Peter Zhang

Tracking Time

Introduction

Hey chump. You feeling inefficient? Overwhelmed? A little bit of both?

It’s hard to be productive. Like a tennis machine, technology-powered society constantly pelts us with distractions and temptations. Wading through your daily to-do list can be a daunting challenge.

Let’s suppose that you nevertheless manage to succeed. There’s a good chance you’re still unhappy. Productivity tips, alarm clocks, and dog mentality can only help you do more things, more efficiently. It won’t help you allocate your time where you really want it; it definitely won’t help you achieve balance and avoid burnout.

In June of last year, I faced that exact predicament. In search of productivity tips, I stumbled on a lifestyle-changing blog post by James Ma. The post outlines a lot of instructions for living a better life, and one of them is to track your time. I’ve followed through religiously on this advice, and I suggest you do too.

What is it? It starts with a basic observation: time is valuable. Whether you choose to sleep, study, or watch Minecraft gameplay, you are making an investment with your time. The cost watching a Technoblade stream—I know you do it too—is the all the other options you set aside: going for a jog, reading a book, starting a new project.

We should try and make the best investments we can with our time. But if I asked you how many hours you slept last week, you might give me a ballpark estimate of 7 to 9 hours a day. If I asked you how many hours you spent binging YouTube or scrolling through Instagram last Tuesday, you’d have an even tougher time.

Any solution to poor time allocation must begin by clarifying the problem. Time tracking does just that: it gives you an overview of what you actually spend your time doing.

Guide

Setup

So how do you do it? A lot of this is explained by James’ post so I’ll won’t spend too much time here.

The first step is to start keeping a daily log. I used Google Keep—it’s free. Each day, I make a new note that starts at midnight. As I go through my day, I write down the times when I start a new activity. I keep 15-minute intervals for simplicity, but you get as granular as you want. Don’t make too many categories.

The second step is to transcribe it. I keep all my time tracking data in a Google Sheet. Make a row for each activity (this is why you want to keep your list of categories short) and then group them however you’d like. I recommend sleep, work, fo(od)sho(wer)trans(portation), exercise, personal, and media (YouTube, Netflix, etc.). Keep a “missing” row so you can easily tell if you’re not accounting for all 24 hours. For additional insights, you can also add mood, various goals, and more. If you like my template, you grab a copy here.

Commitment

If you haven’t realized yet, it’s a pain in the ass, at least at first. Following through on this is time-consuming and sometimes frustrating. But that’s the thing—collecting all this data is only helpful if you do it every day, and the key insights come from the days when you feel like shit.

Start at the beginning of some month and commit yourself to just getting through those first four weeks. A shorter time frame can make it easier to motivate yourself. You can make it the transcribing step a lot easier by building it into your morning ritual. For me, transferring yesterday’s log is just what I do before making my to-do list.

I want to point out here that time tracking isn’t just about analysis; it’s also about changing how you think. Logging my time has made me a lot more conscious of what I spend my time. It makes me a lot more appreciative of the time I have and lot more wary of throwing away my time on junk activities. Once you make that mindset shift, time tracking becomes a lot more natural.

Analysis

My time tracking journey began in June of 2020. In mid-September, I began keeping note of the different types of work I was doing (instead of just a general “work” category). I decided to use the data starting then.

My dataset spans from September 19th, 2020 to January 6th, 2021. That’s 110 observations. Each day has a breakdown of my time across things like sleep, work, and personal improvements (planning, reading, meditation). It also includes “mood,” a totally subjective rating of how I felt I day on a scale of 0-10. Yeah, not great, but this isn’t a paper.

Here are the summary stats for the different categories.

