Making Sense of "Conflicting" Commuting Numbers

Tuesday, September 1, 2015
Author: 
Jeff Rosenthal

Recently, the U.S. Census Bureau released updates in commuting data from two different sources: Household Survey (American Community Survey or ACS) and Data from Employers (LEHD Origin-Destination Employment Statistics or LODES)1. Our colleagues at Carolina Demography have posted great tabulations and descriptions of some of the findings from the ACS, and LEAD recently posted an article on how commuting has changed since 2002 using LODES data. Since the data show numbers that some may feel tell conflicting stories, we briefly show differences between the two data sets and make suggestions for their use in comparable situations. For instance, ACS should be considered for transportation and long-term research needs, while LODES is better used for private sector labor market identification and spotting recent trends.

The Data Sets

The results differ because the information comes from different sources with different data elements: design and definitions of research. Such data elements may include:

Which One Should I Use?

It depends. You may need:

  • Additional information that only one data set provides.
  • Particular types of workers, regions, or time periods.

The following cases would suggest using the ACS:

  • For transportation needs: The ACS covers all workers and has information on the mode of transportation they use to get to work.
  • Where you need additional demographic variables: The ACS has a rich variety of information found through the American Fact Finder.
  • For long-term historical patterns: One can compare ACS data with the worker flows from the Decennial Census from 1990 and 2000.
  • For analysis of areas where there is a lot of employment not covered by Unemployment Insurance — namely military bases: The ACS results show overwhelming percentages of workers that work and live in the same county for Onslow (Camp Lejeune) and Cumberland (Fort Bragg) counties, while the LODES results show a more modest percentage of those working and living in the same county, especially for Onslow County.  

The following cases would suggest using the LODES:

  • For labor market area identification. This may occur through either:
    • Front-line workers helping clients find private sector employment with appropriate commuting flows.
    • Researchers determining commuting flows and labor sheds across the state (see Interconnected Markets section here).
  • When the focus relies on private primary workers for the needs of governmental programs or focus on the private sector. From Unemployment Insurance to the new Workforce Innovation and Opportunity Act (WIOA), there may be a need to focus on Unemployment Insurance covered employment, be it for training or staffing purposes.
  • For spotting recent trends. The LODES is better when you need the most current data for comparison. The most recent release of LODES data was for 2013, while the most recent ACS data was from 2009–2013. The LODES data for North Carolina comes out each year, and goes back to 2002.
  • For analyses within a county.  With OnTheMap, one can examine commutes at smaller geographic levels (i.e., cities, zip codes, or census tracts).

By using different data sources, we will necessarily have (at least slightly) different results. The key to any reporting or research using data is to understand the data and how it’s collected. If you’re unsure, consider consulting with a social science professional at LEAD or Carolina Demography. You’ll be glad you did!

Entry co-authored with Patrick McKemie.  This entry relies heavily on a descriptive paper written as a collaboration of the Census experts of each source that can be accessed here:  ftp://ftp2.census.gov/ces/wp/2014/CES-WP-14-38.pdf