Book titles

My collaborators and I have had some successes and some failures; here are some stories, with the benefit of (varying degrees of) hindsight.

“Bayesian Data Analysis.” We thought a lot about this one. It was my idea to use the phrase “data analysis”: the idea was that “inference” is too narrow (being only one of the three data analysis steps of model-building, inference, and model checking) and “statistics” is too broad (seeing as it also includes design and decision making as well as data analysis). I hadn’t thought of the way that BDA sounds like EDA but that came out well, even though the first edition of BDA was pretty weak on the EDA stuff–we fit more of that into the second edition (in chapter 6 and even in the cover). Beyond this, I was never satisfied with “Bayes” in the title–it seemed, and still seems, too jargony and not descriptive enough for me. I’d prefer something like “Data Analysis Using Probability Models” or even “Data Analysis Using Generative Models” (to use a current buzzword that, yes, may be jargon but is also descriptive). But we eventually decided (correctly, I think) that we had to go with Bayes because it’s such a powerful brand name. Every once in awhile I see the phrase “Bayesian data analysis” used generically, not in reference to our book, and when this happens it always makes me happy; I think the statistical world is richer to have this phrase rather than the formerly-standard “Bayesian inference” (which, as noted above, misses some big issues).

“Teaching Statistics: A Bag of Tricks.” Should’ve been called “Learning Statistics: A Bag of Tricks.” Only a few people want to teach statistics; lots of people want to learn it. And, ultimately, a book of teaching methods is really a book of learning methods. Also, many people have told me that they’ve bought the book and read it. I actually think it’s had more effect from people reading it than from people using it in their classes. Sort of like one of those golf books that people put by their bedside and read even if they don’t get around to practicing and following all the instructions.

“Applied Bayesian Modeling and Causal Inference from Incomplete-Data Perspectives.” The title seems fine, but something went wrong in the promotion of this book. Xiao-Li and I collected some excellent articles and put a huge amount of effort into editing them. I think the book is great but it hasn’t sold a lot. Perhaps we should’ve structured it slightly differently so it could’ve been used as a course book? And of course we shouldn’t have published with Wiley, who are notorious for pricing their books too high. (I notice they now charge $132 (!) for Feller’s famous book on probability theory.) Why did we go with Wiley? At the time, Xiao-Li and I thought it would be difficult to find a publisher so we didn’t really try shopping it around. In retrospect, we didn’t fully realize how great our book was; we were satisfied just to get it out there without thinking clearly about what would happen next.

“Data Analysis Using Regression and Multilevel/Hierarchical Models.” The awkward “Multilevel/Hierarchical” thing is Phil’s fault: I wanted to go with “multilevel” (because I felt, and still feel, that “hierarchical” can be seen as implying nested models, and it was very important for me in this book to go beyond the simple identification of multilevel models with simple hierarchical designs and data structures), but Phil pointed out that “hierarchical” is a much more standard word than “multilevel” (for example, “hierarchical model” gets four times as many Google hits as “multilevel model”). So I did the awkward think and kept both words. (And Jennifer was fine with this too.) Also we needed to put Regression in there because a multilevel model is really just regression with a discrete predictor. And Data Analysis for the reasons described above. The book has sold well so the title doesn’t seem to have hurt it any.

“Red State, Blue State, Rich State, Poor State: Why Americans Vote the Way They Do.” I think this was a mistake. First, as some people have pointed out and as we realized even at the time, we don’t actually say why Americans vote the way they do. I really wish we had chosen our other candidate subtitle, “How Americans are Polarized and How They’re Not.” Beyond this, I’m actually down on the whole Red State, Blue State thing. Sure, it’s grabby, but I fear it makes the book seem less serious. Given that we didn’t become the next Freakonomics and we didn’t sell a zillion copies, if I could go back in time I’d give it a more serious title, such as, hmmm…, “Geographic and Demographic Polarization in American Poliitcs”–no, that’s too serious-sounding. Maybe “Democrats and Republicans: Who They Are, Where They Live, and Where They Stand on the Issues.” Or “American Voters, Red and Blue: Who They Are, Where They Live, and Where They Stand on the Issues.” Something that is a bit grabby but conveys more of our research content. (Many people were misled by our title into thinking the book was merely a retread of our Red State, Blue State article, but really it was full of original research that, to this date, has still only appeared in the book.)

“A Quantitative Tour of the Social Sciences.” I can’t imagine a better title for this one. And I love the book, too. In addition to having wonderful content, it has a great cover that was contributed by a blog commenter (who I still have to send a free book to; sorry!). We’ve gotta do a better job of promoting it, but I’m not quite sure how. Here’s a nice review.

I have a few more books in (various stages of) the pipeline, but I’ll hold off telling you their titles until they’re closer to done.

8 thoughts on “Book titles

  1. I didn't know you have a book called "Applied Bayesian Modeling and Causal Inference from Incomplete-Data Perspectives." And I am a reader of your blog for about one year. Maybe I missed some post about the book, but I don't see a lot of references to it in your blog.
    On the other hand, Bayesian Data Analyis and ARM deserver a lot of mentions. By the way, I bought both books.

    I will take a look at amazon about this "new" book.

  2. I took a look at amazon and the book "Applied Bayesian Modeling and Causal Inference from Incomplete-Data Perspectives." seems really impressive. Though, as you pointed, the list price is $120 (now they're selling it for $96).

    If it were cheaper (around 50 bucks) I would order one. Maybe is a matter of add and price!

  3. I agree that the 'problem' of "Applied Bayesian Modeling and Causal Inference from Incomplete-Data Perspectives" is its high price. This is also true for many (most) stats books :(

  4. "I felt, and still feel, that "hierarchical" can be seen as implying nested models"

    Yes! And that's exceedingly confusing for applied users of statistics who often simultaneously get hit with explicit discussion of nested model comparisons, and varying slopes and intercept models!

    Incidentally, what's your preferred terminology for a LMM or GLMM with crossed random effects? Are these still an instance of multilevel models?

    And while we're on the subject, do you view residuals as varying intercepts at the level of individual observations?

  5. I bought Red State Blue State partly because the title seemed informative. But then it is my area of interest. As far as likely crowd pleasers, I like What's the Matter with Connecticut. Or perhaps David Broder is a Big Fat Idiot. (I know that's not right but if the point is curb appeal…)

Comments are closed.