Abstract
With data from service delivery requests in twenty-nine cities, this article reports on a comparative analysis of city service request systems (311 systems) and their operations. The study uses actual 311 service request data available through cities’ open data portals. A typology of thirty service type categories was created. Various hypotheses were tested. As city population increases, the number of service requests also increases. Garbage/recycling is the most commonly requested service, followed by code enforcement requests, parking, pickup of bulk items, and abandoned vehicles. Cities are no longer just using telephones for their users to submit requests but have also now incorporated other service channels. Service resolution times vary across cities and service category but there is some evidence that safety and health types of services receive priority and are resolved more quickly. The article ends with managerial and policy implications.
Public administration doesn't get any more fundamental than the delivery of services to users upon their request. Users contacting 311 systems to report that a service is needed among the most basic coproduction of government services is crowdsourcing community needs. Following one typology of coproduction (Nabitchi, Sancino, and Sicilia 2017), 311 service request systems represent the co-delivery of services at an individual level.
With data from service delivery requests in twenty-nine cities, this article reports a comparative analysis of city service request systems (311 systems). The 311 service request centers have become an essential part of many cities’ service tool portfolios, growing in number and importance across the country. A Virginia Beach study investigated consolidation of their multiple call centers; they determined that 84 percent of their departments had specific employees answering user requests (and in 23.7 percent of the departments, those employees spent up to 91 or more percent of their time on those tasks). Further, 72 percent received twenty-six or more requests for service during just one week. By consolidating requests for services into one 311 call center, cities eliminate the need for users to struggle to answer questions like, which government do I ask to do this, what department do I go to, and what is their phone number? (Virginia Beach Department of Management Services 2011).
Once based entirely upon telephone requests, these systems have expanded to include multiple means of collecting requests, organizing and identifying them in “call” centers, and forwarding them to the appropriate city agency for action. This referral system has become a welcome shift for cities and users, as they replace users trying, on their own, to navigate sometimes unclear websites or phone directories to try to figure out where to direct their problems. Cities’ open data portals now allow us the opportunity to directly study these systems with the data generated by the 311 systems themselves.
The article will provide a literature review on the research to date using 311 systems, identify a typology of the most commonly requested 311 services, and test hypotheses about these systems across major American cities. The methods and data, 311 call/click requests obtained from open data portals of various American cities with 311 systems, will be described, and conclusions are drawn.
Government Service Delivery and Service Requests
Public sector services are delivered to users via a limited number of channels or modes: face-to-face, telephone, postal mail service, e-delivery through the Internet (web or email) and, through m-government or mobile devices. Services can also be characterized by their complexity level: uncomplicated, limited complexity, and highly complicated.
Service Request Systems
The idea behind 311 systems was to apply technology to users’ need to request services to increase responsiveness and accountability and, ultimately, trust in government. These systems create more transparency in service delivery, provide more customer-oriented services, and positively focus on citizen engagement. The first 311 systems were developed during the Clinton Administration as community policing strategies to relieve pressure in cities where users called 911 lines with nonemergency and emergency requests. In 1996, the Federal Communication Commission (FCC) set aside the 311 number for nonemergency uses. (U.S. Department of Justice Office of Justice Programs 2005).
After much planning, Baltimore developed the first 311 system and processed the first 311 service request on February 13, 2001 (City of Baltimore 2019). The opening of the nonemergency service request system had an immediate impact upon their 911 system: the number of calls to the 911 number declined, the amount of time it took operators to answer calls went down, and the number of users hanging out without speaking to an operator declined (Schwester, Carrizales, and Holzer 2009). In Baltimore, the system was also integrated into their CitiStats system and, ultimately, strengthened that, too. Other cities soon followed, and by 2009, at least thirty-two cities had the system (Schwester, Carrizales, and Holzer 2009). By the time of this study in 2017, at least eighty-four cities had adopted 311 systems.
