Halloween Data

Given that the Alameda Unified School District made Halloween a student holiday this year, we prepared for first blood earlier. Our hypothesis was confirmed by our first spooky family landing at 4:53 p.m. Over 4.5 hours, our team of five data scientists and five candy wranglers distributed 82 pounds of candy and about 120 stickers to 1,164 trick-or-treaters.

Our data analysts tracked a number of categories per usual. We did not introduce any new categories this year, and retired the height category due to logistical challenges. While previous years have seen a smooth ramp up and steady decline, this year we saw a flatter, steadier peak period and noticeable dip between 8 and 8:30 p.m.

Inflatable dinosaurs and alien abductions remain common amongst the inflatables. There were a number of new inflatables this year, including Mr. Potato Head and Rex from Toy Story, Hyacinth Hippo and Ben Ali Gator from Fantasia’s “Dance of the Hours,” Squidward, an airplane, and an axolotl one of our researchers initially thought was a Mudkip. That said, inflatable costumes were down by a third from last year, and many of the inflatables we did see were already deflated even early in the evening. These costumes may be on the decline due to their mobility-inhibiting nature, deterioration due to costume age, or the increase of their use by adults.

Baskets have begun to rise in popularity as candy receptacles, and the analyst who has been tracking this for several years has suggested spinning those out from the “Other” category to replace hats, which have been steadily declining in popularity. Trader Joe’s seasonal canvas totes were well-represented among bags, along with the perennial pumpkin containers, Boo Buckets, and plastic shopping bags.

For once, our data scientists were on top of a cultural trend and went into the evening fully expecting to see a massive number of costumes from the Netflix property KPop Demon Hunters. We were not disappointed. From HUNTR/X to Saja Boys, Derpy to Sussie, children and caregivers pushed the prevalence of the nascent intellectual property beyond either Marvel or DC. Likely a direct result of the rise of the Honmoon, Elsa’s stranglehold on Halloween seems to have come to a close. We only saw a handful of Elsas in her Frozen 2 togs. White costumes also increased in popularity due to the rise of KPop Demon Hunter candy seekers in the trio’s “Golden” livery, and the prevalence of black costumes dipped slightly.

Color data otherwise remained stable from years past, with black continuing to shroud the lion’s share of trick-or-treaters, including a particularly delightful bat made from a cardboard box. Once again, virtually all of the yellow costumes we saw this year were Minions, though we did enjoy the presence of several Grus. Modern costumes (post-1900s) are becoming more common, beating out retro ones (pre-1900s) at about a 3 to1 ratio rather than the 2 to 1 ratio seen in previous years.

Last year’s Joker prevalence inspired us to track DC properties this year, but we did not see a single Joker. Supermen dominated the DC landscape (sparse despite the James Gunn 2025 film release), and Spidermen (and the rare Spider-Gwen) continue to comprise most of the Marvel costumes.

Our team now feels confident saying that Chucky and Ghostface have been fully integrated into the classic horror costume oeuvre. Based on the difficulty of classifying these costumes, along with those like Freddy Krueger and Michael Myers, into the science fiction/fantasy dichotomy, we will be adding a third “horror” category next year at the suggestion of the analyst who has been tracking this category on an ongoing basis. Most of these classic horror costumes, including more generic options like creepy clowns and ghouls, are currently inflating the fantasy category numbers to 85%, and we believe creating more granular data in this domain will more accurately represent our community of candy seekers.

In terms of trends, over the course of the evening, we saw what amounted to a full herd of shot-in-the-head deer costumes. When we mentioned it to one posthumous deer, curious about whether this was a meme or intellectual property we did not know about, they were surprised to learn there were others prancing the streets. Our data scientists keep an eye out for the influence of TikTok on costume theming and point to that influence for the sudden rise of Max Goof costumes this season, a trend our team found, well, goofy.

We also saw an increase in Cheetah Girls, an astoundingly accurate Vyvyan from The Young Ones, and a single Prince from Katamari Damacy, likely following a similar trend wherein we saw Left Shark costumes rise as parents shared their beloved media moments with their offspring. Of special note in this category was a bedraggled Cornelius, ready to give you a hug at the local support group so you could cry it out and get a good night’s sleep. What made this costume particularly horrifying was the wearer referring to 1999’s Fight Club as an “ancient text.” Didn’t that just come out?

We continue to post our findings first in the Alameda Post!

2024

First blood was at 5:02pm this year, kicking off 4.5 hours of distributing candy to 1,166 trick-or-treaters. It was a briskly cold evening, and at various points the accuracy of our data collection was compromised by the presence of jackets or shivering children and their escorts running full bore to and from the candy bowl.

Our team of six data analysts made our triumphant return to tracking candy containers for the first time since 2019. Pumpkins, be they hard-sided buckets or soft bags, remain a perennial favorite among children. The canvas tote has edged past the pillowcase within the bag category, but plastic shopping bags continue to enjoy popularity due to their accessibility.

We welcomed a new experimental category this year, using the CDC’s height chart to determine that 3.5’ is the median height for 5-year-old boys, as well as the height of the ruler we were able to jam into the dirt by the front door. Using height as a proxy for age (as it’s gauche to ask a guest how old they are), we estimate 17% of this year’s trick-or-treaters were under five.

