Skip to content
 

If textbooks in the form of comic books are useful, what does that say about how textbooks could/should be written?

Aleks links to “The Manga Guide to Statistics” and commenter David Warde-Farley links to the similar-looking “Cartoon Guide to Statistics.”

My thoughts

Based on the example shown above, the point of the comic-book format seems to be to allow a punchy, power-point sort of delivery. The picture conveys essentially no content, which would suggest that the entire contents of a 222-page comic book could be presented in a 10-page pamphlet of text. The remaining 212 pages are essentially a reader-friendly trick to get students to turn the pages. It’s the printed analogy to a power-point presentation.

So . . . let’s take the customers’ word for it that these cartoon guides are good. If so, this suggests that the useful content of a typical introductory statistics book can be captured in 10 pages. And, if this is the case, it in turn suggests that textbook writers could do a better job with those other 212 pages. Maybe it would be better to have a 10-page textbook and 212 pages of examples? Presumably a good textbook author could do better than those silly cartoons.

It’s a tricky issue. Thinking about my own 600-to-700-page textbooks, it’s hard for me to see what I would cut to bring it down to 30 pages. At the same time, the actual material that students learn in the class can probably be written in 30 single-spaced pages, so maybe it would be a good idea to try to pull that out.

(My books aren’t directly comparable to these comic books, as I’m covering higher-level material. But the issues of presentation can’t be that different.)

15 Comments

  1. Blake says:

    It might take a comic book 20 pages to have as much substance as 2 pages of textbook material… but if I can read it 10 times faster and hold onto the information longer, I'll still call it a win for the comic book. (assuming your 21.2:1 ratio is exaggerated.)

  2. Daniel Lakeland says:

    I think a hierarchy of presentation is a really important thing that textbook writers fail to appreciate.

    The 10 page summary, using general descriptions and the minimum of essential equations is a very important tool. It gives the reader a sense of what direction to think in. It answers questions that the reader may have like "why do we care" and "where is this used", and "how does this relate to something in the rest of the book" etc.

    I think textbooks should have a 10 page introduction that tells you what you will learn throughout the course. Then each chapter should have a 1 to 2 page introduction telling you what you will learn in this chapter and how it relates to previous and future chapters. Then give the detailed explanations of the material in the chapter.

    The key to success in this kind of approach is to answer the right questions in your introductions. Many introductions are more an opportunity for the authors to show how broadly informed they are, or how interesting they think the material is, than to inform the student of the plan of the course and the "take home" messages they should learn.

  3. Even though I was a math major as an undergrad, I found Bayesian Data Analysis very difficult to read, even after reading and understanding Larsen and Marx's Mathematical Statistics. It wasn't because the math per se was hard, but rather because so many steps were presupposed and left out in the discussion. I suppose I was looking for a longer introduction that connected more of the dots for me.

    The incredibly overloaded notation doesn't help, where variables are used for means, and for subscripts on variance or other related parameters. This may be due to my computer-science-centric way of thinking. Now I like just using p for everything, and even use overloaded subscripts; I now find things like Jeff Gill's book's notation with a different function name for likelihoods, priors and posteriors really hard to follow. So it's probably partly just what you're used to.

    Now that I understand the material, BDA is providing a whole new level of insight, and I'm able to work through all of the elided steps on my own because I can tell where everything's supposed to be going conceptually. I usually find it faster to read a book that actually goes through most of the steps explicitly.

    I found Albert's pop baseball stats book Curve Ball a very nice intro to topics like posterior distributions, because I already understood the examples (unlike say, home radon or rat tumors), and found the leisurely presentation of lots of examples easy to follow. More importantly, he developed the intuitions in a way that made things "click" for me in a way that nothing I'd read before did.

    If I remember correctly, I found even the first hierarchical beta/binomial example in the BDA book very mysterious, because it used a notion of "moment matching" estimation that I'd never seen before and was never explained. I found myself having to read the book with a search engine and the web by my side. Other topics, such as EM and Gibbs sampling, were presented at a very abstract level for someone first trying to grapple with the intuitions (or at least someone with my level of comprehension). More examples would've helped me immensely, especially if they were even more explicit and step-by-step than the ones provided.

    I knew almost nothing about regression other than the chapter in Larsen and Marx when I read Data Analysis Using Regression and Multilevel/Hierarchical Models. But I found it very easy to follow because of the code-based examples, which require a degree of explicitness that a regular textbook doesn't. In fact, with no other help, I found myself able to install R and BUGS using the instruction and even create my own latent variable item-response like models and fit them in R and BUGS. I understand the equivalence of models with redundant variables, but still don't understand conceptually why they make the samplers faster in BUGS in some cases.

    I don't think comics would've helped, despite being a huge fan of the genre in general.

  4. vit_r says:

    Pleae notice the lines, the hand and the big manga-eyes in this picture. Manga is about emotions. If some information is bound to emotions, it cannot be forgotten.

    Western comics are more like power-point presentation.

    "a reader-friendly trick to get students to turn the pages" is the interest. The form is not important. I think the key are the storytelling skills. I had seen boring mangas and exciting textbooks.