Today's 311 systems are sophisticated systems with multiple input channels (including mobile options and incorporating Web 2.0 interactivity) for receiving resident requests, back-end CRMs (Citizen Relationship Management) databases, integrated delivery systems transferring requests to all departments of a government, and feedback systems allowing citizens to follow their requests to resolution. “There's been a shift away from 311 being a glorified switchboard that handles basic questions to where it can handle multiple channels, moving requests, forward automatically, even closing out the process and informing the customer that the job is done. …311 is now about work order management, field services and customer satisfaction.” (Newcombe 2017).
Further, “the idea is that people should not need to know what agency is responsible for fixing problems.” (Minkoff 2016, 216). Today, channels used in U.S. cities’ 311 systems include everything from the traditional telephone, mobile apps, Web interface, Twitter, Facebook, face-to-face walk-ins, fax, and mail, and even a mobile van circulating the city.
Literature Review
There has been little comparative research examining and comparing the actual structure of the 311 systems across cities. Instead, the 311 literature mainly has concerned why 311 systems were adopted or used the 311 data to explore hypotheses on equity, impact on city budgets, and increased resident trust. The literature surrounding 311 service delivery can be categorized in several ways, first by the research methodology utilized (disagreeing with the literature categorization laid out by Chatfield and Reddick (2018) or according to several important themes. The methodological categories first include the very numerous case studies of individual cities with 311 systems, although some of these studies have different purposes and goals to direct their studies: Boston (O’Brien et al. 2017); Edmonton (Lu and Johnson 2016); Houston (Chatfield and Reddick 2018); Lynwood (Barnhouse 2008); Miami-Dade County (Schellong and Langenberg 2007); Minneapolis (Fleming 2008); New York City (Minkoff 2016); San Antonio (Fleming 2008); Philadelphia (Nam and Pardo 2014); San Antonio (Fleming and Barnhouse 2006); and San Francisco (Clark and Guzman 2017)].
Several scholars have also used survey research to explore 311 (Schwester, Carrizales, and Holzer (2009), Clark and Rokakis (2014), and Reddick (2011, 2009)). For instance, Reddick (2011, 2009) surveyed 311 system programs in 2009 to understand how and why cities adopted these systems.
Another sizeable category of literature about 311 systems were non-comparative technical papers [examples are Fleming 2008; Schultz 2003]. These technical papers provide essential details about the systems’ actual operations so that scholars can understand issues important in implementation and service delivery. Of these, most were case studies but, in evaluating the City's six major call centers to determine options for their consolidation, in 2011, Virginia Beach, Virginia's Department of Management Services conducted a small comparative study of 311 call center characteristics. This study and the case studies comprise the only research conducted on the operations of 311 systems, and no effort was made to determine if other cities shared operations.
A more helpful way of categorizing the literature about 311 systems is to explore common substantive themes. In this case, topics in the literature included why 311 systems were adopted and then managed to thrive, 311 programs as coproduction service delivery systems, and various outcomes (like increased resident trust, subsequent equity for users, why users participated in 311 operations, and impact on budgets) (Chatfield and Reddick 2018; Clark and Guzman 2017; O’Brien et al. (2017); Lu and Johnson 2016; O’Brien (2015); and Fleming 2008).
Adoption and Successful Implementation
Reddick (2011, 2009) investigated factors common to the adoption of 311 systems, finding the form of government, the managerial capacity of the leadership, the support of upper management, and the ability to collaborate across the city were all critical factors. Additional survey research identified essential characteristics of 311 systems from a survey of systems in fourteen cities in 2009 (Schwester, Carrizales, and Holzer 2009). We note many of these factors are significant in predicting other technological adoptions by cities. Like other technological solutions, Nam and Pardo (2012) found that strong support from organizational leadership, flexible and adaptable staffing, and partnerships with external organizations were factors that helped the program survive early challenges.
