I’m using the single-year ACS, and I have a question about the reference period of the employment status and income. Let’s say I use employment status (EMPSTAT) for 2002 ACS. I assume this information was collected at some point in 2001 and the respondents reported their employment status a week before the survey - right? Likewise, if I use the worked last year variable (WORKEDYR) for the same wave, 2002 ACS, does this information refer to anytime between 2000 and 2001 or 2001 and 2002?
The methodology of the ACS is such that it includes 12 monthly independent samples where data collection for each sample lasts for three months. This document from the Census Bureau documents the data collection process. The data collection begins at the beginning of the year and extends January and February of the next year. So, in your example, for the 2002 ACS, data could have been collected at some point in 2002 (January through December) or in 2003 (January or February).
The reference period for EMPSTAT is last week, so the week prior to when an individual was surveyed. As for WORKEDYR, the reference period for the ACS is the previous 12 months. So, for someone who was surveyed in November 2001 (for the 2001 ACS) this would refer to the 12 months prior, covering parts of both 2000 and 2001. For someone who was surveyed in February 2002 (again, for the 2001 ACS) this would refer to time in both 2001 and 2002.
I hope this helps.
Thanks for your response.
I’m afraid I still can’t get my head around with the “12 monthly independent samples” and the figure from the linked document. In figure shows the data collection for the 2013 ACS began November/December 2012 and lasted until June 2013, suggesting July through October 2013 is not included. Could you please clarify if this indicates data collection stopped during this period?
Yes, that figure is a little confusing. To clarify, everyone in the 2013 ACS is surveyed between January 2013 and February 2014. The important detail is when people are selected into the sample. If they are selected into the ACS sample in 2013, for example, they will be in the 2013 ACS. It can take some people up to three months to actually respond to the survey, so that is why some people who respond in January or February of the following year (2014, for example) may still be in the previous year’s sample (2013, for example). These details should clarify how to interpret Figure 7-1 in the document linked above.
Thanks for the clarification. That makes more sense.
By the way, what Michelle says was that the data could include information on the samples selected two months ahead of the survey year. But now I hear that it is actually the other way around. I’m a little confused… Would you mind defining the data collection period for the 2004 ACS for example, please?
Sorry about the confusion. For the 2004 ACS, all individuals are selected into the sample in 2004. Each individual has up to three months to respond to the questionnaire from the time of selection into the sample. So, individuals selected into the sample in November and December could respond as late as January or February of 2005.