Methodology

WMATA kindly provided eight months (May – August, 2015 and 2016) of card-level Metrorail data with obfuscated serial numbers from unregistered SmarTrip cards. (129.26 million observations).

Exhibit 1: Classifying “Home” & “Work” Stations

I examined the data per person on a weekly basis; weekends were dropped.

“Home Stations”

1. When an individual’s first trip of the day originated at the same station where their last trip of the day terminated, the originating/terminating station is a potential Home Station. Days include trips ending after midnight.

2. For persons with qualifying Home Stations, their most frequented Home Station in a given week is their Home Station for the week. If a person doesn’t have a predominant Home Station in a week, their data for that week is excluded.

“Work Stations”

3. For days with exactly two trips, in weeks where riders have a Home Station, Work Stations are the Metro stations where a person’s first trip of the day terminated and their last trip of the day originated.

4. The most frequented Work Station in a week is a person’s Work Station for the week. If a person doesn’t have a predominant Work Station in a week, their data for that week is excluded.

“Home Stations” & “Work Stations”

5. There is a single Home Station and Work Station combination per person and week. If a person had only one qualifying Home Station and Work Station week through the entire data, their observations are excluded. (This is to avoid the inclusion of tourists.)

Summary Stats

6. Home Station and Work Station trips are credited once per rider per week, regardless of how many trips per week that rider made between the stations. This counts riders commuting between Home/Work Stations without overweighting riders who traveled the route more times per week than others.

7. Outlying trip times between Home/Work Stations are excluded based on the interquartile range. Specifically, trips with times exceeding seven times the difference between the upper and lower quartiles of the distribution are excluded.

8. Displayed travel times are an average of the actual commuting time between Home Stations and Work Stations. Individuals’ morning commutes to work are averaged with their evening commutes home, then averaged again against all individuals with the same Home/Work station pair.

Exhibit 2: Time / Distance Ratio and Trips per Week

This analysis relies on the population of Riders with Home/Work Station pairs and with average trip times per previously described methodology.

“T/D Ratio” (Time/Distance Ratio)

1. The T/D Ratio is the average Metrorail travel time between Home Stations and Work Stations divided by the geodesic (“crow flies”) distance between the stations. Geodesic distance is calculated using the X Y coordinate radians of each Metrorail station. Ratios are rounded to the nearest whole number.

Trips per Week

2. Trips per Week are the number of trips riders take per week between Home Stations and Work Stations, averaged across all Home/Work Station pairs with the same rounded T/D Ratio.