Outcomes
Through investigating a series of case studies, Fleming (2008) was able to show that 311 systems improve citizen satisfaction and, ultimately, improve trust in government. Investigating a different question about the impact of 311 systems on citizen trust, Schwester, Carrizales, and Holzer (2009) surveyed thirty-two 311 systems in 2009. They concluded that the systems impact citizens’ trust in government, and ultimately, citizens believe the government is more responsive and effective.
Equity in any service delivery system is an important outcome; there has been increasing work on this question involving 311 systems (Clark, Brudney, and Jang 2013; Lu and Johnson 2016; Clark and Guzman 2017; Kontokosta, Hong, and Korsberg 2017; Chatfield and Reddick 2018; Clark and Brudney 2018; Clark et al. 2020). Researchers examined possible budgetary outcomes in studies comparing Boston and San Francisco with multiple data types but could not find any effects (Clark and Guzman, 2017). However, Clark and Rokakis (2014) found that users’ use of the 311 system at any level led to significantly higher and positive views of the quality of city services. Kontokosta et al. (2017) found in New York City that neighborhoods with more minority and fewer English speakers, higher unemployment, and male residents tended to underuse the 311 system. But overall, the more impoverished the neighborhood, the less likely users were to use the system, a finding that is certainly in line with that earlier research in citizen contacting behavior (generally a middle class-based behavior).
Clark et al. (2013) found an unequal spatial distribution of urban service delivery and citizen contacting behavior. Analyzing Boston's 311 service requests, they found little difference in how the contact systems may benefit one group over another, but some differences in the method of contacting, as Hispanics used 311 less as the systems moved more and more to the Internet and smartphones. Clark and Brudney (2018) found the differences in demand for services were not related to socioeconomic status. In an expanded study of fifteen cities, Clark et al. (2020) found slightly quicker service response rates by city departments in minority neighborhoods except for San Francisco. They found departments responded faster in higher income neighborhoods. Otherwise, they concluded that they found no systematic biases, whether the requests originated in poorer, less educated, or more minority neighborhoods.
Reasons for Requesting 311 Service Requests
There has also been research conducted on why users contact 311 systems. Using 311 request data, a user survey, and voter registration records to examine motives for service requests, O’Brien et al. (2017) and O’Brien (2015) found higher levels of concern for the area around them lent motive or custodianship. Because users perceived “ownership” of the city, they were more likely to report the need for services in that area. Wu (2020) utilized a survey of San Francisco citizens to find that the level at which citizens accept technology and use city services plus their level of satisfaction significantly influenced their use of the 311 system.
An essential pathway of research covers the spatial distribution of 311 contacting. Minkoff (2016) theorizes three types of service requests—government goods, noise, and graffiti. He also suggests that “311 contacting volume is driven by both the condition a space is in and the attributes of the people who inhabit that space” (Minkoff 2016, 236). Minkoff finds that socioeconomic status and the city's resources are weakly connected to the number of requests to the 311 system in New York City. He also saw more requests in areas with older housing stock, high foot and vehicle traffic (an alternative explanation is that these areas are part of the central city), and population growth.
