It's a Date

Hey, I’m Jim! I’m not the regular author of this blog, nor do I know as much about movies or metrics as Nick. However, he was still kind enough to lend me this space for a minute to share one of my musings.

Tracking The Office

It always happens to me halfway through binge-watching a show. I see Rick Grimes six-gun another ~~zombie~~ walker in The Walking Dead and I think “I should have been keeping track of his kills this whole time”. Or keeping track of the rules to True American in New Girl. Etc. But by the time I think to do this it’s too late, and I don’t watch shows over again – at least not paying the same amount of attention.

But if there’s one show that’s good enough to warrant multiple rewatches, it’s The Office. I thought about what I could measure – there’s no killing (at least [1] nothing [2] confirmed [3]), and most of the silly games only last one episode. Despite how amazing The Office is, the plot itself is pretty boring. So, I decided to make an equally boring choice and keep track of which dates and days of the week episodes take place. It isn’t a zombie body count or a game to play with friends, but it’s still a) The Office, and b) data, so it’s fun (for me).

Jokes about boring material aside, The Office is actually a good show to do this with, for a few reasons:

1. Episodes mostly occur on weekdays, so the pool of potential dates is cut down by nearly 30%.
2. There are a lot of holiday-themed episodes.
3. The show is semi-realistic in that people age, have kids, etc. in a reasonable amount of time in the show compared to how much time is passing in the real world.

Sometimes this was easy. For example, in season 3 Andy starts a 10-week anger management class. He comes back exactly 10 weeks later in real life. Perfect timing. But at other times The Office was mind-numbingly bad with its timeline. The very next episode Andy claims that he was only gone for 5 weeks (probably because he was gone for 5 episodes).

It’s also inconsistent with more general facts (like whether Pam or Jim started working at the office first), and sometimes you have unreliable narrators. In one of the most inconsistent recurring cases, Michael’s birthday is either in February or March, or sometime between May and August, depending on which episode or character you listen to.

In fact, Jim and Pam even acknowledge the inconsistency of the writers in a meta sort of way in season 5:

Jim: “Andy put down a bunch of deposits on stuff for his wedding with Angela, but then she was sleeping with Dwight for... several years. Wait, no, that can't be right.”

Pam: “The timeline's messy."

In short, I know this is a TV show and I’m overthinking it. I probably put more thought into the timeline than the writers did. I also know that the arbitrariness of some of my decisions makes parts of this analysis almost worthless. But if you’ve made it this far you might as well keep reading.

All of the charts below were made in Power BI. You can find the PNG files for those as well as a CSV of the data in my Git repo.

What day of the week do episodes take place?

To start, let’s look at my best guess for every episode:

Sunday and Saturday being low makes sense (the show takes place in an office, if you didn’t get that), but why is Thursday so high? Most episodes aired on a Thursday, and in the absence of any evidence to the contrary, I assumed the episode took place the day it aired.

Since I can’t really make a good guess for those episodes, let’s just take them out. Here’s what the distribution looks like if we focus on episodes where I’m semi-confident about what day of the week it took place:

First thing’s first – I’m not confident in that many of the episodes. Because of the air date bias, Thursday loses nearly all the days I had given it before. We still end up with another interesting distribution though. Most episodes here take place on a Friday or Monday. The reason is similar for both – characters like to talk about their weekends (before it starts or after it ends). In the most extreme case, 4 of the 6 episodes in season 1 definitively take place on a Friday (and a 5th possibly does, though there’s no evidence for or against it).

Friday is clearly the king though. While a decent number of distinct episodes do take place on a Monday, its overall numbers are bolstered by the fact that 2 episodes (the weight-loss competition) stretch across 8 Mondays.

Days Per Episode

Like I just mentioned, some episodes take place across multiple days. How common is that?

About 2/3 of episodes take place on one calendar day. At most episodes take place across 4 days (not counting the gap between days where nothing happens), but really anything more than 2 is uncommon.

There is once again some bias here with the initial chart, though it’s more on the writers this time. For whatever reason, a lot of the cold opens clearly take place on a different day from the rest of the episode. If we don’t like counting those because they don’t usually affect the plot or take up much time, we get a distribution that’s pushed slightly more to the left:

Excluding cold opens mostly just moves half of the 2-day episodes down into the 1-day category. That’s not a very exciting change, but at least now we have 2 colors in our chart.

It’s the Most Wonderful Time of the Year

If you’ve watched even a handful of episodes of The Office, you’ve probably seen them celebrate a holiday. If you’ve seen the whole series it won’t be a surprise which holidays are celebrated the most often:

Christmas, Valentine’s Day, and Halloween clearly take the cake. Despite not being shown in the first season, Christmas is still celebrated 9 times because it was celebrated twice in season 7 (Michael delayed it when he found out Holly was coming back).

Anecdotally, I’d expect a typical office to celebrate Thanksgiving more, but maybe my coworkers are particularly big fans of potlucks.

Diwali and St. Patrick’s Day make the list with one occurrence, as does Secretary’s Day. That last one I was unsure whether to include. Is Secretary’s Day a holiday like Christmas or a “holiday” like National Ice Cream Day (which I didn’t count)? I decided to include it, though I think Stanley would be heartbroken that I gave more attention to that than to Pretzel Day.

Fans of the show will probably notice that I didn’t include birthdays. That was basically just an oversight on my part while taking notes.

What Else You Got?

I mentioned at the beginning that I tried to determine what dates episodes occurred on – and yes, in the end I really did assign a specific date to each episode. Again, many of the selections are somewhat arbitrary, but at an aggregated level you still see some interesting trends.

There’s no mathematical reason why this ends up looking almost like a normal curve. 2005 starts off low because the first season only included 6 episodes. The count ends even lower in 2014 because the show didn’t actually air that year – it was a flash forward during the finale.

Aside from the tails, the rest of the curve is partly a coincidence. 2009 does have the most episodes that aired in a single year, but years in the middle also seemed to have more multi-day episodes.

Breaking things down by month, you can see a big drop-off in episodes during the summer (when The Office never aired). That isn’t a surprise, but it does make me wonder what other kind of plots and shenanigans we would have seen if they hadn’t just skipped the summer months most years.

September, December, and January are also lower due to fewer air dates. Similarly, October ends up being the highest just because it has more air dates (since it’s a 31-day month and generally didn’t take weeks off).

In the end, I could have saved a lot of time and just plotted out which dates episodes aired instead of when they took place. That would have ended up being a very similar chart.

The End?

This analysis only covered events that happened entirely within the timeline of the show, but there were enough clues dropped to determine long-term events – when people were born, how long they worked at Dunder Mifflin, etc. It isn’t a lot of extra info, but it is a lot of extra work given how it must be pieced together one little bit at a time.

What are some things that could have made this analysis better?

1. More time. The show featured 201 episodes and 9 webisodes. That’s a lot of content to pore through, and I only watched each episode once for this exercise. I know I didn’t catch everything the first time, and I’m sure my methodology in choosing dates changed slightly across the months it took me to do this.
2. A better fashion sense. Multi-day episodes would have been easier to notice if I had been paying attention to the clothes that characters were wearing (assuming a change of clothes generally means it’s a new day). But I didn’t realize this opportunity at first, and I didn't want to make the effort to pay that much attention later.
3. I didn’t cover deleted scenes, bonus footage, etc. I could make an argument like “deleted scenes don’t count because they were deleted”, but the real reason is that I watched this on Netflix and didn’t see any deleted or footage (aside from the webisodes, which I did seek out on YouTube).

Feel free to keep those factors in mind if you also decide to spend months of time on this. We’ll compare notes when you’re done lol.