepts without challenging the mind to go an extra step? For instance I have an example similar to the doping one you mention above, that I go over in class. I try to make the students understand P(Test=+|Disease) NE P(Disease | Test =+) and they tend to get surprised the numbers are so different. I try to make them recognize why this occurs, without mentioning the B word. Hopefully this allows them to understand rules of prob better. How do you approach, say teaching descriptive statistics, without boring them but also not losing them?
Rambling can work well, if you've got a good title to either preview all the content or one of the most interesting subtopics.
I like the man as hero versus method as hero distinction. I'll have to remember that one. While I liked the Incredibles I wasn't a huge fan because it was a bit too far on the man as hero side for me by setting up the method type as the villain in part for being a method type.
yay for rambling from me, too.
Also – Iron-man vs. Superman is an excellent metaphor.
I'll add it to my list of ways how to bash Freakonomics ;-).
The efficiency gain of Fast Pass depends on how you spend the wait time. Many people will get a Fast Pass for one ride, then wait in line for another. At the end of the line, you get two rides in quick succession (you're at the front of this line and your Fast Pass timer is coming due). But of course, because there are Fast Pass riders on the ride you actually waited for, you might have waited twice as long as you would have. If you spend your wait time leisurely strolling about you had a more pleasant wait, but because other people are standing in two lines, you'll end up riding fewer rides then if everyone had physically stood in each line.
Regarding the focus of blog posts, I'm partial to rambling ones. Focusing on a single topic you can lose all the connections to other ideas that excited you about the topic in the first place.
If you ever decide to write the "Iron Man Suit" book, count me in on your team.
Bring on the fan-boy bait.
Iron Man's just a cheap Marvel knockoff of Batman!
If Superman were a scientist, he'd be a physicist, not a statistician!
PS: Robert Downey Jr. was awesome as Iron Man — better than any of the Batmans. If only they'd foregone the Jar-Jar-Binks-like robot, it'd have been a near-perfect Marvel superhero genre movie.
I think I will steal the Superman-Ironman analogy and I like the rambling posts because they show how (1)the ideas are related among themselves (e.g. applications that could have been ignored) and (2)they are fun to read. As a behaviour scientist I am also intrigued by how the brain connects one idea to the other.
I think it depends on the way your readers like to take it all in.
I like to read through a bunch of information at once, rather than checking back for small posts every day, so I am more inclined towards the longer versions with lots of ideas. Fewer introductions, conclusions and so on. That said, whatever you like writing I seem to like reading, so go for it.
Love the rambling, several ideas thing. Interesting to see your thought process.
i too vote for rambling. its been great sofar…
Andrew: thanks for your nice review. I too vote for rambling. This is a blog after all, not a journal article.
Your Superman/Ironman analogy strikes a chord. Certainly, in my work, I love projects that have methodological challenges but as you stated, it's really hard to make methods engaging.
The SAT section is a good illustration of this. I'm very happy about that section because I think the lay reader will get very little out of Curley and Schmitt's original research article, and what I've written makes their ideas clear. However, I chose to stop there, and not discuss the methods (standardized differences vs. IRT models). I just haven't figured out how to do that effectively without losing the reader in technicalities. Maybe if there is a sequel, I will figure it out. Or maybe the book on methods has a different target audience.
On the Fast Pass, I think the psychological aspect is that people don't like to stand around – if we are allowed to roam, even aimlessly, we don't think of it as "waiting". Like Paul said, with fixed service rate and an amount of work in front of us when we arrive, it's just not possible to "jump the line". Another great example that didn't make it to the book is that they deliberately put baggage claim belts very, very far from arrival gates because we don't count "walking to the baggage claim areas" as "waiting".
> Sure, I accept the point that the rides have a finite capacity, so in that sense you can't really shorten waiting times very much, but if you can wander around while waiting for your ride instead of standing on line, that's a positive gain, no?
In the Disney parks, more space is provided for comfortably buying merchandise and food than is provided for making the attraction lines non-claustrophobic. The FastPass takes advantage of this, to the delight of Disney shareholders.
I like the rambling entries. I'm not sure if I like them 4 times as much as 4 smaller entries.
I prefer the rambling entries.
i very much agree to the idea of statistics as a suit to put on. The reason we do statistics is not for gaining other people's respectful look, or to dominate a very concentrated area so we become more important, building borders so that not many people has the access to it – just as the accountants and doctors do.
I go for separate blogs idea. it might be easier for people to comment too.
I do read lots of popular intros to statistics, trying to think about what would work in a novel. The thing that baffled me for a long time was that they were pretty consistent in being thin on graphics – startlingly worse than just about any introductory statistics textbook. I couldn't really understand this, because good graphics can often make a point clear in a way that no amount of user-friendly anecdotage can. I don't know that any statistical textbook lives up to the ideals of Edward Tufte, but they go much further in that direction than any popular book I've seen. I refer to my bafflement in the past tense because at some point the penny dropped – and as it happened you had a post which was relevant. Producing good graphics is HARD. And showing readers how to produce their own datasets and play around with statistical software is also HARD. It's hard for me to review this kind of book (I tried once) because they all have the same shortcomings, and I can't quite believe that a general readership will rush out and buy a textbook on my say-so. Argh.
Bob: I thought about Batman but decided that Iron Man worked better for the point I was making. Sure, Batman isn't superhuman, but he has all sorts of impressive physical abilities. Renting the Batmobile and strapping on the utility belt won't make you Batman the way that putting on the Iron Man suit will make you Iron Man. (At least, that's my impression; I haven't actually seen any of the movies in question).
Helen: I'll try to do it right in my forthcoming intro statistics book. On the plus side, I have a good mix of theoretical, computational, and applied experience and a real desire to get things right. On the minus side, I haven't actually started writing the damn thing yet.
Rambling works for me, at least occasionally.
Also, I don't know if he's the first to do it, but Gerd Gigerenzer has spearheaded a good amount of the research on the efficacy of 'natural frequencies' in understanding probabilities. Here's one example (that I haven't read, but assume is reasonable):
http://mdm.sagepub.com/cgi/content/refs/29/3/368
Gigerenzer came to IU a few years back to discuss some of this research, and it was one of the best talks I've ever been to.
Yes, go for the "rambling blog", always. Your stuff is good. If I want structure, I'll buy your books, and do.