System Operations
Little research throughout the surveyed literature compared the essential characteristics of 311 systems across the country or on cities’ responsiveness to 311 service requests. What results were found tended to be a by-product of other findings. As stated earlier, the Virginia Beach study provides some results on early call center operations (2011). In doing so, they compared data from sixteen cities across the country (including Virginia Beach). The results, from the point of view of the call centers themselves, included the average annual number of calls per 100 residents, the average yearly call handling time (123 seconds across cities), the average wait time to speak to a live agent (32 seconds), the average percent of calls abandoned before the agent answered (11.5 percent), and the average percent of calls transferred to another department (18.6 percent). (Virginia Beach, Virginia Department of Management Services 2011)
The Virginia Beach survey of all their call centers indicated 63.8 percent of the requests received were for information about City meetings and events, 35.9 percent were requests for City services, and 48.7 percent were complaints about services. Requests for hours of City services were 43.6 percent, while information about billing and taxes was almost a third of contacts (30.8 percent). We also note that these results are from all of the City's call centers—not just the 311 call center, so we cannot generalize them to the 311 call center itself. (Virginia Beach, Virginia Department of Management Services 2011)
In their study, White and Trump (2018) mainly focused upon whether 311 data was helpful in studying citizen participation; they also reported on the service request category, providing the top twenty-five types of services from the cities they studied. As these were from New York City, requests for housing authority services were also included and dominated the top part of the list. The top five were Heating, General Construction, Street Light Condition, Street Condition, and Plumbing; three of these are related to individual housing issues and not more public concerns. Clark et al. (2020) utilized seven service request types to study equitable coproduction across fifteen cities. They indicated that trash/recycling/litter was the most common request type across all fifteen cities, followed by requests related to animals, trees, potholes and roadways, and streetlights. We can conclude there is some variation across cities in the types of services. Still, there is evidence that streetlights, street condition, garbage and recycling, animals, and trees are among the most significant concerns.
Clark et al. (2020) studied whether governments respond differently to 311 service requests based on the location of the service request; in other words, are there inequitable outcomes for requests coming from minority or poor neighborhoods. They included fifteen cities in their analysis and concluded that mostly, the outcomes are not inequitable.
Hypotheses
We will move on to developing hypotheses from the literature reviewed here. From the literature review, we realize that we do not know much about the operations of 311 systems, including some of the most basic descriptive information. We have no information about the kinds of services users most request or even the volume of service requests received by cities. More importantly, we do not know what patterns in service requests might exist across cities; do users in different cities request the same types of services, or does that differ across cities? We see that technology is moving on, and more and more cities are using technologies besides the telephone and call centers to receive requests, but we do not know which channels are being used by which cities. These obvious omissions lead us to specific research questions and hypotheses based upon what we know about technology and its usage in government and cities.
Hypothesis 1
We expect that the number of service requests will be some linear function of the size of a city's population, based upon research on technology in government and the limited results of the Virginia Beach study. That study suggests that cities with more resources will use those resources to advance their technological services (Virginia Beach Department of Management Services 2011). Our hypothesis is thus:
Hypothesis 1—Number of Requests
Hypothesis 2
White and Trump (2018)'s research suggested streetlights and street conditions were among the top services requested. While focusing on equity and coproduction issues, Clark et al. (2020)'s research also provides results on the types of services most commonly requested; garbage and recycling were the most frequently requested, followed by animals, trees, potholes, roadways, and streetlights. Anecdotal observation of city services also suggests urban services like garbage and street repair and improvement would be necessary to users across cities.
However, beyond that, we would expect that there would be differences across cities in the pattern of services requested, based on the specific needs of their location, weather, and surrounding environment. For instance, some cities would experience high levels of requests for snow removal, but other cities would have no need at all for that service. Other cities have other quirks, like being closer to an urban wildlands interface, meaning animals are more likely to travel in the city. Thus, more animal-related services might be needed. Our working hypothesis will be:
Hypothesis 2—Service Types Requested
Hypothesis 3
While the 311 service request system began as a phone number (i.e., 3-1-1- rather than 9-1-1-), many cities now use mobile apps, web based-apps, or even Twitter or Facebook as a channel for their users to make requests. We hypothesize that using the telephone to make service requests will be retained by all cities, but additional service channels will also be utilized.
Hypothesis 3—Service Channels
Hypothesis 4
We also note that service resolution time is likely a function of both the time needed to complete the actual process of resolving the service request and the priority given to the request. The responsiveness of various departments to 311 service requests is just beginning to emerge as an important research topic (see Clark et al. 2020). But their research looked at geographic differences between service request response times and differences across service times; they expected service times to differ across different service categories. We expect services involving repairs (like streetlights or street repairs) to take longer to complete than drop-off services (like dropping off garbage cans). We also expect that cities would give service types involving safety or health a higher priority. These would include services like dead animals, sewage, hazardous materials, and traffic signals.
