Analysis

Descriptive

The data alone that this study will yield is of great value. There are numerous uncertainties and myths in the voting community – often-cited statistics with no origin, generally held misconceptions – which hard numbers could debunk or confirm. Two favorites are briefly described below.

Machine Efficiency

One very interesting outcome will be how many voters, and at what pace, a machine services in a day. In previous studies, researchers have estimated the number of voters a machine services in a typical voting day using just the first two parameters: ballot length (in minutes) and hours polling station operation. The best current estimate is that one DRE can serve 277 voters on Election Day; this is based on a 3.5-minute ballot and a 15-hour election day.   In contrast to the DRE, an optical scanner can process about 360 ballots an hour and can therefore service 5400 ballots in the same 15-hour day.  However, there is probably a large discrepancy between the maximum efficiency of a voting system and its efficiency in regular use at polling stations.

Poll worker Demography

The median age of poll workers nationwide is thought to be 72, though it is unclear where this number originated. In surveying the poll workers at our 32 polling stations, we will be able to provide a mean poll worker age as well as the range and distribution of their ages. Though this may not be perfectly representative, it will be transparent and accurate.

Sample Outcomes

• Counties can assess their ballot length before an election and use our CA efficiency estimates as a guideline when allocating machines.
• Counties can begin to examine the age of their election workers, how they compare to other counties, and what effects these factors might have on the overall operation of the election.

Statistical

Collecting data on mean waiting time for potential voters at a polling station allows us to build a regression model to explore variables correlated with increased wait times.  From this analysis, we can derive how wait time varies with a host of variables related to the precinct, voting process and technology.  The coefficients of the exogenous variables in the regression will suggest what aspects may cause long wait times and indicate the magnitude of this relationship. As we plan initially to include a wide range of variables (not all of which will significantly affect mean wait time), we will generate a number of models to ensure that the standard error of the regression is inefficient.

Sample Outcomes

• Seeing our decisive claims about the relative import of a vector of variables on wait time, counties can adjust variables within their control. For instance, if poll worker competence is strongly associated with shorter lines, counties may explore additional or revised training.
• Seeing how different types of voting technology contribute to overall time to cast a ballot, counties might consider switching to more efficient technology.