Estimating the Effective Population of Nantucket

Sponsor:	Nantucket Data Platform
Sponsor Liaisons:	Peter Morrison, Joe Smialowski, Alan Worden
Student Team:	Frank Campanelli, Tyler Donovan, Alec Wehse, Sam Winter
Abstract:	This project, in collaboration with the Nantucket Data Platform (NDP), studied the effective population of the Nantucket community. We gathered data from several sources on Nantucket and the StreetLight Data service to gain insight into the dynamic population changes Nantucket experiences every year. The main deliverables from these endeavors were to communicate to the NDP viable methods for estimating the Nantucket population, recommendations for future analyses, and more specific population and travel data requested by community stakeholders.
Link:	Report: NDP Final Report Presentation: NDP Final Presentation

Executive Summary

Public, private, and nonprofit sectors need population data to make informed decisions on everything from marketing and planning strategies to policy decisions. Most towns and cities rely on government census data, but census data is insufficient for resort towns like Nantucket that experience highly variable seasonal populations. There is a lack of accurate population data on Nantucket, which can cause policies and planning decisions to be implemented with inadequate supporting evidence. In 2017, the editor of the Nantucket Inquirer & Mirror, Joshua Balling, wrote “The Town of Nantucket is essentially a $100 million corporation. Yet the Board of Selectmen, Finance Committee, and other agencies are often forced to make decisions on the fly, relying on outdated or incomplete information, particularly about the island’s population size” (Balling, 2017).

The Nantucket Data Platform (NDP) was founded in the spring of 2017 with the goal of collecting population and demographic data on Nantucket and making it publicly available to organizations for use in the decision-making process. The NDP plans to consolidate data from a variety of local and other data sources. The goal of this project, in cooperation with the NDP, was to determine effective methods of evaluating population data specific to the region of Nantucket. We primarily investigated methods of population and demographic data collection within the StreetLight Data platform. StreetLight is a third-party company that purchases cell phone mobility data from service providers and provides it to customers for analysis. We compared our findings from StreetLight to sources on Nantucket, such as the Steamship Authority passenger counts and municipal solid waste reports, in an effort to validate the information collected from StreetLight. To summarize, we worked with the NDP to complete the following objectives:

1. Objective 1: Determine best practices for the collection, analysis, and use of population data in resort communities
2. Objective 2: Evaluate stakeholder needs for demographic data to strengthen evidence-based decision making, and to learn about seasonal employment on Nantucket
3. Objective 3: Evaluating the usefulness of Streetlight data, combined with other data sources, for population estimation.

During our study of the Nantucket population, we analyzed a variety of data sources to determine their effectiveness and reliability. We deemed the following sources of data relevant to determining the effective population of Nantucket:

1. Nantucket Street Census: The Street Census from the Nantucket Town Clerk’s office is the most complete count of permanent residents on Nantucket, as it is the most recent and the most thorough.
2. Transportation Data: Ferry and airport transportation data from the Nantucket Town Planning Office displays total ferry ridership and airport departures for each month of the year.
3. Solid Waste: Solid waste data from the Nantucket Department of Public Works (DPW) shows municipal solid waste generation going back to July of 2014.
4. StreetLight Data: Cell phone mobility data from StreetLight depicts the movements and certain characteristics of people to make inferences about the demographic characterization.

StreetLight tracks people that use certain cell phone applications and will present information on these people when they pass through zones that are set by the user. Zones are areas of interest that can be set to any shape and size for analysis. StreetLight takes the data from the selected zones and represents it on a Visitor Activity Index, a metric used by StreetLight to display data while maintaining strict privacy for the individuals tracked. StreetLight defines the Visitor Activity Index as “a measure of the relative volume of visitors to the zones” (StreetLight). The values are provided on an index and do not indicate the exact number of visitors. Values can be compared to other Zones in the same Project, or to Zones in other Visitor Projects for scaling and comparative purposes.

The analyses in this report relied on both direct and indirect approaches to population estimation. Direct approaches include methods that individually count a population. Many of the analyses conducted are based around one such direct approach which is the yearly street census conducted by the Nantucket Town Clerk. This census is the most complete list of permanent residents on Nantucket, and is the backbone for many of our estimations. Indirect approaches use symptomatic variables, which refer to measurable factors affected by population fluctuations. We produced population estimates using two different symptomatic variables: StreetLight Visitor Activity Index and solid waste production.