Hypothesis 4—Patterns of service request resolution will differ across service types, but services involving safety or health will have shorter service resolution times when measured across all cities in the study.
Methodology
The unit of analysis for this study is the individual service request in each city. Each city with a 311 system containing the actual 311 data in its open data portal was included in the study. This project is a comparative analysis of the request data from users in each of those cities’ 311 service request systems.
The Data and Data Collection
The data utilized in this study were the self-reported service request data streamed through cities’ open data portals; the unit of analysis was the individual service request. White and Trump (2018) examined the “promises and pitfalls” of 311 data; they cautioned against using 311 data to study participation levels but believed they would be meaningful to measure service demands; this is how this study uses the data. To begin this study, the cities with working 311 systems were identified; to obtain the service request data itself, the websites of those cities were then examined to determine if they had open data portals. If they did have open data portals, those portals were searched for 311 service request data. Only those cities with both 311 systems and open data portals containing 311 data were used in this study.
A total of twenty-nine cities were identified that provided service request data in 2017, the year studied. From each of those twenty-nine cities’ open data portals, the datasets for 311 service requests were downloaded. Chicago broke down their data into eleven different datasets separated according to significant service categories. In each city's open data portal, the 311 datasets were identified and filtered to include dates within 2017. Together, these datasets resulted in a total of forty-one different datasets (containing 7,418,374 service request cases), all with varying numbers of variables and cases, that were analyzed for this study in what could best be described as an analysis of multiple analyses. Table 1 contains a list of the cities whose 311 data was found using this methodology and utilized in this study.
U.S. Cities with 311 Service Request Cities and Their 311 Data Available Through Their Open Data Portal.
Analysis
After the data was collected, each dataset was separately analyzed using Excel and the data visualization software Tableau as data management and analysis tools; Stata and SPSS were utilized for statistical analysis. Tableau is quite powerful at allowing flexible and easy recategorization, so it was beneficial with this sometimes unruly data. In particular, Tableau made it possible to combine the sometimes tens of different types of service requests.
The analysis itself consisted of exploratory data analysis using the capacities of Tableau and Excel for each variable. Then descriptive statistical analysis was conducted across all cities and variables using Excel and SPSS. Finally, Tableau was utilized to develop the graphics to illustrate the patterns found across cities.
Variables
The study's variables included service request type, methods used by the city to receive requests (service channel), date request was received, date request was resolved, and geographic location of the request. The department to which the request was referred was often included in the dataset, but not for every city. The date the request was received and the date it was closed were used to calculate service time, the length of time (in days) cities took to resolve the issue. To reiterate, in all datasets, the unit of analysis case represented one individual service request.
The most important variable analyzed was service type, and significant effort was put into developing a schema of urban service types used in this analysis. Of all the variables , service type was the most difficult to code because it differed so much from city to city. In all cities, some kind of classification system of the various types of services requested was used. Each request was coded according to that city's design, there being no standard classification system across cities. A classification system might have only ten to twenty codes in a strict categorization system in some cities. Then, there might be no real effort to standardize in other cities, resulting from tens of possible codes. The fact that different cities utilized their 311 systems for various purposes also made this process difficult. For instance, New York City collected requests for the housing authorities through their 311 system (these were eliminated in this analysis). Other cities collected inquiries for drivers’ licenses and vehicle registration. The analysis was made even more chaotic because some cities receive requests via email (without a form to structure the request), some receive requests by Twitter, and others take walk-in requests. When a city does not have a structured intake form, the service request takes whatever freeform the requestor or city staff member likes. For example, in one city with many requests for attention to abandoned vehicles, they included each vehicle's actual license tag; this results in hundreds of category entries in their database, not only one standardized entry (such as Abandoned Vehicles).
