Shared e-scooter programs often generate complaints about improper parking because it creates unappealing clutter and can make it difficult for pedestrians to use the sidewalk. We examine how cities can address noncompliant scooter parking and explore public perceptions of this problem with two related studies.
First, we conducted field experiments in Washington, DC (hereafter “DC”), and Auckland, New Zealand. We observed scooter parking and then evaluated the efficacy of three interventions to mitigate noncompliant parking. We find low rates of noncompliance, but the precise rate is highly dependent on how we define noncompliance. We also find that in-app message reminders and sidewalk decals marginally lowered rates of noncompliant parking, but the largest improvement in compliance occurred after DC implemented a requirement that scooters be physically locked to bike racks.
Second, we examined public perceptions of improper parking by asking people how much noncompliant parking they think occurs and what practices they think constitute noncompliant parking. We find that the public overestimates improper parking of scooters and underestimates improper parking of bicycles and cars. Given their unfamiliarity with scooter parking regulations, respondents use pedestrian accessibility and tidiness as heuristics to determine what counts as proper scooter parking.
Our results suggest that intuitive parking solutions that align with perceptions of orderly parking, such as bike racks or on-street parking corrals, can improve rider compliance and may reduce public dissatisfaction with shared scooter parking.
Parking, sidewalk riding, and vandalism are typically the greatest sources of public conflict and contention for shared e-scooter programs (
Cities’ experiences with shared scooters are short, and both the understanding of parking problems and the best approaches to address noncompliant parking are evolving. Understanding fundamental facts about scooters—including rates of noncompliant parking—is critical to ensuring appropriate approaches to regulation of shared scooter programs and developing suitable methods to curb negative impacts. In a context where shared scooters are mainly provided by private companies, parking policy is an example where urban planners and other policy makers must “continually (re)negotiate the state-market relation” to ensure the mobility benefits of shared scooters (
We ask three questions: How can cities and scooter companies reduce noncompliant parking? Do our observed measures of noncompliant parking match public perceptions? And what counts as noncompliant parking to the public? We answer these questions through two studies that we conducted in Auckland, New Zealand, and Washington, DC (hereafter “DC”).
In the first study, we examined how scooter riders responded to three interventions designed to reduce noncompliant scooter parking. The interventions consisted of (1) in-app messages to educate and remind riders of parking rules, (2) sidewalk decals directing riders to nearby parking corrals, and (3) physical locks so scooters can be secured to bicycle racks and other parking infrastructure. Using field data collected over several weeks, we evaluated the relative effectiveness of each of these interventions in reducing rates of noncompliance. We complemented the observational study with a second study focused on public perceptions of improper scooter parking. We used intercept surveys conducted in Auckland and DC to compare our observed rates of noncompliant scooter parking with public responses to probe the public’s understanding of scooter parking.
Though it is unusual to include two distinct studies in a single paper, the findings from the two studies complement and enrich each other. The observational study provides objective data about the frequency of noncompliant parking and points toward policies to mitigate this issue. But we have reason to believe that the observational data does not match the public’s perceptions of improper scooter parking. In our earlier field study of scooter parking, we observed low rates of scooters impeding pedestrian access (
The following section summarizes the scholarly and gray literature on scooter parking. We then present the methods and findings from the first study and the second study in sequence. We conclude with a general discussion that ties the two studies together and offers suggestions for policy.
In this paper, we use three main terms to describe scooter parking behaviors. The first two terms,
Studies of shared micromobility media coverage suggest that scooter parking behaviors are a common concern. Gössling (
Researchers have also surveyed riders to understand scooter parking. Multiple surveys have found that parking rules are not well understood (
Scooters are not the first disruptive mode to cause parking consternation. The advent of automobiles spurred new parking solutions: garages, on-street parking metering, and other measures (
Cities have adopted three main approaches to shared scooter parking: regulations, incentives, and infrastructure. Across all three approaches, cities have sought to enforce proper parking through shared micromobility companies’ digital platforms and traditional approaches (e.g., ticketing).
