# Summary

Seems like every few months some random academic tweets about how they work a lot of hours, and then other people tweet about how they are doing fine despite working fewer hours. I’ve never participated in these conversations because I didn’t know how much I worked, so I tracked my work at 30-minute intervals for 2019 and 2020 to try and answer two questions:

1. How much do I work?
• Quick answer: ~40-45 hrs/week, and never more than 50 hours unless I’m away from home for a conference, workshop, or field work. But, I don’t take very many days completely off.
1. When I do work, how do I spend my time?
• Quick answer: research has the plurality of my time over the full 2 years, but has been overtaken by logistics (emails, paperwork, etc.) recently.

The years analyzed (January 2019 - December 2020) had several important personal and professional changes including (i) going from a postdoc to a PI at a new institution (August 2019); (ii) transitioning to working from home due to COVID (March 2020); (iii) and having our second kid (June 2020), so I also decided to look into whether my working time or activities changed through time.

I did this analysis mostly for curiosity’s sake, but then realized I should make a couple data-driven resolutions for how to improve my work-life balance in 2021:

• Take at least one day per weekend with 0 work hours (preferably both) - I was often spending an hour or two on emails during the weekend.
• Reduce the amount of time spent on logistics. I think a recent shift to not checking my email or taking any meetings before 10 AM has been helpful in this regard, but other tips would be welcome.

Update 2022/01/02: See here for the updated stats with another year of data.

# How much do I work?

Answering the first question (How much do I work?) is pretty simple; I can just add up all the hours I worked.

Over the course of the two years, I worked 3908 hours, or an average of 37.9 hours per week - pretty close to the standard 40 hr work week. Sounds pretty good! Until, of course, you consider the fact that I took some vacations during this time, and also was on part-time leave for 3 months with the new baby.

The hour count is pretty consistent week-to-week, fluctuating around 40-45 hrs:

In a typical week I work between 40-45 hours. My median weekly work time is 42 hours and the mode is 44 hours. The three vertical red lines correspond to starting my new job (August 2019), switching to work from home (March 2020), and the new baby (June 2020). Only the new baby seemed affect hours worked, as I went on partial leave for 12 weeks.

There are a few outliers on the high end: 9 weeks in the past two years where I worked more than 50 hours. Cross-referencing to my calendar reveals that all of those are weeks where I was away from home for a conference, workshop, or field work for at least part of the week. I also noticed that there are no weeks with 0 hours worked, and the least I worked in a week was 2 hours, which is troubling.

Now, let’s take a look at the daily data, colored by the type of day:

Pretty clear pattern - usually working right around 8 +/- 1 hours on work days, with the extreme highs occuring during conferences/workshops/field work, and lower values on the weekends. This is also evident in a histogram of daily hours worked:

As we can see, a peak right at 8 hours (vertical black line) and a pretty strong fall-off on either side of that. But, how about that left side of the histogram - less than 75 days with 0 hours worked over the past two years? Yikes. In this time period alone, there were 205 weekend days and another 81 days off for holiday, vacation, parental leave, etc., meaning that I actually worked on 70% of my days off.

This matches well with my priors - in the histogram we can see a large cluster of days < 2 hours, which is about the length of my kids’ nap time, and I will often sit on the couch and check some emails during nap time on weekends. This pattern is even evident if we look at the hours worked broken down by day of week:

Those dark red weeks correspond to the times when I was away for a conference/workshop (note how they stopped in early 2020). For some reason Mondays and Thursdays are the days when I tended to put in a bit of extra time. I’m proud of the fact that I never put in a full work day on a weekend (>= 6.5 hrs), but less happy about the fact that there are a lot of light yellows on Saturday and Sunday, which leads to not a lot of gray boxes.

I found that the maximum consecutive days where I worked (61 days, from March 31 to June 3, 2019) was 12x greater than the maximum consecutive days I didn’t work at all (5 days, from August 24 to August 31, 2020). Since it’s healthy to take days off, this is a problem that needs solving. The low-hanging fruit is to just eliminate all the quick bits of work I did during weekend nap times, as mentioned above, under the assumption that those tasks (typically responding to emails) can be folded into the regular work week, which would probably be more efficient anyways. To ballpark the potential impact of this, I calculated that if I converted all days where I worked <2 hours to 0s, I would have a maximum of 13 consecutive days worked and 16 consecutive days off - much more to my liking!