A qualitative, iterative process of content analysis and code development was utilized to develop the coding schema for this study. Service categories were created by going through the data and examining the types of service requests across all cities, taking notes on the types of services listed throughout that process. Then the service categories were grouped together and further adjusted, and their number was reduced. Eventually, a total of thirty service categories were developed. Table 2 provides the final list of city service categories, as developed through the coding process, in alphabetical order. Potholes were included as a separate category even though they are also types of street repair because potholes are an iconic type of urban service, and so many cities utilized them as a separate category.
Categories of Service Requests Developed for Study.
Results
Hypothesis 1: Number of Requests
We begin with analyzing the number of service requests across all our cities to test our first hypothesis. The report of results begins with basic descriptive information about 311 service requests and delivery systems. The number of service requests varies enormously by city and city size, of course (Table 3). New York sees the most requests by far (2,466,757 in 2017), followed by Los Angeles (1,135,608) then Baltimore (645,769). Memphis (1,312), Evanston (16,214), and Tacoma (24,037) had the fewest requests in this study. Statistical analyses determine that the 2016 population of each city had a 0.94 correlation with the number of service calls. Therefore, Hypothesis 1, suggesting a connection between population and number of calls, is supported.
Numbers of Service Requests and Other Request Information by City.
However, once the number of requests was standardized by city population to create a measure of service requests per capita, a different picture emerged (Table 3). Somerville, MA, had the highest number of service requests per capita (1.29), followed by Baltimore, the first city to develop a 311 system, with 1.04 per capita. Memphis had both the smallest number overall and the lowest per capita number of service requests. The linear correlation coefficient between population and requests per capita is statistically insignificant, at 0.087.
As Figure 1 indicates, the pattern when viewing requests per capita is instead a nonlinear one, taking on more of the shape of a negative exponential curve. In this pattern, the populations of the largest cities (New York City, Los Angeles, and Chicago) exceed the “average” service requests expected of cities because of their large population. They are in the far left corner of the graph and curve. Smaller cities with well-developed 311 systems (such as Baltimore, Somerville, Sacramento, and San Francisco) exceed the number of service requests expected of cities of their size; they are around at the top of the graph with much higher requests per capita. The remainder of the cities cluster in the middle on both population and requests per capita axes.
Population by requests per capita.
Hypothesis 2: Service Types Requested
The same table (3) also includes data on the number of service categories in the original 311 files from each city; in other words, these are the number of unique types of service requests in the original data file found in the open data portal. These categories indicate how structured the 311 service request system is in each city; the more categories, the less structure imposed by the city. Some cities impose a rigid structure on resident requests, asking that they choose from relatively few categories. Others impose little at all. The most significant number of types for a city is San Francisco's at 420, and the smallest is 12; the average number was 138.0. (Note: Chicago's data is not organized like other cities; instead of one dataset, there are twelve datasets, one for each category they collect.)
After the author organized the service requests into the 30 service categories determined in this study, we can see how effectively these new categories cover the types of services in these cities (in the fourth column of Table 3). The higher percentage indicated in Austin showed 115 service categories recoded into the 30 service types identified by this study. Once the data was recoded by the author, 97.7 percent of all the service requests were covered by the 30 service categories, indicating the 30 service types developed in this study were a good fit for this city. At the low end of the spectrum, Philadelphia had fifty-two categories, but once the service request data was recoded into the 30 services, only 24.8 percent of those requests were covered by our thirty service categories. Overall, 84.0 percent of all service requests (7,418,374) could be categorized into thirty service request categories. This result provides some suggestion that the 30 categories had some explanatory power in identifying types of services.
The following results indicate which services are requested the most by users. Table 4 presents the top thirty services list, ranked by the percentage of requests made for those services across all cities. Of all thirty categories of service requests and across all cities (Table 4), garbage/recycling was the most often requested service (representing 18.5 percent of all services requested), three times as many as the next most commonly requested service (code enforcement). Garbage/recycling includes requests for new trash or recycling carts or bins (or for repairs), pickups or stray carts, requests for different carts or containers, and complaints about missed collections. Rounding out the top five most commonly requested services after garbage/recycling and code enforcement were parking, requests for pickup of bulk items, and abandoned vehicles service requests. These thirty categories of service requests represent 84.0 percent of all the service requests across these cities.
