America has more then 3,200 counties and around that number of County Elections Boards that mostly operate independently based on state and to a lesser extent federal law. Each County Elections Board has dozens if not hundreds of employees each with a little bit different way of inputting data often from poorly handwritten forms. Itβs a recipe for mistakes and the data is often disorderly and hard to access.
The rather disorganized way America runs elections makes any widespread fraud difficult. But on the other hand, it makes mistakes much easier and localized fraud much harder to detect. Voter rolls are riddled with errors, most of them clerk typos and mistakes due to confusion on the part of voters, but also incompatible data types, software conversion errors, and a million other bugs creep in. At least in New York, the State Board does not check or correct addresses, nor do counties. If somebody writes a bad address on their Voter Application or somebody inputs it incorrectly, it is recorded as it inputed.
Having bad addresses in the State Voter file leads to all kinds of problems. For one, it makes it harder for candidates and elected officials to communicate with their constituents. Indeed, a major initiative of my office is to clean the voter file to assist Assembly-members with their communications. Itβs a very labor intensive job as you can only automate so much of address corrections. It also means that Board of Elections officials canβt provide notice of polling place locations and changes, mail ballots to peopleβs homes, and mail cards to confirm that infrequent voters are still alive and still reside at the recorded address.
The state voter file is loaded with mistakes too, like incorrect Town, Ward, ED, or political geography for individual voters. While certainly such data is fixed over time, itβs surprising how many mistakes one can find in the voter file. Some of that has to do with all the changes with redistricting, but some of it also has to do with typos in inputting and the non-address corrected addresses making it difficult to properly check political geography with geocoding. While such mistakes may only represent a half a percent statewide, that is still thousands of voters with mistakes in their file.
As voter records are not linked to vital records or social security numbers β itβs actually difficult to remove dead people from the voter rolls unless people return correspondence from Board of Elections saying the person no longer resides at the address, at which point they are moved to the inactive list. Itβs not possible to remove deceased voters from the rolls based on names, because there is a lot of people named Alex M. Rodriguez and M.D. Mohammad on the rolls. Even when you add date of birth, with such common names there are legitimate duplicates representing different people.
In addition, while the state attempts to track voters when they change addresses from county to county by creating a state voter number, the matches are built on full name, date of birth and previous address reported by the voter, so you have several cases where the voter still appears on the rolls in multiple counties or there are duplicates of the voter number. Itβs actually pretty hard to track 13 million New Yorkers as they move from home to home. The state board does itβs best, and they also run names through the federated Voter Information system to try to remove people who have moved out of state, but if you are matching on full name, date of birth, and known addresses, you can still accidentally flag duplicates where they donβt exist and miss others. And even social security numbers and driver IDs are not a perfect system, as they are duplicates and mistakes in that system too.
Election District Maps, while in most counties are digitized, and for the first time are available on gis.ny.gov after a recent law change made it possible for State Information and Technical Services department to aggregate the data are still often inconsistent and not always kept up to date. Each county has different fields in their file, a different way to record the town, ward, and election district. NYS Information and Technology Services (ITS) aggregates that data together into one file, but it is not at all consistent from county to county. The codes in the Shapefile from the counties and ITS does not match up with the official enrollment data Election Districts put out by the State Board of Elections.
There is also no state-wide source for election-district by election district election results. Many counties now post election-district by election-district but not all do. Many do not post the data in a tidy format easily processed by a spreadsheet or database, rather posting it in PDFs or in non-standard format that not easily machine-readable without writing a custom code for nearly every county. This makes it hard to analyze the data, especially on a state-wide or national fashion. People do, but itβs a high bar with a lot of data cleaning required. Moreover, the codes used by counties to describe election districts rarely correlate with the codes used by the ITS on their Shapefiles.
A lot of the disorderly way election data is stored is not just the fault of the highly federated nature of elections in America, but also because the technology for managing big data was uncommon and required expensive servers and advanced knowledge how to operate them. There is also hesitancy by some local election officials, as they want to know who is looking at elections, so they can have an advanced notice of those who might be thinking of running for office or if an embarrassing news article about the disorderly and poorly kept nature of elections data is about to come out. That gives them time to draft a response and not be caught flat footed when a reporter calls about about discrepancies in the elections files.
There is very little evidence of widespread fraud in elections or even much in way of small-scale fraud. But the often disorderly nature of voter and elections data sows distrust in the public and leads to mistakes being made, sometimes motivated mistakes based on partisan loyalties. In that sense, disorderly data causes or at least enables fraud. As unpopular as it might be to local board of elections to be forced to a single standard from the federal or at least the state government to keep their all their voter and election data in a tidy format and well cleaned and maintained, it would go far for preventing mistakes and even fraud, and help restore public trust in elections.
To be sure, things are getting better in recent years with the enactment of state-wide NYS Voter database, state-wide absentee ballot tracking, counties posting election district and ITS compiling election district shapefiles into one, the system could be better, especially when it comes to data cleaning and making state-wide preliminary and final election results available district-by-district as soon as the polls close.
