4 Discussion
We readily acknowledge several limitations of this study, particularly in the survey design and methodology. The interviews were conducted as a convenience sample rather than with a rigorous sampling strategy, with the statistical caveats resulting from that design decision. The sample is also too small to have substantial statistical power, particularly in statistics calculated on multiple grouping dimensions. Finally, the survey collected self-reported responses with no verification or validation of any kind.
Most survey responses were collected on fixed rail transit station platforms. Passengers of UTA rail services were assumed to be the primary audience for the microtransit service, and these riders were presumably more likely to be available to complete a survey while waiting for a train. Additionally, UTA is interested in supporting its fixed rail transit investments in the service area. There is, however, no requirement that microtransit passengers use other UTA services; data supplied by the microtransit provider but not included in this study suggest that only 58% percent of microtransit trips began or ended within 500 feet of a UTA rail transit station. This population might have preferences or patterns that either match or contradict the initial findings of this research.
A final limitation of these findings is the onset of the COVID-19 pandemic. Government-imposed shutdowns and voluntary work stoppages related to the pandemic did not begin in Utah until the week of March 15th, after data collection for this project had completed. As such, the survey responses are likely unaffected by changes in behavior related to the pandemic. However, the pandemic has drastically affected the subsequent operations of both UTA and Via and is likely to change many of the stated behaviors and attitudes reported in this study. Many findings of this study will need to be reconsidered should “normal” operations resume.
In discussing the responses to the question of what mode microtransit passengers would have used were the service not available, we suggested there is anecdotal evidence that commercial TNC rides are the primary competition. There are still questions, however, of how use of this microtransit service might affect conventional transit services. Table 4.1 shows the average weekday ridership during November, December, and January for the period the microtransit service was operating as well as the same three months in the two prior years [uta2020boardings}. Total system ridership was remarkably stable during these three periods. The microtransit service area – in this case defined by ridership on routes F514, 218, 526, F504, F518, F534, F546, and F547 – was declining before the microtransit service began, though the decline accelerated during the first three months of the service’s operation. By comparison, the microtransit service carried approximately 316 passengers per day during its first three months, more than compensating for the recent observed decline in transit ridership were this to be identified as a major contributing factor.
ridership <- read_csv("data/UTA_Route-Level_Boardings%2C_Monthly_Counts.csv") ## Parsed with column specification:
## cols(
## Mode = col_character(),
## LineAbbr = col_character(),
## Month = col_character(),
## Year = col_double(),
## ServiceType = col_character(),
## AvgBoardings = col_double(),
## City = col_character(),
## County = col_character(),
## ObjectId = col_double()
## )lines <- c("F514", "218", "526", "F504", "F518", "F534", "F546", "F547")
rdf <- ridership %>%
mutate(
date = lubridate::as_date(str_c(Year, Month, "01", sep = "-")),
area = ifelse(LineAbbr %in% lines, "area", "other")
) %>%
filter(ServiceType == "WKD") %>%
filter(Month %in% c("November", "December", "January")) %>%
mutate(
period = ifelse( Month %in% c("November", "December"), Year + 1, Year )
) %>%
filter(period >= 2018) %>%
group_by(period, area, LineAbbr) %>%
summarise(AvgBoardings = mean(AvgBoardings)) %>%
group_by(area, period) %>%
summarise(br = sum(AvgBoardings)) %>%
mutate(ch = (br - lag(br)) / lag(br) * 100)
uta_ridership <- rdf %>%
gather(key = count, value = value, -area, -period) %>%
unite("key", area, count) %>%
spread(key, value) %>%
mutate(period = c("2017-2018", "2018-2019", "2019-2020"))
if(knitr::is_latex_output()){
kable(uta_ridership,
col.names = c("Year", rep(c("Avg. Weekday Boardings", "% Change"), 2)),
digits = 2, caption = "Average Weekday Ridership, November through January",
booktabs = TRUE, format = "latex") %>%
add_header_above(c(" ", "Microtransit Service Area" = 2, "Other UTA Services" = 2))
} else {
kable(uta_ridership,
col.names = c("Year", rep(c("Avg. Weekday Boardings", "% Change"), 2)),
digits = 2, caption = "Average Weekday Ridership, November through January") %>%
add_header_above(c(" ", "Microtransit Service Area" = 2, "Other UTA Services" = 2)) %>%
kable_styling()
}| Year | Avg. Weekday Boardings | % Change | Avg. Weekday Boardings | % Change |
|---|---|---|---|---|
| 2017-2018 | 1179.33 | 147410.0 | ||
| 2018-2019 | 1125.00 | -4.61 | 146743.0 | -0.45 |
| 2019-2020 | 970.33 | -13.75 | 147009.8 | 0.18 |
In spite of these limitations, the findings of this research suggest potential paths for transit agencies considering the deployment of a transportation mode of this kind. First, the negative result with respect to income is somewhat promising: an inability to reach out to low-income individuals was a factor in the failure of Kutsusplus (Weckström et al. 2018). The significant findings — a relationship with age in the original data and suggestive relationships with also with household size and transit frequency — also hold meaning for transit providers. Of particular note is the absence of a middle ground or neutral opinion on the service for the largest age group in the survey, those individuals between 25 and 44 years old. In the next older group (45 to 64 years old), a neutral opinion is considerably overrepresented. Does this mean that members of this older group could be a target of successful marketing efforts? How much of these attitudes are actually tied up in covarying household conditions such as vehicle availability and household size? More research is necessary.
4.1 Conclusion
Microtransit services are regularly put forward as means to support last-mile / first-mile trips on fixed route transit systems, and several such systems have been deployed in the recent past. This paper presented initial findings from a quick response survey aimed at learning who was most willing to use a new service within weeks of the system launch. These initial findings suggest first that younger adults are most willing to consider using microtransit services, especially in larger households. Additionally, these services compete most directly with commercial TNC ridehail offerings in addition to fixed-route transit services.
Though preliminary, it is worth considering how these findings might transfer to projects in other cities. The spatial and infrastructure context of the region has played an important role in the UTA On-Demand’s overall success. A low-density but rapidly developing suburban region bracketed by multiple high-frequency and high-capacity rail lines provides an ideal environment to test the potential of microtransit as a first-/last-mile access technology. The results of this study specifically suggest that younger adults and those with larger households express a higher willingness to use microtransit services. Salt Lake County and Utah more generally has a large population that matches this description, with a high share of young adults and a high birth rate relative to the United States average (Utah Department of Health 2020). Planners considering implementing microtransit services to support station access might consider the demographic characteristics of the population in their target areas to maximize the project’s success.
Transit passenger intercept surveys are an important method to determine who is and who is not using a microtransit service, paired with demographic characteristics and trip purpose information. To understand the rider characteristics and trip purposes specifically of microtransit users, by contrast, better survey methods are needed. In particular, a survey pushed through the smartphone application used by the passengers would help in reaching a considerably larger sample. It would also be theoretically possible in that case for the researchers to pair the survey responses with actual observed trip patterns for distinct users including origin, destination, and route GPS points, regularity of use and variance in use patterns, and many other data variables. Obtaining these data and conducting responsible research with them should be a priority for the service operators and their agency partners.