Image of a San Francisco cab courtesy of Flickr user bluewaikiki, under a Creative Commons license.

Uber and Lyft have become so popular so quickly in large part because they've filled a big gap in the urban transportation market. The smartphone-enabled ride services have offered more predictability than cabs, faster travel than transit, and fewer hassles — no parking, for one — than personal cars.

For a glimpse of exactly how this is playing out in a city that perennially struggles with each of these problems, researchers at the University of California at Berkeley recently conducted a survey of Uber, Lyft and Sidecar users in San Francisco that offers some of the first academic research on their behavior and preferences.

The study, conducted by Lisa Rayle, Susan Shaheen, Nelson Chan, Danielle Dai and Robert Cervero, surveyed 380 people intercepted in the evening in three popular San Francisco neighborhoods, where they had either just exited a car with one of these apps or used them recently. The group isn't representative of all "ridesourcing" users in the Bay Area, as the researchers call them. But other data makes it possible to directly compare their experiences to taxi and transit use.

In one of their more striking findings, Rayle and co-authors found that 66 percent of the trips taken by people who use these app services would have been twice as long if taken by public transit instead (that's if nearby transit was at least available). That number includes all of the time spent just waiting for the trip to begin. This graph, which draws on estimated travel times by car and transit using Google Directions, succinctly captures the comparative advantage of Uber and Lyft. Many trips that would have taken 20 minutes by "ridesourcing" would have required 60 minutes or more by public transit:

"App-Based, On-Demand Ride Services: Comparing Taxi and Ridesourcing Trips and User Characteristics in San Francisco," L. Rayle et al.
"App-Based, On-Demand Ride Services: Comparing Taxi and Ridesourcing Trips and User Characteristics in San Francisco," L. Rayle et al.

An Uber car doesn't move any faster through traffic than a taxi does. But the study also found that passengers were likely to wait significantly less for a "ridesourcing" car than a taxi. Using data from a taxi survey conducted by the San Francisco Municipal Transportation Agency, the researchers compared self-reported wait times for "ridesourcing" passengers and taxi riders, depending on the time of day and location in town.

During weekdays, 93 percent of "ridesourcing" passengers said they waited less than 10 minutes for a ride. For passengers ordering a Taxi, the same was true for only 35 percent of them:

This was the breakdown at night during the week:

And on the weekend:

When the researchers further divided this data by taxi zone across the city, they found that taxi wait times also differed significantly depending on location. Alternatives like Uber and Lyft, in other words, provided better service and service that was more consistent across time and location. Users prominently cited these advantages, along with the convenience of electronic payments, when asked why they used these apps:

"App-Based, On-Demand Ride Services: Comparing Taxi and Ridesourcing Trips and User Characteristics in San Francisco," L. Rayle et al.

It's clear in all of this data that companies like Uber are filling an unmet need for better transportation. By doing so, they're both complementing and competing with existing options. Many of these trips could not have been made easily by public transit, suggesting that these services supplement the transit network to some extent. Thirty-nine percent of people surveyed said they would have taken a taxi instead if these apps weren't available. Another 8 percent, though, said they would not have made their trip at all if Uber, Lyft or Sidecar weren't around. That means these companies are modestly expanding the market for rides, even as they steal business from taxis.

Ultimately, these companies may well force taxis to improve their service in response to these trends, a possibility that could ultimately quash demand for these alternatives. In this rapidly shifting marketplace, the Berkeley researchers are quick to note that much has already changed since they gathered this data in the spring of 2014. San Francisco cabbies are now more widely using a comparable app called Flywheel.

This research also raises a key policy question that the authors acknowledge they can't answer: Why do passengers report better experiences with taxi alternatives than with taxis themselves, among other things? "Is this due to characteristics of the services themselves," Rayle asked at a recent transportation conference where the research was presented, "or is it due to the fact that they’re exempt from taxi regulations?"