Parking regulations for scooters vary substantially across different cities: what is permitted in one city may be prohibited in another. Some regulations are common, such as the requirement for users to park scooters upright. Other regulations are more parochial; only one-third of US cities prohibit parking on vegetation or landscaping, in part to reduce the likelihood of a parked scooter falling down on an uneven surface (
Beyond rules and regulations, cities are also exploring approaches to encourage parking compliance. For example, many cities have installed designated scooter parking corrals, which take various shapes and sizes, on streets and sidewalks. A recent before-and-after study in Oslo and Trondheim, Norway, analyzed the effect of adding scooter parking racks and painted corrals on public rights-of-way (
The third approach to scooter parking is using vehicle design standards. Some cities have enacted vehicle lock-to requirements aimed at encouraging proper parking. As of October 2021, just four US cities (Chicago, Minneapolis, San Francisco, and DC) had imposed lock-to requirements (
Attempts to regulate micromobility parking compliance and enforcement via digital platforms met with mixed success. Dunn (
Cities across the globe are looking for ways to regulate scooter parking but have little guidance on what works. In this first study, we evaluated three common approaches intended to reduce rates of noncompliant parking in two cities. The first two interventions focus on rider education, given the difficulties previously described with digital enforcement approaches (
We conducted field observations and experiments in Auckland and DC. We chose these cities for their relevance to peer cities, their long histories with shared micromobility, and the feasibility of collecting data during the ongoing COVID-19 pandemic. The cities were also the settings for previous research on scooter parking (
In each city, we conducted systematic field observations of all parked
Our data collection instrument builds on previously developed methods used by Fang et al. (
Before conducting any experimental treatments, we collected baseline data to capture existing scooter parking behaviors across all shared scooter companies in operation. In each city, study areas included parking corrals, where riders could choose to park scooters but were not required to do so. Likewise, riders were permitted to park in the furniture zone (but not in bike racks in Auckland). In Auckland, some companies messaged simple in-app reminders to riders about local parking rules; by contrast, no such in-app message reminders were sent to riders in DC.
One week after collecting baseline condition observations, we implemented in-app messages for one scooter company: Lime. Implementing messaging for a single scooter company allows us to compare parking behaviors both against the baseline data and across riders who received targeted parking messages (Lime riders) versus those who did not (non-Lime riders).
In-app messages in Auckland (week 2 intervention).
In the third week, we installed temporary sidewalk decals in each city’s study area and returned the in-app messages in the Lime app to the baseline state (i.e., reactivated the old in-app messaging in Auckland and removed in-app messaging in DC). The sidewalk decals were placed at about 330–660 feet (100–200 meters) from the focal parking corral sites (see
Sidewalk decals in Auckland and DC (week 3 intervention).
On October 1, 2021, DC began requiring all shared scooters to be equipped with locking mechanisms and that users lock the scooters to bike racks or other street furniture. We took advantage of this new requirement to evaluate how this new regulation affected scooter parking. To avoid collecting data while micromobility operators, riders, and the city alike were adjusting to the new parking regulation, we waited until the third weekend of October to collect field observations with the lock-to requirement in place (we conducted our observations for the baseline and first two treatments during the previous month). We also conducted two weeks of observations of the lock-to period to increase the sample size for this phase.
Our findings correspond to the times and locations where we conducted observations. Although they have strong internal validity to identify whether experimental treatments were effective, they may not represent parking patterns in other times or places, even within the same city. As others have shown, aspects of the built environment can strongly influence parking compliance (
The in-app message was shown only to Lime riders. It is possible that Lime riders differ from users of other shared scooter services. However, scooter companies generally charge similar prices and offer comparable vehicle models, leading to relatively small differentiation. Likewise, riders tend to choose which scooter they will ride based on availability rather than strong brand preference. We therefore suggest that differences in behavior across different shared scooter companies are likely to be minimal.
In September 2021, local TV and print outlets in DC reported on the coming regulatory changes that would require riders to physically lock scooters to bike racks (see, for example,
Finally, our findings are limited to examining the effects across three policy interventions: we are not able to infer the potential efficacy of other parking-oriented policies that cities have also enacted, such as no-parking zones and providing discounts for trips that end in preferred parking areas.
In this subsection, we describe how the three interventions affected scooter parking. Throughout our discussion, we focus on the rates of scooters parked out of compliance and scooters impeding access (
Noncompliant parking and impeding access across three policy interventions.
Scooter parking locations by intervention.
Prior to the interventions, 5% to 6% of shared scooters impeded pedestrian access in both cities, consistent with previous research. A higher proportion—roughly one in six or five (15% or 19%)—of scooters violated local parking regulations. This estimate, however, is influenced by local parking rules, which vary by city (e.g., parking a shared scooter in a bike rack is banned in Auckland but not in DC).