Virtually all of the yellow costumes we saw this year were Minions. Color data otherwise remained stable from years past, with black continuing to shroud the lion’s share of trick-or-treaters. Modern costumes (post-1900s) continued to beat out retro ones (pre-1900s) at about a 2:1 ratio. Our researchers are aware that our decision to categorize Star Wars as retro due to its “long, long time ago” setting remains one of our most controversial practices.

Technological advances in costuming have led to a noticeable increase in inflatable costumes. Past years have exclusively featured the tyrannosaurus rex (with and without rider) and alien abductions, but this year we saw nearly twenty different puffy designs, ranging from a capybara to a unicorn skeleton to Teletubby Dipsy.

Science fiction’s share over fantasy continues to dwindle, as does the presence of Disney properties including Pixar, Marvel, and Star Wars, previously the top priority for youth costumes. While Elsa remains well-represented, most Marvel costumes were from the Spiderverse, with only a handful of Captains America and a single Killmonger to share the load. Marvel Phase 5 and 6 fatigue has now clearly set in and DC is rising in the ranks. We sincerely regret our oversight in not predicting this given the release of Joker: Folie à Deux, and plan on tracking the DC properties as well as the Disney empire in future years.

Prompted by a large number of Chuckys over the course of the evening, our group extensively discussed costumes centered around intellectual property that the costume-wearer is unlikely to have seen. We recognize Freddy Krueger and Michael Myers are considered to be "classic" scary costumes separate from their hard R movies, but were surprised to see that Ghostface and Chucky seem to have also entered that hallowed hall.

We recognize that an apology is a down payment on making things right, and last year we apologized to parents and caregivers for singing "Baby Shark" in response to shark costumes, only to engage in the same behavior again this year. At this juncture, we anticipate continuing to do this on an ongoing basis, so going forward we will no longer be apologizing, doo-doo, doo-doo, doo-doo. In fact, we went so far as to customize the lyrics as "cozy shark" to acknowledge how warm these onesie costumes looked. We also sang the Spongebob Squarepants and Teenage Mutant Ninja Turtles themes when candy seekers in those costumes stopped by, so the problem will only get worse.

Lastly, as data analysts, we are advised to avoid building relationships with the subjects of our research. However, sometimes a data point is such an outlier in our eyes we cannot help but praise their ingenuity within the data set. While the family of four cranking Double Platinum while striding through the neighborhood in full KISS makeup and attire made us suspect they would rock and roll (our rankings) all night, the clear analyst favorite this year came in the form of a homemade costume: a vending machine, complete with interior illumination of its tasty snacks. It is possible we should leave the lab more often.

We continue to post our findings first in the Alameda Post!

2023

After a whirlwind first year at a busy Alameda intersection in 2022, Halloween 2023 saw us well-prepared for an evening of data collection and candy distribution. We had a large team of seven data analysts in lab coats parked out front of the house, and started the evening with enough candy to handle whatever the evening threw at us (plus extra from our generous neighbors!), which ended up being a whopping 929 trick-or-treaters.

We upgraded our data tracking system from hash marks on clipboards to attendance clickers, pausing every 30 minutes to note the current standings. This was significantly easier to manage while logging the largest waves of trick-or-treaters, and while we have always room to refine our methodology in the service of accuracy, we're improving every year.

Our first blood was a wee black cat at 5:07pm, even earlier than last year! We collected data until 9:07pm, at which point we shifted to handing out candy to night owls (and repeat customers who weren't as sneaky as they thought they were) and getting ourselves inside to unwind and process the data before bed. We once again smashed our previous all-night records, and were it not for a commitment to a reasonable bedtime and warm environs, MacKenzie suspects we would have broken 1,000 candy seekers.

Color data remained relatively stable from years past, with black continuing to dominate the community couture. Modern costumes (post-1900s) continued to beat out retro ones (pre-1900s, which includes Star Wars due to its "long, long time ago" setting) at about a 2:1 ratio.

This year, we had the pleasure of partnering with the Alameda Post to introduce a new category. Their editorial team was curious about whether we'd see a rise in political costumes given that we're about to enter into an election year. Based on discussions with them and our team of data scientists, we determined that political costumes include any and all agents of the State, ranging from royalty (princesses) to first responders (law enforcement, Paw Patrol, SWAT, fire fighters) to armed forces (Army, Navy). Much to our surprise, we saw zero political figures this year, whereas we had seen at least one RBG and 45 in previous years.

We saw a number of Nightmare Before Christmas characters (primarily Sallys), half a year's worth of Wednesdays, and a couple Coralines. It remains clear that Alameda is where goths go to retire and have kids.

Despite the movie coming out in 2020, we saw a noticeable rise in Sonic the Hedgehogs, as well as the anticipated bump in Teenage Mutant Ninja Turtles characters with their 2023 flick. The biggest shift we saw was a massive drop in science fiction costumes such as Iron Man; the youth of Alameda appear to be unimpressed with Marvel's Phase 5, and the rise of Spiderverse costumes was not enough to compensate, allowing fantasy costumes to dominate.

