CONNECT:
SHARE THIS:

Methodology




The 2014 "Best Cities for Successful Aging" report builds and expands on the 2012 index methodology. The two are not very different; however, comparisons should be made with caution. In this section, we focus on some of the major differences between the two, and provide a full list of indicators and their respective weights.

 

The overall composite rankings for 2014 "Best Cities for Successful Aging" are based on the same eight subcomponents that were scored in 2012: general indicators, health care; wellness; living arrangements; transportation/convenience; financial well-being; employment/education; and community engagement. Each subcomponent is scored according to multiple individual indicators. Across all eight, we used 84 individual indicators, up from 78 in the 2012 report.

In 2013 the Office of Management and Budget, which derives its population estimates using Census Bureau data, published updates to its geographic delineations of metropolitan and micropolitan areas. (The former surround urban areas of more than 50,000; the latter include urban areas of more than 10,000 but less than 50,000; both statistical areas include at least one county or equivalent entity.) While the OMB specifies that there are 381 Metropolitan Statistical Areas (MSAs), we adjusted the list of metros. Using the same list of 359 metros from the 2012 "Best Cities for Successful Aging" report, we adjusted to 352 metros to follow the latest metro definitions. This is the result of several metropolitan areas changing definition to become micropolitans, and a few aggregating into a neighboring metro. For example, the New York-Newark-Jersey City, NY-NJ-PA metropolitan area, the largest in the U.S., now includes Poughkeepsie-Newburgh-Middletown. Our 2014 index does not include the OMB's newly added metropolitan areas, due to data limitations.


Determining weights for the overall composite index

The overall weights used for each of the eight subcomponents are very close to those of the 2012 version, with some adjustments.

We used factor analysis to determine the relationships between indicators within each of the eight subcomponents to generate standardized regression coefficients. Standardized regression estimates were further used to develop weights for each indicator. In order to smooth the weights for use in this index, we averaged the newly generated weights with the weights from the 2012 index.

To create the overall composite index, we again used factor analysis to identify the indicators that contributed the most to each of the eight subcomponents. We used regression analysis with the chosen indicators to measure the relationship with the percentage of population age 65 and older. The 2014 overall weights were created by adjusting the weights used in the 2012 index in accordance to the ordinal ranking of standardized coefficients from this regression.

Data creation and resolving data issues

This index uses publicly available data at the metropolitan statistical area (MSA) level. When only county-level data were available for indicators, we aggregated the data to the MSA level. This methodology involves the creation of some new indicators, and presents some challenges that we addressed in the following ways:

Changes in indicators The 2014 "Best Cities for Successful Aging" features some new indicators; we removed the indicator Medicaid eligibility from the wellness subcomponent and transferred a few other indicators from one subcomponent to another. This implies that the weighting of each subcomponent in calculating its overall score will now be slightly different. As such, we modified the average weight, calculating from weights in the 2012 index and the original subcomponent factor analysis, to reflect the ordinal rankings of the new indicators.

Missing/inadequate data Several subcomponents did not have data available for all indicators, or only provided state-level estimates. This updated index follows a similar approach as the 2012 index by splitting up the 100 large metros, and the remaining (252) small metros. For some indicators, data was only available for the large metros, and we were not able to include the indicator for the small metros. In some cases, we developed proxy measures, such as shift shares or averages of close proximity MSAs, depending on applicability of the specific indicator.

Using state-level data Some indicators, however, use state-level estimates since data were not available at the metropolitan level. For example, state and local tax burden per capita is available at the state level, and each metro within the same state is assigned the corresponding state value. In cases where a metro is situated in more than one state, these values were aggregated accordingly.

Go to General Indicators