Methods: Analysis

Analysis Strategy

We analyzed various indicators over nine geographic regions. Our first four regions included the lower 48 U.S. counties (parishes, etc.), the New York State towns (boroughs, etc.), the 61 Park Towns wholly inside the Park, and the 31 Split Towns that straddle the Park boundary.

We used the USDA Rural-Urban Continuum Codes of all U.S. counties (parishes, etc.) “4” through “9” to define the subset of U.S. Counties comprising our USDA rural regions.

We used the weighted mean population density of the 61 Park Towns (the overall mean population density) to define other regions. We selected the subset of U.S. counties whose median population density precisely matched the weighted mean population density of the 61 Park Towns. We called this subset of counties the “low density” counties. To select a set of New York State towns matching the Park Towns, we instead targeted the median. The resulting subset of 47 “low density” New York towns were comparable to the set of 61 Park Towns in median town population density.

We created two further subsets of the aforementioned USDA subset and “low density” U.S. County subset by including just those counties inside the nine northeast states from Pennsylvania and New Jersey up to Maine. We identified these regions with the suffix “Northeast” or “NE”.

We used all of the above nine defined regions in our analyses. We have posted on the Protect the Adirondacks’ website lists of census areas that comprise each of these geographic regions used in this report.

Throughout this report, we employed descriptive statistics. When aggregating multiple counties or multiple towns in a region of interest, we generally computed a population-weighted mean. Thus, we obtained the same result as would be obtained if the region of interest were a single county or town. The exceptions all involve medians. The U.S. Census median age data were only available for 1970 and 2020. The mean of a set of medians is not the same as the actual median of the aggregated underlying data. For median age, we successfully solved this statistical challenge by summing age range populations and then linearly interpolated within these narrow (five-year or narrower) age ranges to obtain median age. We verified that our median ages computed by this population age-range interpolation method agreed with the available U.S. Census median age data. For median household income, given the available U.S. Census data, we computed the weighted mean of all of the median household incomes for all the counties or towns in the region of interest as our best estimate of median household income for that entire region.

This study largely uses two time frames for analysis; 1970 to 2020 and 2020. For our trend analysis, we computed the change from 1970 to 2020. In some cases, e.g., percent poverty, it was most meaningful to express the change as an arithmetic change in percent poverty (change in percentage points). In other cases, e.g., per capita income, it was most meaningful to express the change as a ratio of the 2020 income to the 1970 income. The individual census area comparisons were later used in map creation as described below.

For our 2020 “snapshot” analyses, we compared the mean of the Park towns with all of the aforementioned regions. In these analyses, we examined only the 2020 data. We compared all New York towns or U.S. counties in the various regions with the mean of the Park towns. We also noted the total number of towns or counties that fared less well than the mean of Park Towns. We reported the total number of those towns or counties, and their total population. No statistical computation was involved other than computing the means. The individual census area comparisons were later used in map creation as described below.

Using U.S. Census data, we analyzed the following economic indicators: median household income, per capita income, poverty rate, employment rate, and self-employment rate. We selected these indicators because sampled data were available at the local (town, city, borough) level in New York and the county (parish, etc.) level across the U.S. Median household income, per capita income, and poverty rates are also useful in assessing economic performance over time, though incomes need to be adjusted for inflation. Multiple unemployment measures exist. For this reason, we avoided unemployment and instead used ratios of employment to population.

We also included an analysis of total real property taxable assessed value of all towns across New York from 1982 to 2020. The total taxable assessed value is the total assessed value of land and improvements (e.g., buildings) in a town, city or borough. In these 38 years, all towns, etc., across New York saw a growth in their total assessed value, but there were significant variations in the rate of this growth. The relative changes in assessed value are useful for analyzing the experience of Adirondack communities compared to other communities across New York.

With respect to population, we analyzed four indicators from 1970 to 2020: population growth, age group recruitment/loss, median age, and ratio of children to adults of childbearing age. These indicators have been fundamental to the position of those who argue that conservation in the Adirondack Park has led to unusual population loss.

To better understand the population trends, we analyzed the experiences of individual age groups (or age cohorts). What was novel was our calculation of population recruitment or loss of the Park Towns of various age groups compared to other regions. This type of age group analysis has not been done before in the Adirondacks.