The average job tenure is about 4 years in the United States, but varies by age. The Canadian WES ( Workplace and Employment Survey) in their 2001/2002 panel found that 24% of workers made some change in one year: 9% different employer 9% not a paid worker anymore 6% same employer, but different job But, I am not sure how much work patterns in Canada reflect patterns in the U.S. Also, "job tenure" is NOT quite the same question as finding how often people change their workplace LOCATION. Because people move their residence, change their participation in the labor force, change their workplace location and change their means of transportation to work, the risk of identifying an individual in aggregated CTPP tables (for Counties and Places with 20,000 or more) should be reduced after accumulation over multiple years, even for means of transportation that have small shares. I am trying to quantify these changes over time. We don't have a transportation/travel behavior panel study that measures this directly! A longitudinal survey (or panel) is the best way to capture dynamics, such as changes in residence location, changes in employment status and changes in workplace location. These kinds of surveys ask questions of the SAME population over time, so the change is measured directly, rather than inferred from cross-sectional samples. Mr. Gottschalck at the Census Bureau has provided me with information on how the Survey of Income and Program Participation (SIPP) could be used to analyze SOME of these changes, but it is not straightforward and would take some effort! Also, one could have the SAME employer (in the SIPP) but change their work location, for example, to work at one CVS drugstore or Bank of American and switch to another branch, so any estimates from SIPP would be somewhat lower than "truth". Have you worked with the SIPP, or do you have another data source that might provide some light on this question? Do you have a graduate student who might be interested in pursuing this question? Even if my main interest lies in analyzing risk for CTPP tabulations, it is clearly relevant to travel demand management issues, as the best time to influence travel mode to work might be when the workplace or residence location changes. Thanks! Elaine Murakami FHWA Office of Planning 206-220-4460 -----Original Message----- From: alfred.o.gottschalck(a)census.gov [mailto:alfred.o.gottschalck@census.gov] Sent: Wednesday, April 23, 2008 7:48 AM To: Murakami, Elaine Cc: thomas.j.palumbo(a)census.gov Subject: RE: Labor Force turnover from SIPP Hello Elaine -- OK, I now fully understand your question. You should be able to use SIPP to construct a turnover rate pertaining to the length of the panel. You would have to link together the individual waves of the panel (nine in total), and then create monthly variables for the entire panel. The most recent complete panel would be the 2001 panel, which is three years in length, so you would have 36 monthly variables (e.g., AGE01, AGE02, ..., AGE36). The 2004 panel is still ongoing and not scheduled to be complete until the latter half of 2008. You would have to do some programming to produce such estimates, essentially calculating a job change when a job ID changes. The job ID you would have to create, and you would define a change as you see fit. SIPP collects information for up to two jobs, and I would suggest that you simply limit your analysis to job 1 for the time being until you have your code set. You should have sufficient sample size for each of your age groups at the national level. The SIPP data dictionary is available at: http://www.bls.census.gov/sipp_ftp.html#sipp . The SIPP homepage can be accessed at: http://www.sipp.census.gov/sipp/ . Data can be dowloaded via FTP from the SIPP home page, as well as technical documentation on how to use the SIPP files. Below are some variable data dictionary entries that would get you started in identifying job "changes" as you specified in your prior email: D EENO1 2 883 T JB: Across-wave employer index/number Unique job number that will remain the same from wave to wave. NOTE: See entry for ACAROVR1 below. U All persons 15+ at end of reference period who had a job during the reference period. EPOPSTAT = 1 and EPDJBTHN = 1 and (EJOBCNTR > 0 or ECFLAG =1) V 01:99 .Job ID V -1 .Not in Universe D ESTLEMP1 2 885 T JB: Still working for this employer Is ... employed by this employer now? U All persons 15+ at end of reference period who had a job during the reference period and who were not contingent workers. EPOPSTAT = 1 and EPDJBTHN = 1 and EJOBCNTR > 0 and ECFLAG not equal to 1 V -1 .Not in Universe V 1 .Yes V 2 .No D TSJDATE1 8 888 T JB: Starting date of job When did ... start this job? Year digits 1-4 Range 1934:2008 Month digits 5-6 Range 01:12 Day digits 7-8 Range 01:31 NOTE: See entry for ACAROVR1 below. U All persons 15+ at end of reference period who had a job during the reference period but were not contingent workers. EPOPSTAT = 1 and EPDJBTHN = 1 and EJOBCNTR > 0 and ECFLAG not equal to 1 V 19340101:20080131 .Date V -1 .Not in Universe D TEJDATE1 8 897 T JB: Ending date of job When did this employment end? Year digits 1-4 Range 2003:2008 Month digits 5-6 Range 01:12 Day digits 7-8 Range 01:31 U All persons 15+ at end of reference period who had a job during the reference period, but were not contingent workers, and whose job ended during the reference period. EPOPSTAT = 1 and EPDJBTHN = 1 and EJOBCNTR > 0 and ECFLAG not equal to 1 and ESTLEMP1 = 2 V 20031001:20080131 .Date V -1 .Not in Universe If you have any other questions, please feel free to contact me. Sincerely, Alfred Gottschalck Alfred Gottschalck, Ph.D. Economist HHES U.S. Census Bureau ------------------------------------ Dear Mr. Gottschalck- What I am looking for is the PERCENT of workers WHO CHANGE THEIR JOB LOCATION once (or more) within a 3 year period, preferably by AGE (under age 30, 30 to 59, 60 and over). For simplicity, we can assume that people who change their employer simultaneously change their workplace location. Also, people who leave the labor force in the 3 year period would also be counted in the proportion of changers. Even better would be the PERCENT of workers WHO CHANGE EITHER THEIR RESIDENCE LOCATION OR THEIR WORKPLACE LOCATION in a 3 year period. But I don't want to be greedy. I found your paper http://www.sipp.census.gov/sipp/p70s/p70-96.pdf so I am hoping that you can help me out! I got some data from the Canadian WES, but would prefer to use U.S. data, even if it is older. Thanks in advance. Elaine Murakami FHWA Office of Planning 206-220-4460