In defiance of the associated mobility challenges, inflatable costumes remain popular. In addition to a dozen dinosaurs, we saw a large number of the inflatable costumes showing a kid being held by an extraterrestrial. One of our data analysts commented to one such trick-or-treater "There's a lot of kids being abducted tonight!" and, let me tell you, the side eye and shade from an adjacent group of tweens at that particular comment was the kind of delightful shaming that this age cohort makes famous.

Our team can only assume that the Super Bowl babies of 2015 are old enough to pound the ground for candy because we saw a stunning four Left Sharks (note that it is possible we only saw three, as two of the sharks came as a pair, and one may have in fact been Right Shark). Baby Shark had a strong showing, and our data analysts would like to issue a formal apology for responding to each Baby Shark by singing "Baby Shark." We pride ourselves on creating a safe space for trick-or-treaters and their families, and now understand our response resulted in parents and caregivers hearing their children sing "Baby Shark" for extended periods of time. This isn't aligned with who we want to be as scientists, doo-doo, doo-doo, doo-doo.

Happy Halloween!

We now post our findings first in the Alameda Post!

2022

Halloween 2022 marked our triumphant return to data collection after several pandemic-induced years off. Our location changed and, let me tell you, we were NOT prepared for the wildness that is Grand Street on Halloween. This year boasted our largest team of data scientists yet: four clipboard-ready adults and a candy-wrangler. We had first blood at 5:21pm, earlier than we ever saw at our previous location. If this is what Monday looks like in our new neighborhood, we live in fear of the next time Halloween falls on a Saturday.

We smashed our previous all-night records within hours and had to rely on an emergency infusion of candy from our generous neighbors to keep going (thank you!), still withdrawing into the house at 8:41pm while families roamed the neighborhood.

Color data remained relatively stable from years past, with black continuing to dominate the community couture. Most trick-or-treaters were unmasked, but we still saw some COVID protections, some of which were well-integrated into the overall lewk. This year saw a distinct rise in candy bags with a skeletal hand attached at the top, though traditional bags and pumpkin buckets continued to dominate.

We last tracked retro vs. modern costumes in 2018 and percentages there remain stable, though we did see a drop in fantasy costumes as compared to science fiction ones since last looking at those numbers in 2016. Anecdotally, MacKenzie believes this is due to a rise in Spidermen and Spidergwens. One of our data scientists, Laura, owns an inflatable t-rex costume, so we also chose to track dinosaurs roaming the street. There were several. Stay safe out there, that's all we're saying.

Happy Halloween!

Following the posting of this report in the Alameda Peeps Facebook group, the Alameda Post reached out to us to publish our findings!

2019

Halloween 2019 delivered our best showing since we started collecting data in 2015, with first blood at 5:57pm and lights off at 9:05pm.

Our 191 trick-or-treaters trended later this year and, anecdotally speaking, older. We still have not bested our 2015 apex, but MacKenzie has high hopes for 2020 when Halloween will again fall on a Saturday.

Working off a hypothesis from one of our data scientists, we tracked costumes that were part of the Disney empire: Disney, Star Wars, Pixar, and Marvel. We were shocked to discover that Disney's dominance was far less pervasive than we assumed. We also tracked costume color and candy receptacles, which we've done several times before.

Happy Halloween, everyone!

2018

MacKenzie was teaching on Halloween this year, so many thanks to the team for collecting data in her absence.

There were 170 logged trick-or-treaters and 178 logged candy receptacles. Our data collectors indicated wild rushes of children that interfered with fully accurate data collection. Hopefully, next year's return to 3+ data collectors will support even better data collection.

Like in 2017, we looked at how the trick-or-treater collected their candy, as well as whether the costume could best be classified as "retro" (operational definition of pre-1900) or "modern" (operational definition of post-1900).

2017

Data from Halloween 2017 are in! We had a disappointingly low number of trick-or-treaters this year, fewer than we had when it was pouring rain in 2016. The usual haunted house on our block wasn't around this year, which is likely partially responsible.

Drac valiantly tracked a number of data points despite MacKenzie's constant micromanagement. We looked at how the trick-or-treater collected their candy, as well as whether the costume could best be classified as "retro" (operational definition of pre-1900) or "modern" (operational definition of post-1900).

In a move that got MacKenzie some side eye, she opted to classify Star Wars costumes as "retro" because it took place "a long, long time ago".

2016

The weather gets in the way! Last year, Jon and MacKenzie hosted 261 trick-or-treaters. This year, it was a paltry 167 due to pouring rain. Here's a breakdown of whether the costumes were best classified as science fiction or fantasy.

The hardest costumes to classify were children dressed as food. MacKenzie thinks a hot dog or cookie the size of a five-year-old is a fantasy, whereas Jon rightly pointed out that it could well be the result of futuristic genetic modification, so we placed those five kids with science fiction.

2015

We disbursed from 5:30pm until 9:45pm, and didn't track the total amount of candy distributed. We had a total of 261 visitors, with black as the dominant color by far. Our per-color breakdown is at left.

We left our home entirely undecorated, relying on the porch light to bring folks our way.

Google Sites

Report abuse