# taw's blog

The best kittens, technology, and video games blog in the world.

## Tuesday, December 22, 2015

### Review of The Manga Guide To Statistics by Shin Takahashi

It's been a long time since I last did any book reviews on this blog, or for that matter any other reviews. I usually just use Twitter or G+ for this as it's faster.

Here's a book I bought for the lulz - The Manga Guide to Statistics by Shin Takahashi and it turned out to be better than novelty value I expected.

### Plot

The book follows a story: dad invites a cute coworker for dinner, who does some statistics at work for marketing reasons or something like that. The high schooler daughter wants to meet him again because he's cute, so she gets the idea in her head to ask her dad to get someone from work to tutor her in statistics (because she's totally interested in what dad is doing at work, honest) - and the dad obliges except getting some completely different dude as tutor. Basically it would be not out of place in any high school manga, but then again the last one I seriously followed was Aa! Megami-sama back when I was in actual high school million years ago, so what do I know.

The story is not breaking any grounds, but it's way more amusing that in any statistics textbook.

Also I'm reasonably sure that's already more plot than Mad Max: Fury Road had, and that was a really good movie.

### Educational Value

The level of statistics in the book is fairly introductory, and unfortunately the book has some annoying mistakes, such as:
• completely incorrect explanation of what it means to reject a null hypothesis (also knows as the most common error in statistics textbooks)
• assuming normal distribution for data which is definitely not normally distributed such as normalized school test scores (also extremely common error)
Most concepts like histograms, distributions, correlations, etc. are explained relatively cleanly, but book's attempt at explaining what chi squared distribution is used for and what's the meaning of independence testing feels like a total miss, and I doubt anybody would have more than just a vague idea what the hell they are just read after that.

There's a bunch of math formulas, and I was not impressed by typography - with parentheses not being correct size to match enclosed content. Of course one could make a valid objection that this is a manga not a math textbook, and it was probably not typeset in TeX, which is fair enough I guess.

Am I the only one annoyed by bad typography here?

Of course this leads to obvious followup issue - is it even possible to teach non-Bayesian concepts like null hypothesis testing at this level beyond just "run this calculations, don't worry what they mean"?

Another problem with the book is that it probably focuses too much on the kind of silly content that predates spreadsheet software like formulas and looking up stuff in distribution tables.

Oh the books also has exercises for the reader and appendix on Excel, which are both potentially useful too.

### Summary

Overall I'd say the book is probably not the most amazing textbook, but it'd be a pretty sweet novelty gift. Now non-Bayesian statistics is a particularly messy subject to teach, so I guess other books from the series might do better.

tl;dr 4/5 stars