Recently I wrote about my presentation at a U.S. Election Assistance Commission discussion on improving their survey and data collection efforts in 2006 and beyond. During that discussion, Thad and I provided copies of a draft report we have been working on, “Improving the Election Day Survey”

Today we made the final version of this report available on the VTP website. In this report, we discuss a variety of issues we have observed as we have tried to use the EAC election day survey data in some of our research.

Most importantly, we discuss twelve recommendations for improvement of future election day survey efforts:

1. Define the terms and concepts to promote uniformity across jurisdictions.
2. Refine and clarify the survey instrument.
3. Seek additional important information.
4. Reduce the cost and complexity of the survey instrument.
5. Improve the technology of reporting.
6. Insure interoperability across datasets.
7. Automate data collection where possible.
8. Improve compliance.
9. Improve data distribution.
10. Provide data at precinct level.
11. Methodological development of tools for fixing and analyzing future datasets.
12. Build collaborations.

For researchers, we also have made available at http://vote.caltech.edu/data.htm five versions of the EAC election day survey data. As some who have tried to use the EAC data have noted (and as we noted when the data were originally released), the data are only available in state-by-state files. So our diligent VTP research team pulled the state files together into one large dataset, which is available at the above website in Excel, comma-delimited, STATA, SPSS, and SAS formats. We’ve made no attempt to check the data, other than some simple tests that attempted to verify that the appending procedures across the state files appear to have worked as we hoped. If users have questions with the data, let us know. If users do anything to build additional value to these datasets, we will let our readers know of those developments.