In both cities, the most common parking location, by far, was the furniture zone (64%–71%). In Auckland, parking corrals were the second-most-common parking location (19%), while they were used infrequently in DC (3%). In DC, the second-most-common parking locations were the middle of the sidewalk (12%) and against a building (10%). Scooters were rarely parked at bicycle racks (1.5% in Auckland and 3% in DC).
We observed a weak response to the in-app message in Auckland and a somewhat larger improvement in parking compliance in DC. Unlike Auckland—which already had simple in-app message reminders about proper parking prior to the research intervention—DC had no in-app reminders at baseline. Therefore, in-app messages may improve parking compliance, though it may be the introduction of a reminder—rather than the use of specific wording or imagery—that sways behavior.
In Auckland, introducing refined versions of in-app message reminders for Lime riders led to only a slightly lower rate of shared scooters impeding access compared to baseline (4.9% to 3.1%; Fisher’s exact test: p > 0.05). Furthermore, rates of impeding access were similar for both Lime vehicles (whose riders received the message) and non-Lime vehicles (whose riders did
After the introduction of in-app messages, noncompliant parking in DC declined from baseline for Lime scooters (19% to 3.8%; Fisher’s exact test: p < 0.05) but only slightly (19% to 13%) for other companies’ scooters. Likewise, the rate of shared scooters impeding access dropped from 5.9% to 0% for Lime scooters (though the sample size lacks sufficient power to identify a statistically significant effect; Fisher’s exact test: p > 0.05), while other companies’ vehicles continued to impede access at rates similar to the baseline (5.9% to 5.2%).
The introduction of sidewalk decals did not lead to a statistically different rate of noncompliant parking or impeding pedestrian access in either city compared with the baseline. We did observe increased rates of parking in parking corrals and bike racks.
In Auckland, sidewalk decals had a small effect on parking behavior. The percentage of shared scooters that impeded access declined from 4.9% to 2.5% (Fisher’s exact test: p > 0.05), while the percentage of parked scooters noncompliant with local regulations fell from 14% to 11% (Fisher’s exact test: p > 0.05). The effect of decals on overall measures of parking compliance in DC was similarly minimal: impeding access stayed effectively constant (5.9% to 5.8%; Fisher’s exact test: p > 0.05), and noncompliant parking according to regulations dipped from 19% to 16% (Fisher’s exact test: p > 0.05).
The decals appeared to be effective at increasing parking in corrals in Auckland. The proportion of shared scooters parked in a corral increased from 19% to 26% in week three, a 26% increase (Fisher’s exact test: p < 0.05). In DC, corral parking
The lock-to requirement had a large effect on parking compliance and location. Following the implementation of the lock-to requirement in DC, the rate of parked scooters impeding pedestrian access decreased by over half (5.9% to 2.4%; Fisher’s exact test: p < 0.05), and we observed a robust one-third decrease in noncompliant parking (19% to 13%; Fisher’s exact test: p < 0.05).
We also observed major shifts in where shared scooters parked. Bike rack use increased more than twelvefold compared to baseline (3.2% to 40%; Fisher’s exact test: p < 0.001), fueled by sharp declines in parking in the furniture zone (71% to 47%; Fisher’s exact test: p < 0.001), parking against a building (10% to 4.4%; Fisher’s exact test: p < 0.005), and parking in the middle of the sidewalk (12% to 5.3%; Fisher’s exact test: p < 0.01).
The experiment yielded two key insights. First, rates of noncompliant scooter parking are low, but the actual rate of compliance greatly depends on local regulations. Second, while in-app messages and sidewalk decals can encourage marginal changes in parking behavior, the introduction of lock-to requirements had the greatest improvement in parking compliance and shift in parking locations.
In our second study, we sought to understand perceptions of scooter parking by surveying members of the public. We focused on two questions: How much improper parking does the public think is happening? And what do respondents think counts as improper parking?
Research assistants in Auckland and DC conducted intercept surveys of pedestrians in the same study areas as described in Study 1. Research assistants invited participants to take the survey immediately by scanning a QR code and completing the survey on their smartphone or later on their smartphone or computer. In Auckland, 144 respondents started the survey, and 125 completed it; in DC, 71 people started, and 58 completed. Survey respondents were nearly evenly split between female (51%) and male (46%), were 31 years old on average, and overwhelmingly (92%) worked, lived, or went to school in the study city. The majority (60%) had never taken a trip on a shared scooter.
