Personal Science Book

A thread to discuss Personal Science: Learning to Observe, a book by me, Thomas Blomseth Christiansen, Jakob Eg Larsen, Martijn de Groot, Steven Jonas, and Sara Riggare.

(The early release version of the book can be purchased here: Personal Science: Learning to Observe)


Will this book be made available on Amazon for purchase?

Great book. Thanks for doing this!

After reading it, I have a number of comments but first I’m wondering who you are intending as the audience? Or to put it another way, which aisle of the bookstore? If it’s self-help, then I thought the meat was in the “Observing” chapter and could use more discussion and examples. If it’s a health and wellness book, then there are numerous good before/after examples but I wonder if you could organize more by subject (e.g. sleep, digestion, mental health, etc.) . If it’s more of a popular/general science book, I’d like more references and more cool unexpected examples (e.g. the Shangri La Diet)


Thanks Richard! This feedback is already very useful. It may not be clear enough what kind of book this is! The answer is that we intend it to be self-help; that is, option 1 on your list. The meat is definitely in the Observing section. You’ve seen a lot of talks and done a lot of self-tracking. Are there some projects think should be included (not excepting one or more of your own, please don’t be hesitant to point to your own work), or some issues that deserve fuller explanation?

Eventually we do hope to have the book widely available through all the regular channels. This is a pre-release to the Quantified Self community of an almost finished version.

1 Like

I’d like to see more examples of A/B testing, which I think is one of the best techniques. Something like Seth Roberts’ butter experiments (or my replication using fish oil) where you try something for a while, then a flush period, then try somethings. One nice thing about A/B testing is that there are easy statistical analyses you can do to measure how likely any effect is due to chance.

Sleep tracking is one of the most popular QS projects. I really like this summary from Tomáš Baránek, unless you’re you’re trying to stick to content from (I’d have to search, I’m sure there are great examples there).

That Azure Grant study of cholesterol throughout the day is a QS classic that shows the importance of sampling frequency. If that’s what you mean by “cadence”, perhaps that section could include that study as a warning about the importance of measurements at the right times of day, frequency, etc.

Instruments might include a discussion about the importance of consistency across devices. You can’t assume “heart rate” means the same across Fitbit and other devices for example.

A section on spurious correlations, plus something about how even a negative result is a good learning experience. (Any of the Zenobase home page examples are good: eg. phases of the moon).

This sort of a book begs for more discussion of the *how – people will want specifics about which app, which which device, nuts and bolts for specific experiences – but maybe that’s the next book!

p.s. note typo in Features section “explore it’s features”


" are hungry for technical advice they can apply right away. "
“working through what might be trackable and checking
whether your proposed phenomena satisfy the criteria”
“There are uncountably many. There is no way to
make a list of “things in the world.””
" skill required to bend these instruments to our own ends goes way up. "

Why not wiki, ie other people’s advice? Lots of the ‘background’ phenomenon are pretty common and should be pretty easy to just read about and use. Rare needs would not apply much. In my case finding the wearable accelerometer was the issue.

I am not sure about book’s definition of “Variability”. I think rare but very serious changes such as moving to a new country was mentioned. However all continuous variables have some variability and for many reasons that can get complicated. I think this should use terms like ‘signal to noise’ rather than variability though the amount of that inside a variable will only be found out after data gathering and analysis. Consider the learning curve of a cognitive test; It must be removed to be of use comparing days because it is absolutely the greatest source of variance. Target / main variable of interest should vary enough to matter to user but this fact is not useful because proxies would probably not be any better.

Foreground vs background tracking. I first thought this was about active vs passive tracking. Maybe primary vs secondary would be clearer?

For many/most people, the act of tracking itself (active, or even passive) is enough to change behavior. This might warrant some more discussion?

In any case, it’s a perfect book for anyone who wants to deepen their understanding of this topic a bit. It’s no “Personal Science for Dummies”, but also not overly academic, so just about right!

Curious if there is any data indicating how many people actually change behavior based on tracking data. Does anyone have any research on this?
We often get investor pushback that tracking apps won’t change behavior – would be great to have some data to refute :slight_smile:

Here’s an example of a controlled trial for weight loss with a tracking app, and a meta-analysis of interventions using a Fitbit.

I don’t think there is any question that self-tracking can alter behavior. What’s less clear is how often self-tracking results in long-term behavior changes.