What I Learned From Tracking My Anxiety

With some real world data.

At work, we are pious worshipers in the church of data. We rely on it to track progress and to make decisions. What gets measured, gets managed, all that.

But we rarely make good use of data to improve our lives. We often just kind of fumble around, like a drunk college freshman at a party, hoping to meet someone, or find friends, or impress the world with mean dance moves.

Sure, some of us set income or travel goals, assuming that achieving them would make us feel a certain way.

But I wanted to cut to the chase: track how I feel first, and work to improve that directly.

The Story of Joy and Anxiety

Every morning since August of last year I asked myself: how much joy did I feel yesterday? how much anxiety? and then scored both on a 5 point scale.

I chicken-scratched it daily in my journal, before moving it all into a spreadsheet to do the analysis. This is the first chart I plotted:

It shows how my levels of joy and anxiety fluctuated over time, on a 7-day rolling basis. And it matches up really well against real life. 

Let’s take the month of September:

I spent two weeks vacationing with my parents, the first time I got to see them in over a year. In the first couple of days, we see the stressful run up to flying in the era of COVID: certified PCR tests, documents, questions, lines of people. “Security”.

Then my joy skyrockets and anxiety plummets as soon as I get to see my parents and steadily climbs till it peaks towards the middle of my time with them. It then slowly tapers off through the last few days, as the daily routine starts to normalize (as an aside, this is why I think the dream of a permanent beach vacation is unlikely to solve our everyday woes).

Are Consistent Levels of Joy and Anxiety A Good Thing?

As expected, there’s a strongly inverse relationship between anxiety and joy: -0.82, where -1 is perfect inverse correlation. This means that, on balance, days with higher joy came with lower anxiety, and vice versa. But I was surprised to see that often I felt roughly similar levels of joy and anxiety on the same day. In fact, almost 44% of days had scores within 1 point of each other.

If almost half of our days are both anxious and joyful in roughly equal amounts, then maybe the data of our days is less important than which part we choose to highlight when we tell ourselves the stories of lives.

It’s also remarkable how narrow the band of my joy and anxiety tend to be. More than half the time, my anxiety is between 2.1 and 2.8 and my joy is between 3.4 and 4.0. Those are pretty tight ranges. 

It’s even narrower if we look at uneventful periods, where I don’t do much except work. Most of the time, my anxiety vacillates only between 2.5 and 2.9, and my joy even less, 50% of the time captured between 3.3 and 3.6 (this is the IQR for the geeks reading this).

That could seem like a good thing — like I’m emotionally stable or something, but it seems life is fuller when the band of emotions is more elastic. That’s the opposite end of the catatonic monotony of existence that many of us fear falling into.

Or maybe a balance between consistent, quiet periods and some emotional rollercoasters is what, paradoxically, keeps us sane.

Data Confirms: Emotions Are Impermanent

Let’s look at the data again, this time shifting to December.

After a long time of intense, high-pressure, heads down work, things got harder. I found myself more anxious than usual, more easily irritable and less excited about the work I was doing. For the first time in the four month period I tracked, I felt very little joy and lots of anxiety consistently, rather than here and there.

For the time in my professional career, it felt like I was burning out.

Wait, Burning Out? But You Seem Pretty Happy

I showed a draft of this article to a close friend who pointed out that this chart actually looks like I was doing quite alright, even through the worst of it. That my joy never dipped below 2.5.

That was a real eye opener for me because I knew how tough many of the days in November and December were, and it didn’t square with the story the data was telling.

First I blamed the data: it’s a 7-day rolling chart, meaning it averages the day-to-day fluctuations into single points, hiding some of the really tough days. But the truth is, that wouldn’t explain the broader trends.

As I digested it more, two things struck me:

First, a day with a low joy score feels like failure to me.

Like I was writing a day off. And, maybe it’s the Damn Optimist in me, but when I was reflecting, I could always find a small corner of positivity, no matter how fleeting. The taste of warm soup or a text from an old friend amidst a typhoon of a day. Having this perspective increased the scores I gave my days, thus not one day had a joy score of 1.