The survey focused on two topics. First, we asked respondents to estimate the share of scooters, bicycles, and cars that are improperly parked in the study city. Second, we assessed their perceptions of scooter parking by showing them only three randomly selected parking scenarios out of the ten scenarios (see
Estimates of improperly parked scooters in Auckland and Washington, DC.
Scenarios displayed in survey to test knowledge of parking regulations and assess perceptions of clutter.
Our survey samples are likely not representative of the public that has seen scooters parked in the two focal cities, but assessing the representativeness of the intercept survey is difficult. The relevant population of interest was individuals who live, work, or otherwise pass through the study areas in the two cities; the attributes of these populations are unknown. Approaches to identify a close match for the population (e.g., the Census statistics of those living in the study areas) for weighting would simply introduce additional uncertainty in our estimates, so we present unweighted survey results.
Respondents overestimated the prevalence of improper scooter parking (
Respondents also overestimated rates of noncompliant bicycle parking, but they said they believe it occurs less frequently than noncompliant scooter parking. The median response was that 5%–10% of bicycles are improperly parked;18% of respondents said more than 30% of bicycles are improperly parked, and 33% of respondents estimated that fewer than 5% of bikes are improperly parked. By contrast, previous studies suggest that fewer than 1% of bicycles impede pedestrian access (
While respondents
We observe substantial variation in public estimates of improper scooter parking. This may reflect disagreement about what constitutes improper parking. Additionally, parking may be a low-salience issue that most respondents rarely think about and do not have a solid understanding or intuition of. Finally, the wide range of estimates could simply reflect innumeracy among the public (
We also wanted to know whether public perceptions of improper scooter parking extend to the transportation professionals who regulate and manage shared scooter programs. To understand how public perceptions compared to those of transportation professionals, we also surveyed participants at four professional transportation conferences or meetings where one of the authors was presenting completed research on scooter parking. Transportation professionals held similar perceptions about rates of improper parking compared with intercepted survey respondents. The median transportation professional respondent reported that 20%–30% of scooters are parked improperly, compared with 10%–20% of bicycles and cars. We include more details about the survey of practitioners and findings in Appendix B.
To better understand the gap between public perceptions and field observations, we surveyed the public about its understanding of improper scooter parking using ten scenarios (
Response patterns showed clear evidence that pedestrian accessibility is a primary condition the public uses to assess proper parking. The overwhelming majority (85%–90%) of respondents assessed bike racks, parking corrals, and scooters neatly arranged in the furniture zone (“tidy” in
Perceptions of noncompliant parking and clutter for ten scooter parking scenarios.
Study areas in Auckland and DC.
Conversely, the public decisively identified all four of the scenarios that may pose accessibility hazards—tipped-over scooters or scooters parked in front of a doorway, in front of a curb cut, or in the middle of the sidewalk—as noncompliant with local parking rules (85%–100% of respondents indicated the scooters in these scenarios were improperly parked).
The public appeared to conflate “tidiness” with proper scooter parking. We presented respondents with two similar scenarios where scooters were parked in the furniture zone of a wide sidewalk, with one scenario showing “tidy” scooter parking and the other “messy.” The tidy scenario has a wide sidewalk with three parked scooters neatly aligned parallel with one another in the furniture zone. The messy scenario shows the same three scooters on the same wide sidewalk but parked at different angles from one another (see
Our survey results highlight a disconnect between the public’s perceptions of scooter parking and systematic field observations of scooter parking. The public’s perception of improper parking is an overestimate of the observed rate of noncompliant scooter parking, while at the same time, the public underestimates the rate of noncompliant car parking, relative to previous research (see
Furthermore, we identified two key considerations determining the public’s perception of improper scooter parking. First, the public clearly identifies pedestrian access as a crucial determinant of compliant parking. Second, the public strongly considers tidiness when gauging proper versus improper parking. A simple manipulation of one parking scenario to alter the orientation of parked scooters relative to one another into either tidy or messy configurations produced dramatic swings in public perceptions of scooter compliance and clutter. The importance the public places on tidiness in determining compliant parking may explain some of the public backlash against scooters, even when scooters are rarely parked in ways that violate local rules or impede travel by others. But scooters may offend people’s aesthetic sensibilities or pose a nuisance at times (e.g., a crowded sidewalk during peak rush hour).
