Inauguration Attendees Make Significantly Less Money than Women’s March Attendees
Comparing the crowds with movement data reveals interesting differences.
Ryan Fox Squire and Victoria Gu, SafeGraph
Washington D.C. was the site of two major political events last weekend: the Presidential Inauguration and the Women’s March on Washington on January 20th and 21st, respectively.
Using anonymized GPS data from mobile devices (smart phones), we analyzed these events and their attendees.
The SafeGraph dataset included hundreds of thousands of anonymous devices in Washington D.C. throughout the month of January, and we identified many thousands of devices attending each event.
Interestingly, 2% of attendees at the Women’s March also attended the Inauguration.
Watch how crowds spill into the National Mall
Since both of these events dominated downtown D.C. on back-to-back days, it was easy to identify and compare anonymous mobile devices in attendance.
We can visualize the movement of all people during the day of the event as a movie — which is fascinating to watch. Crowds gather, disperse, march and parade. In the videos below, each dot shows an anonymized GPS signal from a mobile device. The slider at the bottom indicates the time of day.
Time lapse of movement on the day of the Inauguration
Time lapse of movement on the day of the Women’s March
Tool credit: Time lapse visualizations were made using Carto.
Inauguration Attendees Make Significantly Less Money
We can estimate the income levels for attendees of the two events by merging SafeGraph movement data with data from the 2015 American Community Survey (conducted annually by the US Census Bureau). Census data is organized by zip code, and SafeGraph data identifies home zip code based on where the device spends the majority of its time. We also used census data to correct for geographic biases in our datasets.
We expected relatively wealthier people would attend these events , the assumption being that traveling and taking time off work are more accessible in more affluent populations. The data confirm this. The average attendee at both the inauguration and the women’s march is estimated to have a household income over $70,000, which is significantly higher than the national median household income — estimated to be around $58,000.
What we did not expect was that there would be significant differences between the groups.
The average Inauguration attendee came from a zip code with $71,000 median household income whereas for the average March attendee the number is $77,000.
Given our large sample size, this $6,000 difference is statistically significant.
Large geographic differences in the two crowds
Shown below are the top ten home states of the attendees of the two events (again, after correcting for geographic biases).
Not surprisingly, the participants for both events came from a combination of states that have large populations and are geographically close to Washington D.C..
We can take these percentages of attendees for each state and calculate the difference between the two events to visualize which states contributed relatively more or less to the two events.
States that are dark blue (e.g., New York) contributed a greater percentage of the total attendees at the Women’s March compared to the Inauguration. For most states, like Illinois, the differences are subtle (< 1%), meaning, for example, that the proportion of attendees from Illinois was about the same at each event. Nonetheless, there is a clear geographic bias.
The Northeast region — stretching as far down as Maryland and Virginia — made up a significantly larger fraction of the Women’s March crowd than the Inauguration crowd.
This visualization is similar to an electoral map — with some notable exceptions. In particular, 6.4% of the Inauguration crowd was from California compared to only 4.5% of the Women’s March.
Locals flee D.C.
Analyzing the total number of individuals in Washington D.C. during the month of January revealed some interesting trends approaching the Inauguration.
During the week leading up to the Inauguration there is a simultaneous influx of non-residents to D.C. and exodus of locals (people who live in the DMV — the District, Maryland, Virginia metro area) out of the city. These decreases in locals are probably individuals who normally commute into D.C. opting to avoid the city in the days leading up to the Inauguration.
Everybody likes Starbucks and Shake Shack
We also explored which restaurants attendees visited during the Inauguration weekend when they were not at their respective events. The patterns were remarkably similar.
The figure shows a few popular and unpopular restaurant categories along with the percentage of attendees that visited that category at least once during the weekend.
The most popular categories in both groups were coffee (e.g., Starbucks), Bagels/Donuts (e.g., Einstein’s Bros Bagels and Dunkin’ Donuts), sandwiches (e.g., Potbelly), and hamburgers (e.g., Shake Shack). There were no major differences in the dining behavior of the two groups of attendees during this weekend.
At least when it comes to sandwiches, both groups agree.
If you found this interesting or valuable — please recommend and share this post.
And you can also follow the SafeGraph data blog.
— —
This analysis was conducted by Ryan Fox Squire and Victoria Gu using SafeGraph data. Ryan is Product Manager at SafeGraph and former Data Scientist at Lumos Labs. Victoria is a Software Engineer at SafeGraph and a former Software Engineer at Yelp.
SafeGraph is building the world’s largest and most accurate ground truth dataset on human movement to power machine learning and human analysts. Co-founded by Auren Hoffman and Brent Perez, SafeGraph is located in San Francisco.