  Mean SD Min Q1 Median Q3 Max
Sleep 8.33 1.07 5.50 7.75 8.25 9.00 12.00
Work 9.45 2.34 3.75 8.00 9.50 11.00 14.75
Class 1.64 2.02 0.00 0.00 1.00 2.75 9.75
Debate 3.42 3.03 0.00 1.25 2.75 4.75 13.75
Voyager 2.02 2.29 0.00 0.00 1.38 3.00 9.00
BER 0.62 1.22 0.00 0.00 0.00 0.75 6.75
CSM 0.37 0.59 0.00 0.00 0.00 1.00 2.25
FoShoTrans 1.60 0.76 0.00 1.00 1.50 2.00 4.00
Exercise 1.31 0.75 0.00 0.75 1.25 2.00 3.50
Planning 0.55 0.62 0.00 0.25 0.25 0.75 3.50
Social 1.00 1.04 0.00 0.00 0.75 1.75 4.75
Reading 0.49 0.53 0.00 0.00 0.25 0.75 2.75
Meditate 0.16 0.13 0.00 0.00 0.25 0.25 0.50
Media 1.11 0.88 0.00 0.50 1.00 1.50 4.25
Mood 5.80 1.61 2.00 5.00 6.00 7.00 9.00
Personal 2.20 1.30 0.00 1.25 2.00 3.00 6.50

I began by plotting different activity categories against each other, where each day is colored according to my mood (orange = happy, blue = sad). Four of those plots are below. There seems to be a negative correlation between sleep and work as well as personal and work, which probably just shows that there’s a time tradeoff. High levels of personal time and exercise seem to improve my mood. Meanwhile, lots of sleeping seems to dampen my mood. The lack of a clear relationship between class and sleep is good news—I guess I’m not losing sleep for class after all.

Next was the fun part. I conducted multivariate regression on different groups of features to see how well they predicted my mood. The “Lifestyle” model used aggregated time allocation across sleep, work, exercise, foshotrans, personal, and media. The “Work” model used my time spent on different work commitments, like class, Voyager (my consulting club), or—was most interested in this—debate. The “Habits” model zeroed in on just sleep, media, and the breakdown of different personal improvements (planning, readnig, and meditating)—it was really just the things I perceived as being good or bad. Here are the results.

LifestyleWorkHabits
(Intercept)4.4676.467***6.855***
(36.989)(0.405)(1.126)
Sleep-0.104-0.191
(1.540)(0.129)
Work0.099
(1.546)
Exercise0.384
(1.581)
FoShoTrans0.088
(1.571)
Personal0.438
(1.520)
Media-0.316-0.404*
(1.528)(0.161)
Class-0.123
(0.080)
Voyager-0.090
(0.068)
Debate-0.146**
(0.054)
CSM0.213
(0.257)
BER0.217
(0.129)
Planning0.650**
(0.234)
Reading0.242
(0.262)
Meditate3.160**
(1.147)
R-squared0.2160.1400.269
adj. R-squared0.1710.0990.234
p0.0000.0070.000

Significance: *** = p < 0.001; ** = p < 0.01; * = p < 0.05

Interesting! The first model had no significant results. Watching YouTube and oversleeping make me slightly upset, and doing more personal stuff make me slightly happy. Fine.

The second model gets more interesting. It seems like class work and Voyager have insignificant negative effects, while tutoring and economics writing have insignificant positive effects. But debate has a very significant negative effect on my mood. For every seven hours I spend on debate, my mood declines by a whole point—I probably should cut back on those debate tournaments!

The third model is the coolest. It has the highest explanatory power of the three, with an adjusted R-squared of 23.4%. Media, no surprise, has a significant negative effect. An extra hour of media will tank my mood by 0.4. Planning and meditation, on the other hand, have big positive effects. Just forty-five minutes of planning will boost my mood by about 0.4. And, to my surprise, an extra hour of meditation is associated with a 3.16 mood-boost. Missing out on my fifteen minutes of meditation really makes me cranky.

Conclusion

Time tracking can help you reclaim some control over your time. By quantifying what you do and when, you get an objective, birds-eye view of your lifestyle. If you’re a geek like me, you can do some cool stuff with the data. But you definitely don’t need regression to figure out what you’re spending too long on, or what you should do more of. Moreover, even if you don’t ultimately find a revelation in your time data, the process of tracking it will make you notice and appreciate your time—I think that’s worthwhile as well. I hope you give it a try; I really hope you get dope regression results.

Built with Jekyll on the Swiss theme.