Improper scooter parking is one of the most common complaints about micromobility across the globe. We believe that this problem is widely misunderstood, which in turn distorts policy responses and distracts from much more important transportation problems.
We conducted two studies in Auckland, New Zealand, and Washington, DC, using field experiments and surveys. In our field experiments, we observed scooter parking and evaluated rates of noncompliant parking (measured as both impeded access and noncompliant parking according to local regulations) before and after several interventions. Our field experiments coincided with introducing a lock-to requirement for scooters in DC, allowing us to analyze how this new regulation shifted scooter parking behaviors.
People overestimate rates of improper scooter parking relative to the share of scooters that we observe violating local regulations or impeding access by other travelers. Perceptions of rampant improper parking are pervasive, and even riders themselves think that many or most other riders are bad actors (
Planning and policy discussions about scooter parking need to start from a place of shared understanding—namely, what counts as improper parking. We find that public perceptions of improper parking are largely driven by concerns about pedestrian accessibility and an aesthetic sense of tidiness and order. Simply varying the orientation of parked scooters in our survey scenarios flipped public perceptions of the scooters from tidy and compliant to cluttered and noncompliant. The appearance of “clutter,” when riders do not park their scooters in a tidy fashion, may lead to higher perceived noncompliance than occurs when judged by local regulations or impedance.
We can test our hypothesis that perceptions of improper parking are shaped by concerns about pedestrian accessibility
Public perceptions are important in both shaping the narrative around scooter parking and influencing the policies that govern scooter parking. A dissatisfied public can engage in policy debate by writing to elected officials, submitting public comment at city council meetings, and complaining to city agencies that oversee shared micromobility services. Complaints and concerns may shape subsequent policy measures, professionals’ perceived salience of different issues, or both. Indeed, we find that transportation professionals’ views mirror the public’s and that many greatly overestimate rates of improper scooter parking, which could affect the types of scooter regulations that cities enact.
Each policy intervention we tested improved scooter parking compliance, but to differing degrees. Implementing in-app messages and installing sidewalk decals led to small but meaningful improvements in noncompliant parking and impeding access, and DC’s citywide lock-to requirement dramatically improved scooter parking compliance. Yet, this latter requirement is not without potential conflicts. City staff often cite bicyclist complaints and concerns that scooters will consume all existing bike rack spaces if allowed or required to park on lock-to racks (
Scooter parking regulations vary from city to city and are, at times, at odds with public perceptions about what constitutes proper parking. Expecting riders to know the intricacies of scooter parking regulations (
To put it simply, we suggest making parking rules obvious and intuitive. Allowing scooter riders to use intuitive solutions as compliant parking is likely to have several related benefits, including reduced noncompliant parking and improved public perceptions. In Chicago, for example, the introduction of physical locks resulted in 97.3% compliance in parking audits and a dramatic reduction (78%) in complaints (
Cities should provide more physical infrastructure to park micromobility vehicles, both shared and privately owned. One key advantage of physical infrastructure, such as bike racks and parking corrals, is that it communicates parking regulations to riders
Micromobility operators can also improve parking compliance through rider education and communication. As this research showed, introducing a simple in-app message reminder can improve parking compliance where no message currently exists. Likewise, the introduction of signage, in this study in the form of a sidewalk/footpath decal, can improve parking compliance through further institutionalizing proper parking, as seen historically with the development of car parking signage. While results suggest that it may be challenging to eliminate all noncompliant parking, a combination of interventions can bring noncompliant parking down to even lower rates.
Finally, we urge policy makers to focus on straightforward, effective solutions rather than flashy technological fixes. Implementing parking regulations and enforcement through digital platforms are unlikely to work and only communicate rules to riders while remaining invisible to the public (
Data for this article are available at
In the case of predicted rain or excessive heat, we shifted data collection to a Wednesday or Sunday or postponed it until the following weekend.
We have made a version of the data collection instrument freely available online for others to use:
In Auckland, a scooter violated local parking regulations if it was (a) parked in the middle of the footpath, (b) parked in a bike rack, (c) tipped over, (d) blocking an entrance to a building, (e) blocking a crosswalk, (f) parked such that a meter or less of footpath space remained to walk past the scooter, or (g) parked on a tactile dot pad. In Auckland, a scooter was determined to impede access if it was (a) blocking an entrance to a building, (b) blocking a crosswalk, or (c) parked such that a meter or less of footpath space remained to walk past the scooter.