Ranking of 30 Most Commonly Requested Services Through 311.
*Superscript = Ranking from most variation to least.
These results support our hypothesis that garbage collection, street repair and improvement, and streetlights would be the most commonly requested services. In the aggregate across all the cities, garbage/recycling is the most frequently requested service, consisting of 18.5 percent of all service requests. However, streetlights ranked 8th, and street and sidewalk repair ranked 9th, with only 3.6 and 3.5 percent of service requests, respectively.
To estimate the amount of variation across cities in Table 4, we turned to a standardized measure of dispersion showing the amount of variability relative to the size of the mean, the coefficient of variation (calculated as the ratio of the standard deviation to the mean) (also shown in Table 5 in the third column from the left). The smaller the coefficient, the less variation, and the more stable and consistent service across cities. Garbage/recycling, street/sidewalk repairs, signs, trees, and parking showed minor variability of their average scores across cities, suggesting more consistency of those types of requests across cities. At the other end of the scale were taxes/payments, noise, transit, homelessness, and hazardous materials; these service request areas had the highest coefficients of variation, indicating more variability across cities. Some cities did not include these service areas at all. The low coefficients of variation for garbage/recycling and street/sidewalk repair support Hypothesis 2, but the coefficient for streetlights is larger, suggesting more variation across cities for that service area.
Numbers of Service Requests and Other Request Information by City.
Pulling information about cities and services together, Table 5 presents the services requested most often for each city, along with the percent of total requests for that service in that city. For instance, service requests around live animals were the most commonly requested type of service in Austin, comprising 22.8 percent of their service requests. These included requests to get animals’ proper care, animals locked up alone in vehicles, requests for animal control, and notification that animals were being illegally sold by the roadside or were stuck in storm drains.
Of the twenty-nine cities in this study, fourteen had garbage/recycling as their most commonly requested service type. Bulk items, code enforcement, and parking services were the next most frequently requested service in two cities. The second most widely requested services were spread out more across the various service types, but five cities had code enforcement as their second most commonly requested item, and three had garbage/recycling.
More informative is the high range of total service requests covered by just the top two service areas for each city (last column in Table 5). Ranging from 8.9 percent for Philadelphia to 81.4 percent in Los Angeles, the top 2 services requested by users accounted for an average of 42.2 percent of all services requested. Across cities, this indicates some stability in service requests by users.
The high number of cities with garbage/recycling as their most commonly requested service partially supports Hypothesis 2. However, street and sidewalk repairs were only first or second in one city—Memphis and streetlights were first only in Chicago and second only in Riverside.
Hypothesis 3: Type of Service Channels
Table 6 presents information on the types of service channels users used to make their service requests, where this data was available. As can be seen, although mobile apps have taken hold and are being increasingly used, in most cities, the telephone is still the method most used by users to make their service request. However, only four cities (Birmingham, Nashville, Montgomery, and Austin) still indicate service request levels by telephone of 80 percent or more. There is a great deal of variability across cities, and Hypothesis 3 is supported. Telephones are used across the cities, and there is also a wide range of other channels being used.
Service Request Channels Utilized by Reporting Cities.
Hypothesis 4: Service Request Resolution Times
Table 7 presents descriptive statistics for the amount of time cities took to complete the different service requests. These are ranked by the mean number of days needed to complete a service request, from the least to the most (an asterisk indicates when those descriptive statistics are based on the service times for fewer than ten cities since not all cities contain each type of service). The service type cities resolve the fastest are hazardous materials, dead animals, bulk items, taxes/payments, sewage, noise, live animals, water, transit, and garbage and recycling. These results suggest some support for the hypothesis that services related to health and safety would be provided more quickly than others since hazardous waste, dead animals, and sewage are among the top five, and bulk items could involve health and safety issues. Further, we see the variation across cities that the hypothesis also suggested.