  5. vit_r says:

    Please notice the lines, the hand and the big manga-eyes in this picture. Manga is about emotions. If some information is bound to emotions, it cannot be forgotten.

    Western comics are more like power-point presentation.

    "a reader-friendly trick to get students to turn the pages" is the interest. The form is not important. I think the key are the storytelling skills. I had seen boring mangas and exciting textbooks.

  6. Hadley says:

    It seems a little unfair to judge a book by a single picture!

  7. Brian M says:

    I noticed the boggley eyes. Also that the finger pointing in space "cannot be measured" while the arm at a triangle point "can be measured".

  8. Helen DeWitt says:

    Several years ago I came across a rather peculiar book, "Who Was Fourier?", which had been produced by members of the Hippo Club in Japan – the Hippo Club being, originally, a group of amateur language learners. Members of the group tackled Fourier analysis, each taking on a different chapter and starting from the very basics; a professional mathematician checked the chapters to make sure there were no howlers.

    Now, this book does not (unsurprisingly) get you very far in Fourier analysis. It shows something rather different. On any given page, about 90% of the space (I'd guess) is devoted to reassuring the student that this is not scary. There is only a tiny nugget of actual mathematical material, and what there is comes in very tiny increments, think a flight of stairs each 2 inches high… The nuggets are embedded in dialogue (Newton is addressed as Sir Newton, awwwwwwwww, and says things like 'By Jove!', awwwwwwwww) and funny pictures.

    What does this really tell us? Well, in this book the ratio of noise to signal LOOKS really high: there's maybe a two or three lines per page with mathematical contents, and the whole rest of the page is essentially saying: Look, don't worry, this is fun, don't be scared! Keep trying! Gambatte! In a "professional" book on mathematics, on the other hand, the ratio of signal to noise looks really high: it looks like the mathematical equivalent of a really good fruitcake, something so stuffed with alcoholically enhanced dried fruit there's barely a mouthful of flour in the thing to hold it together. But there's actually much MORE noise, for a certain class of reader. The text is saying: This is incredibly difficult, this is not for the likes of you, don't even bother trying, you'll never understand this in a million years, get lost you waste of space.

    Members of the Hippo Club apparently got on well with Fourier analysis, or at least as much as they tackled. The thing that's not clear to me is, if someone gets much further than expected using this (of course) extremely inefficient method, what happens next? Is the student then at a point where s/he could take on T W Korner's much meatier Fourier Analysis? If not, is there some extra amount of explanation plus encouragement that would get the student to that point?

    So as far as statistics goes, what I'd want to know is whether the manga approach helps people who would otherwise give up to keep trying – and, if so, what it would take to bring such readers to the point where they could persevere with a text that did not offer constant reassurance and entertainment to the supposedly nervous reader. (Further research needed; how lovely.)

    Having said all that, I do think this sort of approach looks plausible partly because books on statistics don't normally take on board Edward Tufte's work on the visual display of quantitative information. The book I go back to again and again is Jim Pitman's Probability; the graphics are spectacularly good, and this reader, anyway, found that they invited one to go back to things one hadn't quite grasped until it all actually made sense.

  9. Andrew Gelman says:

    Helen: Good point. There's room for low-information, high-motivation intro books as well as more densely packed texts such as Bayesian Data Analysis. This still raises the question of what should be done for the introductory course. If, as I suspect, only 10 pages of material will be learned in any case, these intro textbooks should probably be radically restructured.

  10. Aleks Jakulin says:

    The sample picture has no content if you already understand the notion of measurability: if a reader doesn't know it, he will think a little bit and perhaps learn a new concept that allows him to wonder if something can or can't be measured.

  11. Andrew Gelman says:

    Aleks,

    Yes, but the picture has no content beyond what is in the words.

  12. Also, this picture is rather bad because I think it has been translated poorly – it's using measurability to differentiate between quantitative and qualitative variables, which isn't a connotation that measurable has in English.

  13. vit_r says:

    Hadley,

    if you find the book on Amazon, you can read 5 pages of the "story". This picture do not explains text but is a big "exclamation mark". A lot of textbooks use such pointing hand to show which information is most important.

  14. Andrew Gelman says:

    Hadley: Yes, I suspected as much. I think the distinction between qualitative and quantitative variables is much overplayed. Many variables that are categorized as "qualitative" are in many respects quantitative. Deb and I discuss this in our Teaching Statistics book.

    This is a problem I often see in statistics textbooks: some minor, and perhaps incorrect, point gets emphasized, and that is what students take away from the course. This is not really a strike against the cartoon format; rather, it's no surprise that such a book tha tis unconventional in its presentation will take a very conventional take on the topics to be covered.

  15. Andrew: I'd agree – I think the cardinality of the sample space is much more important, as it basically determines the type of mathematical tools that you need to bring to bear.

    More broadly, however, I think the "manga guide to statistics" should be praised for trying to do something different. There are plenty of other intro stats books that cover the same old tired material in the same old tired manner, and at least these authors have tried to present it in a interesting format. I don't think they have succeeded terribly well, but at least they tried!

    I think the manga format can be successful. One of my favourite books is The adventures of Johnny Bunko" which is a career advice book in manga format. I strongly recommend taking a look at a copy.