In DC, a scooter violated local parking regulations if it was (a) parked in the middle of the sidewalk, (b) tipped over, (c) blocking an entrance to a building, (d) blocking a crosswalk, (e) parked such that 3 feet or less of sidewalk space remained to walk past the scooter, or (f) parked on a tactile dot pad. In DC, a scooter was determined to impede access if it was (a) blocking an entrance to a building, (b) blocking a crosswalk, or (c) parked such that 32 inches or less of sidewalk space remained to walk past the scooter.
The exact method changed slightly between the two cities. In Auckland, respondents were shown one image each from three scenario groups. The three groups were organized thematically: one group that we anticipated most respondents would rate as compliant and tidy, a second group of scenarios that respondents would rate as noncompliant and cluttered, and a third group that was mismatched—either compliant but cluttered or noncompliant but tidy. In DC, we showed respondents four images but did not organize them into thematic groups. We also showed respondents only one of the two randomly assigned questions (compliance or clutter) for all four images. We made this change to avoid biasing respondents to answer consistently for each image regarding the scenario’s compliance and clutter. We did not observe marked differences in responses in DC compared to Auckland and therefore feel confident in combining the two sets of responses despite the slight adjustment to the methods.
We compared public perceptions from the intercept surveys to the perceptions of transportation professionals, whom we surveyed during four webinars between February and October 2021 at which one of the authors was presenting completed research on scooter parking. These events were the International Parking and Mobility Institute (IPMI) conference (February 24, 2021); the Oregon Active Transportation Summit (OATS) conference (April 30, 2021); a Dublin, Ireland, webinar about scooter parking policy (July 15, 2021); and the North American Bikeshare & Scootershare Association (NABSA) conference (October 28, 2021). We collected responses from 158 respondents across the four events (85 at the IPMI conference, 31 at the OATS conference, 21 at the Dublin webinar, and 21 at the NABSA conference). We refer to these respondents as “practitioners” or “professionals,” though we acknowledge that we do not have information about their professional occupations.
We surveyed the audience at the start of our presentations. As the format was restricted in time and survey capability, we only asked what percent of scooters are improperly parked (all four events) and what percent of bicycles and cars are improperly parked (IPMI and NABSA only).
The polls we conducted during our webinars are convenience samples. We believe they include a broad cross-section of groups, including those focused on parking, on micromobility, and on active transportation, though we did not collect demographic or occupational information.
Practitioners’ estimates of improper scooter (
Estimates of improperly parked scooters by transportation professionals and the public.
Estimates of improperly parked bicycles and cars by transportation professionals and the public.
Thank you to Na Zhou, Milad Jabbari, Rahul Kadam, Savitri Savitri, Julia Wicks, Benhaz Raouf, and Ugonna Njeze for their data collection efforts in Auckland and Washington, DC. Niki Dass, Lauren Mentjox, Robert Gardner, and Michael Kurlancheek of Lime and Sharada Strasmore, Greg Matlesky, and Javier Griffiths of DDOT helped with logistics, implementation, and study materials. Thank you as well to the organizers of the International Parking and Mobility Institute conference, the Oregon Active Transportation Summit, the Dublin webinar (Sebastian Schlebusch, Sarah Meade, Hannah Daly), and the North American Bikeshare & Scootershare Association for allowing us to survey attendees. We are also grateful to Joachim Scheiner, Giulio Mattioli, Dan Work, Caroline Janssen, and Subeh Chowdhury for reviewing various drafts of the survey and to Shari Shapiro, Ashley Scott, Sam Sadle, and Brandon Haydu for their support of this study.
Lime funded the research assistants, sidewalk decals, and incentives for the intercept survey.
The authors have no competing interests to declare.
The authors confirm contribution to the paper as follows: study conception and design, A. Brown, N. Klein, and C. Thigpen; data collection, A. Brown, N. Klein, and C. Thigpen; analysis and interpretation of results, A. Brown, N. Klein, and C. Thigpen; draft manuscript preparation, A. Brown, N. Klein, and C. Thigpen. All authors reviewed the results and approved the final version of the manuscript.