Responses to Questions - 2014

Response to Terms of Clearance
1. While OMB acknowledges the improvement in the quality and completeness of the survey
data that EAC collected in 2008, OMB remains concerned about data quality.
For the 2012 iteration of the survey, EAC developed a electronic data
entry template used by the States that contained over 60 checks for data
inconsistencies and errors. Additionally, upon receiving the data from the
States, an additional 140 rules for consistency and errors were applied at
the jurisdictional level (e.g., counties for most States, townships or citywards for some States), and at the aggregate State level (i.e., aggregated
across all jurisdictions to arrive at State totals). This is also noted in the
EAC PRA submission, in Justification, Part A, Question 3.
The improved data entry template and the additional checks for
inconsistencies and errors resulted in an increase in the response rate at the
jurisdictional level of between 5% to 10% as compared to the 2010
responses for many questionnaire items (e.g., from a response rate of
86.4% in 2010 to 92.5% in 2012).
In 2014, EAC will use the same contractor and many of the same
contractor staff as were used for the 2010 and 2012 surveys. This results in
a continuity of technical assistance that the contractor provides; the
rapport and professional relationships established in past surveys will
yield benefits in data quality for 2014.
2. Since EAC is not employing statistical methods to generate survey results that are
representative of the surveyed population, the 2010 UOCAVA Report to Congress should
describe the limits of the survey data and include conspicuous and appropriate caveats about
drawing conclusions from the data.
This information collection is a census data collection effort. That is, EAC
asks every State, four territories, and the District of Columbia to respond
to the questionnaire, and supply data for every jurisdiction in that State,
territory, and the District. There is no sampling involved and, thus, no
statistics that require generalization. EAC obtains data from the entire
population of voting jurisdictions.
EAC compiles the data provided from the States for use in its reports
(pursuant to HAVA Section 703, which relates directly to UOCAVA
data). To provide a context EAC notes its census approach in all of its
reports, along with caveats about missing data and cautionary notes
regarding attempts to draw conclusions based on incomplete data sets.

1

Further, EAC dedicates a portion of its reports (and an online FAQ sheet)
to informing readers about the States’ data and issues related to States’
non-response, demonstrating EAC’s responsibility to inform all readers of
States’ data-related issues. To provide context for the numbers appearing
in the reports, EAC performs some descriptive analysis. However, EAC
makes the raw data publicly available so readers may perform their own
statistical analyses. EAC will continue to include the previously noted
caveats in its 2014 report to Congress and on its website. (This is also
noted in the EAC PRA submission, in the Justification section, Part B,
Question 1.)
Despite the fact that this data collection is a census of all jurisdictions,
there are missing data. Data are missing primarily for three reasons: (1)
the information sought is "not applicable" because the State does not
possess that feature in its election process (such as election-day or sameday registration and voting); (2) the data may be "not available" because a
State's recordkeeping systems do not automatically capture a particular
datum and collating the data manually would be too costly; (3) missing
due to data entry error on the part of the State.
For 2014, EAC will conduct a "per-item response-rate analysis" in order to
further refine and address the issue of missing values.
3. Specifically, EAC should
3.1. (1) implement basic consistency checks (rather than publish negative numbers in tables)
and calculate percentages that have numerators and denominators reflecting the same
responding jurisdictions
As noted above, EAC did this for the 2012 data collection and will expand
on this capability for the 2014 data collection. The 2014 system will have
a refined set of over 200 rules for internal consistency, applied to
jurisdictions and State aggregate totals, and an additional set of rules to be
applied to the FVAP questions that have been incorporated into the
instrument. Furthermore, EAC will examine data obtained on the 2014
elections with those obtained in the 2010 elections (the previous mid-term
election cycle) as a way to catch outliers and more granular errors.
Additionally for 2014, EAC will refine the definition and meaning of (a)
data not available; (b) data not applicable; and (c) missing data.
3.2. (2) acknowledge the limitations of each conclusion in the report that cannot be
supported by robust data. EAC should also identify State-reported figures (by, for
example, stating that "States reported..."), distinguishing them from EAC conclusions.

2

All three reports in the EAC survey report series have described the data contained in the tables.
All statements made in the reports are based on facts found in the data, aggregated at the State
level. EAC does not draw any inferences or conclusions from the data; all data represent Statereported figures. Any further analysis and commentary is left up to the discretion of stakeholders
and researchers to conduct and report on.

4. OMB requests that EAC share a draft of the report prior to its submission to Congress.
The States must submit their initial data for the quantitative portion of the survey by
February 1. EAC runs consistency checks on the data, notifies the States of any identified
issues and the States must return revised data no later than March 1. In 2012, the dataset
contained approximately 7.3 million data points.
For 2014, EAC is combining the NVRA, the UOCAVA, and the EAVS reports into one
consolidated volume, to be delivered to Congress no later than June 30, 2015.
Additionally, EAC will be transmitting the data related to the FVAP questions
(incorporated into the EAVS for 2014) to FVAP on March 2, March 16, and April 1,
2015, so that FVAP can fulfill its reporting requirements.
To deliver the final report to Congress on June 30, 2015, EAC must submit a final
draft submitted to the Government Printing Office (GPO) for design, typesetting, and 508
compliance, by May 15, 2015. EAC will provide to OMB a courtesy copy of that final
draft submitted to the GPO at that time.
5. In addition, OMB requests that EAC coordinate with DoD to identify overlap between EAC's
Election Day Survey and DoD's "Post-Election Survey of Overseas and Post-Election Survey
of Local Election Officials" to avoid duplicative efforts.
This has been done for 2014, (as described in the PRA Justification Part A), and is reflect in the
revised instrument that was part of the OMB submission. Substantial thought and discussion
occurred in order to add the FVAP questions to the EAVS, with EAC and DOD collaborating to
achieve this integration. EAC will use those FVAP questions that are similar to the EAVS items
as additional checks on data quality and consistency.

3