This approach is healthy to the extent it forces me to scour for optimism. But it’s not so healthy when I refuse to acknowledge that a day just really, really sucked. The way to feel more joy is not to fool oneself into thinking that things are not allowed to suck. That’s how I create situations where my level of happiness seems high externally but internally I am a card-carrying rider of the struggle bus. And realize it only after I miss my stop.

Which brings me to my second point…

I have unrealistic expectations of how my days should look like.

As I excavated deeper, this archeologist stumbled on an ancient rule carved into the stone of my consciousness. In this exercise, as in my life, the unspoken rule has been that nothing but the best is acceptable.

That the goal of life (and my tracking) is to have two straight horizontal lines: unchanging 5s in my level of joy, unchanging 1s in my level of anxiety. But does that ever happen? And if it did, would it be truly desirable? Even if it were, I’ve only had 2 days that got close to that since August. Does that mean that the other 135 days were somehow not good enough?

As a takeaway, I would like to learn to accept that some days are tough without interpreting that as a failure. I would also like to recognize that the kinds of days I’ve been striving for come by rarely, and that there’s a lot of life worth living crammed in the large spaces between them.

Anyways, I Was Burning Out.

So I took an extra day off to spend a magical long weekend upstate, disconnecting from my devices — and from the world. When I got back, I took myself on a date: first to the MoMA, where an iPhone app whispered sweet nothings about Donald Judd into my ears, and then to a bougie chocolate store that fed me samples till I felt too guilty to walk out empty handed.

It helped. A lot. I lunged into the week with morning yoga poses, herbal teas and newfound enthusiasm, represented by that peak in December.

And yet, my work challenges were still right there, where I had left them. I began to revert back to the mean of my anxiety and joy, reflected in the steady decline to a local minimum. Alas, the holidays, it turned out, were around the corner, giving me much needed time to rest and reflect — hitting a second peak right before the new year.

That’s a big range of things to feel in a few weeks time. But the 7-day rolling average chart (above), hides a lot of day to day fluctuations.

What if we looked at the data on a straight-up daily basis?

That’s what that looks like.

Both anxiety and joy move around a LOT day to day, and it doesn’t seem like there's a pattern to whether feeling anxiety one day makes us more likely to feel it the next day. 

We can confirm this by looking at a concept called autocorrelation, which indicates how correlated one day’s data is to previous data, and lets us (sort of) capture the concept of “momentum”. For the anxiety data, the autocorrelation is quite low: 0.2, where 1 is perfect autocorrelation.

It’s a good reminder. Every day that’s sad or anxiety-dominated feels like that’s just how things are gonna be for good. But I know, now with the data to prove it, that I am very likely to step out of bed feeling quite different the next morning.

Weekdays vs. Weekends

Most of my life I’ve been living with a subconscious dichotomy: weekdays are for things that are challenging or tedious, like work, and weekends are for relaxation, lethargy, fun. This is a dangerous mental model of time that I’ve been intentionally moving away from. 

Its implication is that we spend 5/7th of our life in the section of time we’ve labeled as “work”, also defined by what it’s not, namely “not fun”. That way of thinking can make life feel miserable.

So it was a pleasant surprise to see that my levels of anxiety and joy actually don’t vary that much between weekdays and weekends:

I read that as a good thing. It means I am living my life during the week, and not simply waiting for the weekend to feel joy and be able to relax. It also makes sense: I try to find balance by doing some “work” or writing every weekend, and finding time for myself during weekdays.

That’s All To Say

There’s a lot of work ahead: both learning more about my life and actually applying what I learn. But that’s the journey, I guess. And I look forward to it.


LinkedIn, And Other Indicators of Anxiety

Aside from Joy and Anxiety, I’ve been tracking a few behaviors that I think correlate with how I feel on a daily basis. Things like spending time on LinkedIn, saying Yes instead of No, checking my phone in the mornings and a couple others.

That’ll have to be a separate article. But if you liked this one, make sure you subscribe so you can get the next one delivered to your inbox, hot off the electronic mailbox printing press.

Gil