Service Completion Times By Type of Service, Ranked By Least to Most Number of Days to Service Completion.
*Less than ten cities included this service category.
We examine these patterns further in Figure 2, illustrating the five services delivered with the shortest service resolution time. These box and whisker plots compare the mean and standard deviation of the number of days taken to complete the various types of services and expand our knowledge of the patterns of service request resolutions. We can see the priority placed on hazardous waste requests by the low average number of days to resolve these requests and the bunching up of all the cities’ results right around this mean. No city took longer than ten days to resolve hazardous waste issues, as might be expected. For other services, the same pattern can be seen of many cities bunching up around mostly a mean of six days or less, but there are several outliers. Overall, we can conclude that there is some support for safety and health services being resolved more quickly, and we can see in the box and whiskers plot the variation that exists.
Service time distribution by category of service.
Conclusions
This article has provided an overview of 311 systems within American cities, focusing on the types of services requested by users and other characteristics. The number of service requests is highly related to city population size. However, the per capita number of requests is not associated with population size, although large cities and those with more history of 311 system usage do see more requests. More of a negative exponential pattern was seen. By far, garbage/recycling was the most often requested service across all cities, followed by code enforcement and parking. The telephone channel was still dominant in over half of the cities which contained the service channel information, but other channels, particularly mobile apps, were growing in importance.
Thirty types of services were identified that are requested by users in the 311 systems of their cities. We hypothesized that cities would prioritize resolving service requests that relate to safety and health first. There was some support for this hypothesis, with hazardous waste, dead animals, and sewage in the top five lowest service times. However, there were also deviations from the expected patterns and widely different contexts for each service within each city. Further, restrictions on the data structure (the sheer number of datasets involved and the various data structures) meant that conclusive statistical tests could not be conducted.
Finally, there is a great deal of variation across cities in how their systems are organized: whether and how cities define possible service options, the service channels used across cities, and how the request portals are built. There are no professional or data standards to organize these systems, so cities are still developing and revising their own. Examining the patterns found across cities, including the thirty identified service types presented here, represents one possible structure for the 311 service request systems.
Additional research on 311 systems and their outcomes could utilize the service type categories developed during this research. One area of further research is more study of various outcomes according to the class of service. Breaking down significant service categories like garbage and recycling, do different subtypes each have lower service times? If not, what patterns exist among these subcategories which are seen are being of greater priority than others in the same category?
Obtaining longitudinal service channel data from individual cities, combined with interviews, could allow researchers to understand the patterns of shifting service categories, including why and when residents turned to use mobile apps or a web page. This type of research can help to smooth the transition from telephones to digital channels in the future for 311 programs just being established.
More research needs to be conducted on the equity of service outcomes. Among other variables, the length of time it takes city agencies to complete requested services has differed by type of service, but only a few services have been studied (Clark et al. 2020). Additional cities and service types need to be investigated, and the questions expanded to incorporate interviews; this research can be critical to ensuring equitable distribution of services across cities.
There is much to be learned about 311 systems and the intersection of technology and service delivery. Why do different types of users use different service channels to request services? How well are services being provided? Why do different cities have such widely varied service resolution times for the same service? Researchers could assist by working on these issues and other avenues for research. Research in these areas, and others, must continue to keep pace with the growing importance of these systems to their cities.
There are numerous managerial and policy implications from this research. Standard service categories across cities would help better understand service delivery and service time performance across types of services. Research on transitions between service channels could aid those transitions in cities yet to make those transitions. Analysis should be used to develop performance measures that can benchmark and develop accountability within and across cities. Finally, best practices should be identified, developed, and shared from future research. The application of focused academic research in this field can improve urban service delivery across American cities as 311 systems spread further, across additional cities.
Footnotes
Declaration of Conflicting Interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The authors received no financial support for the research, authorship, and/or publication of this article.
