While reviewing the 2008 American Community Survey (ACS), we determined there were an unreasonable number of cases with persons in households outside of Alaska who reported living in Alaska one year ago in the data collected through computer-assisted telephone and personal interviews. The electronic equipment for these operations use a drop down list for state of residence arranged in alphabetical order by 2-letter state abbreviations; Alaska is the first entry. We implemented several fixes as we continued to research the cause of error.

2008-2010: Data Correction

We imputed responses for cases using affected collection modes where:

• Respondents currently lived outside Alaska;
• Did not report a mobility status (i.e. mover or nonmover);
• Reported Alaska as state of residence one year ago; and
• Did not report a county of residence one year ago.

2011-2014: Software Enhancement

Initially, we thought the problem involved the accidental selection of Alaska from the drop down state list. We implemented an edit that required field representatives to verify cases where people who reported a move from Alaska and did not currently live in Alaska. The edit applied to both the residence one year ago and place of birth questions.

2015: Additional Software Enhancement and Data Correction

After further review, we found that when a field representative selected a state from the drop down list and later changed it, the field for state of residence one year ago retained the original state name. Most instances of this system error affected data for Alaska. We have implemented fixes that went into effect in July 2015 and January 2016. We also corrected data for the first half of 2015. This was done by blanking all cases outside Alaska where state of residence one year ago was Alaska and we collected no other information for residence one year ago in the affected data collection operations.

With these fixes, the 2015 ACS showed an estimated 40,563 moved from Alaska to a different state for the population 1 year and over compared to 92,637 in 2014.