When extrapolating these two data sets, we anchored the two estimates to the same point in time, as well as the same base population number, in order to have comparable results. We used an “anchor month” to relate Street Census population counts to the population during a single month of the year. Based on solid waste data, we determined that the population was lowest in February each year, making February a suitable anchor month to relate to the known permanent resident population. We regard solid waste as a valid benchmark, given its consistency over several years. The anchor month was used to scale data, such as the StreetLight activity index and solid waste production, from a symptomatic variable into total population counts.

Our best estimate for the February population in 2016 was our middle estimate calculation, which was 18,627 people. We scaled the Street Census population of 13,200 by a 77.7% non-response rate to get 16,984 as a total permanent resident estimate, which also acts as the lower bound for the population of Nantucket. We then utilized ferry and airport travel data, and assuming an average visitor stay of 2 days, estimated that on average there were 1,643 visitors on Nantucket at any given time. This then increases the lower bound to our 18,627 middle estimate. Figure 1 displays the StreetLight and solid waste population estimates, scaled up using our middle estimate.

Figure 1: StreetLight vs: Waste Population Estimates for 2016

We reached these estimates using the following set of assumptions:

• Each head of household on Nantucket is documented by the Town Clerk and received a copy of the Street Census survey.
• The Town Clerk’s office maintains a complete list of non-respondents for each census year.
• The non-respondents to the Street Census have the same average family size as the respondents.
• All permanent residents are present on Nantucket and are counted as spending the majority of February on Nantucket.
• Solid waste data is a reliable symptomatic variable of population, which indicates that the population is lowest in February for 2015-2017.
• Equal trash generation occurs across the demographic spectrum.
• Visitors to Nantucket in February stay an average of 2 days.
• Arrivals and departures are equal across all modes of transportation to and from Nantucket in February.
• There are no permanent residents using the ferry in February.
• The number of visitors traveling to and from Nantucket using private boats and aircraft in February is negligible.

The use of StreetLight to estimate the population of Nantucket will require better understanding of how subpopulations are proportionally represented in StreetLight data. Population estimates using Streetlight deviate considerably from population estimates using solid waste during the summer months, when many visitors are on the island. This implies that StreetLight may track visitors more often that permanent residents. One of the main assumptions used when analyzing data from StreetLight was that StreetLight collects a representative sampling across the different demographic groups. This means that every increment on the Visitor Activity Index represents the same number of people. There is currently no information on whether StreetLight tracks various demographic groups at different rates. This is because StreetLight collects its data through mobile device applications, some of which may be used by visitors more than permanent residents. For example, a tourist visiting Nantucket may tend to use navigational applications more often, as they will more often need assistance getting from place to place, compared to a year-round resident who already knows their way around the island.

StreetLight data can be useful for exploring sub-populations based on information about home and work locations. We created a set of heuristics based around the Visitor Home and Work Analysis, which shows the distance from where a person was tracked to their home and work, to infer the demographic groups of people on Nantucket. These groups include: permanent residents, seasonal residents, commuters, and tourists. Our heuristics are defined in the table below.

Table 1: Demographic Breakdown

Because StreetLight does not recognize a home unless the resident stays there for 18 days of out a month, the seasonal resident population of Nantucket remains unknown. In 2015, 64% of homes on Nantucket were considered seasonal (Nantucket Housing Issues, 2015). Using this data, combined with a study on the average time of stay for seasonal residents, may allow the NDP to achieve a better understanding of the seasonal resident population on Nantucket.

Based on our findings from the development of the anchor month, conclusions drawn from StreetLight and other data sources, and stakeholder feedback, we recommend that the NDP:

• Continue to update and support the anchor month as new analyses and data become available
• Examine how StreetLight tracks various subpopulations, and determine a relationship between visitors and permanent residents that can be used to increase the accuracy of population estimates.
• Refine and further develop the heuristics associated with home and work distance to more accurately define the demographic groups, with particular interest in separating out seasonal residents using StreetLight in conjunction with other available data sources
• Attempt to recover departure and arrival data for Hy-Line and Steamship ferries, as well as Nantucket Memorial Airport passengers, organized by trip
• Consider studies on the amount of solid waste produced per capita to compare the derived use per capita used in our solid waste analysis.