Methodology
Data Gathering
Our data gathering practices have evolved over time. Adequate preparation before the first candy seeker arrives is essential to a successful evening.
Identifying Your Team
The size and skill of your team of data analysts will shape what kind of data you collect and how many data points you can expect to track. If you want good data, your analysts need to be focused exclusively on logging incoming trick-or-treaters. We suggest anyone who enjoys socializing or wants to be able to come and go throughout the evening hand out candy rather than tracking data.
We also suggest having each analyst track a single category, especially if you have large waves of candy seekers, and having each category assigned to a single analyst. Any time you have an analyst switch what they're tracking or switch the analyst tracking a category, inconsistencies are introduced into the data. Part of the fun of the either/or categories is debating how to classify different edge cases. For example, one analyst might assert that a human-sized taco should be categorized as a fantasy because they have dreamt of giant tacos, while another might place it under science fiction due to the genetic engineering required to create a walking, talking taco. You want the classification to remain consistent.
Selecting Categories
Back in 2003, the denizens of the Alpine Butterfly Lodge began tracking their trick-or-treaters. One of our data analysts who knew these folks liked the idea and we joined in beginning in 2015. Initially, our categories matched those ABL was tracking, though over the years we've branched out to select our own categories.
You can track anything that interests you. We encourage you to take a look at our publications to learn about what we track. If you are in a high traffic area, we strongly advise utilizing either/or categories.
Procuring Candy
We see anywhere from 1000-1200 trick-or-treaters for a 4.50-hour evening and spend $200-300 on candy each year. If you'd like to look at numbers over time, ensure you have enough candy to support a full evening of data collection. Our team would like to stress that there is absolutely nothing fun about "fun sized" candies and we recommend using the "mini" form factor as your baseline.
Running out of candy can force an early stop to data collection, but developing an understanding of the traffic patterns unique to your location is part of the learning curve. If you run out of candy during your first years of data collection, do not be hard on yourself.
Beginning the Evening
First blood in most neighborhoods occurs between 5 and 5:30pm on weeknights, trending 30-60 minutes earlier on weekends. We suggest having your team in place by that time. Emphasize the comfort of the data analyst by providing a place to sit with a clear view of incoming trick-or-treaters, beverages, and liberal personal access to the candy bowl.
Who To Count
Our data collection is about trick-or-treaters, so we count anyone who receives candy in our data. We do not care if the trick-or-treater is costumed or not; if a data point asks for candy verbally or with an outstretched hand or container, they receive it, and we include whatever they are wearing in our data, even if it's their street clothing. We strongly encourage other data collection teams to offer candy to beleaguered parents, tired caregivers, and surly adolescents regardless of costume status, and to include those individuals in their data.
How to Count
While at our previous location, we were able to tally trick-or-treaters using hash marks on sheets of paper, and this worked even when tracking complex categories because we had under 300 guests over the course of the evening.
At our current site, we see almost 1,000 trick-or-treaters over the course of the evening, with over 200 an hour at our peak. Our high level of traffic means we have had to prioritize efficient counting methods, and in 2023, we introduced the handheld clicker to our technological arsenal.
The clickers are hot glued to scrap wood so they can be easily held in the analyst's lap or on an adjacent table. Small pieces of glass tile are labeled with the categories each year, and can be erased and updated depending on the points being tracked.
Creating Easier-to-Audit Data
Each half hour, our analysts call out their totals to be logged on a summary sheet. Noting data frequently allows us to compare data from different analysts and find issues during the analysis phase, and isolates problems inherent to the peak traffic moments without muddying the data from the slower moments.
For example, in 2023, four of our five analysts noted 6 trick-or-treaters between 5pm and 5:30pm. This helped us determine that the analyst who noted 7 had inadvertently counted a caregiver escort as a trick-or-treater, and we were able to excise that particular data point from analysis.
Ending the Evening
When to close data selection is ultimately up to you. You can end based on duration (3/4/5 hours from first blood), time of day (8pm, 9pm), or activity (when trick-or-treaters drop to <15/10/5 per half hour).
Trick-or-treating in our neighborhood runs late, with candy seekers moving through the neighborhood well past 9pm, so we usually end based on a combination of duration, ambient temperature, and analyst energy.
In 2023, we stopped collecting data after precisely 4 hours, though we continued casually distributing candy for another 20 minutes or so while striking our seating.
Handling Discrepancies
No matter what you do, if you are located somewhere with large waves of candy seekers, you will have discrepancies in your data. How you resolve those is ultimately up to your team. We use our half hourly data to determine the most likely total number of trick-or-treaters, and extrapolate from the overall data to determine which points to add or remove.
For example, modern costumes typically outnumber retro ones at a ratio of 2:1, so if one analyst logged 3 trick-or-treaters that others did not, we would subtract 2 modern and 1 retro costume from their count. We understand this is ultimately an imperfect method of massaging the data, and we suspect is a principal reason we have not been able to secure grant funding for our research.
Analyzing the Data
Coming soon!
Finding Your Own Best Practices
Coming soon!