So, my resolution from part 1 of the analysis: Take at least one day per weekend with 0 work hours. The potential detriment of putting in those occasional weekend hours (overwork) is likely more than any gains (incremental progress on unimportant things). I don’t want to cut it to 0 hours on both days, because I like to spend a half hour on Sundays planning out my work for the week which helps me start the week focused on Monday morning. I suppose this could be shifted to my last activity of the day on Friday afternoon… maybe in 2022!

# What do I do when I’m working?

Answering this question is a bit more complicated, but I split things up into 6 categories:

• talking = things that involved other people… meetings, seminars, zoom calls, etc.
• logistics = stuff that is generally unimportant (but sometimes necessary). emails, expense reports, etc.
• research = what pays the bills - writing, making presentations, data analysis, reading papers, etc.
• conference = anything at a conference/workshop. I define it broadly so it includes going out for meals with colleagues, etc. - basically, any time I am in the presence of other people.
• service = journal or grant reviews, committees, open science stuff, etc.
• mentees = talking to or working with someone who I am mentoring such as a postdoc or student.

Let’s look big-picture first. My 3908 hours of work broke down as follows:

Research comes in the lead and constitutes 32% of my time - thank goodness, since that’s nominally what my specialized training is in. It’s followed unfortunately closely by logistics (28%) and talking (23%). Before you go thinking I’m an inattentive supervisor, I didn't start keeping track of mentees as a separate category until April 2020, so that one is definitely an underestimate. From April 2020 onwards, it averaged 13% of my time.

Unsurprisingly, there’s a lot of variability from week to week:

The black lines once again denote the key dates of starting new job, working from home, and having new kid. Maybe it’s just the way I’ve visualized this, but it seems like the logistics category has the most similar thickness through time, while research is a lot more variable. This is especially evident whenever there’s a conference (which also includes workshops) - they jut down like little knives cutting away my research time, but leaving logistics more or less intact. This contributes to a concerningly long period of time in the second half of 2019 where my research time was rarely about 25% of total hours worked, which I was fortunately able to remedy - with some help from the lockdown - in 2020.

Focusing in on these top 3 categories, which represent 83% of my time, we can see changes in the relative time investment:

The proportion of my time I’ve spent on logistics has gone up, which is driving down the amount of time spent on research and talking. Based on this overly simple linear trend: logistics has risen from ~25% of my time two years ago to more like ~33% of my time now, and research has been knocked out of the top spot, from ~35% of my time two years ago to more like 25% of my time now.

So, my resolution from part 2: Reduce the amount of time spent on logistics. I’m afraid more logistics may be in my future because of spinning up a big upcoming project that just started, but I’ve tried to recently work just on research for the first 2.5 hrs of my day before checking emails or having any meetings, which has been promising. But, any other tips for cutting down on this stuff would be useful.

# Conclusions

Speaking of using my time effectively, my main conclusion is that I’ve spent far too long analyzing my work habits! But other than that, I'd say my main take-homes are that I work a pretty reasonable amount, but need to do a better job stepping away on weekends/holidays/vacations.

# Some methodological caveats

This was a self-reported study done purely for curiosity’s sake, so I’m sure many holes can be picked in the analysis.

• I just kept track by and estimating what I did for the majority of the half hour period beginning at that time. So, if I worked for 16 minutes on research, I would log it as research. In contrast, if I spent 16 of the 30 minutes doomscrolling on Twitter, I logged it as not work. This is all approximate and was often done after the fact.
• There’s some overlap between these categories (for example, talking with a mentee about their paper would be categorized as mentee though it could also be considered talking or research), so I just tried to be internally consistent.
• However, I wasn’t totally consistent because I introduced mentees as a new category in April 2020, so it is definitely an underestimate.