The Push for Hand Counting: A Step Back in Responsible Election Practices

Dark blue background with blue tally marks, red scribbled question marks, light blue dollar signs and a red-framed clock that is spinning in clockwise in gif-format.

While the apparent simplicity of hand counting ballots to determine the outcome of elections may be appealing to some, there’s something important about the practice everyone should know: it simply does not work. 

Alas, there is no shortage of foolishness to be found in election policy debates today, and the cause of hand count only tabulation has recently been taken up by election conspiracy theorists in states like Arizona, California, Georgia and Nevada, to name a few. 

Although hand counting ballots might seem like a straightforward enough enterprise, the issue is that it really isn’t. It’s incredibly complicated, expensive, inefficient and — most importantly — wildly inaccurate. 

As any voter in the United States knows, the ballot you fill out is not just for one race. Typically, it’s for a multitude of races or ballot questions, dozens even. And your ballot is not going to be the same as someone’s from the next precinct over. Contrary to what its proponents argue, hand counting is not just a matter of checking which candidate received a vote and putting the ballot in that pile — which is how it’s done in other countries, like France, where there is only one election on the ballot. It’s quite a bit more complicated.

In Arizona, the Mohave County Elections Department was instructed to come up with a plan to hand count the results of the 2024 election. As part of this planning process, the department conducted a study of hand counting procedures, where a seven-member board counted 850 ballots from the 2022 election by hand. Here’s what the process for the hand count looked like:

The seven (7) member hand count board consists of one person calling (caller) out the race and candidates’ names; two people watching (watchers) making sure the ‘caller’ calls out the information correctly; two people marking (markers) the race on their separate tally sheets; and two ‘watchers’ making sure each ‘marker’ marks the race correctly. This board is made up of an equal number of people from the two major parties and/or parties not designated.

If all of that sounds like a lot of work, that’s because it is. More importantly, nobody will do that work for free. That same study estimated that the county’s additional cost from hand counting votes would be $1.14 million for the 2024 election alone. 

The Bipartisan Policy Center estimates that to conduct a hand count election at scale in a state like Michigan would require a 300% increase in the election workforce. And even single-race hand recounts, like the ones we saw in Georgia following the 2020 presidential election, cost counties hundreds of thousands of dollars to tally the votes in just that one contest.

But what about efficiency? Surely with people working together we should be able to get some, albeit expensive, results in a timely manner? Nope

No matter what conspiracy theories may abound, all of the available information points to a single conclusion: hand count-only elections just don’t work.

In the Mohave County, Arizona study, it took those seven workers three days, working eight hours a day, to count just 850 ballots, which on average equates to five ballots counted per hour per worker. There were 2,592,313 ballots cast in the state of Arizona in the 2022 election. If we want results on election night using this method, that’s fine, but we’re going to need about 170,000 people — or roughly half of the U.S. Navy — counting nonstop to pull it off. 

But let’s say they do. Let’s imagine 50% of the good men and women of the U.S. Navy pitch in to hand count all the ballots for a single state and they pull it off. At least this complicated, expensive and hilariously unrealistic enterprise will have yielded thoroughly accurate election results, right? Not even close. 

Despite the labor intensive calling, marking and verifying that these election workers engaged in over the course of three days in Mohave County, Arizona, there was still a 5.4% error rate. Other studies of hand counts in New Hampshire and Wisconsin have likewise found that hand counting is substantially less accurate than machine tabulation. 

However, conducting post-election audits by hand still has a purpose, with some very important caveats. The first caveat is that after initial election results are determined by machine tabulation and released to the public, a lack of information cannot reasonably be exploited by candidates or bad actors while a hand count is conducted. If we were to only conduct hand-counts, the information vacuum that exists while the ballots are slowly tallied would prove fertile soil for malicious individuals to sow distrust in our democratic process — as we got a small taste of during what felt like the longest week of 2020.

That leads to the second caveat: post-election audits conducted by hand don’t exist in a vacuum. Their utility comes from the ability to validate or invalidate the results of a machine tabulation. They are only useful when the initial results provide a point of reference. 

Finally, post-election audits by hand do not involve tallying all of the ballots cast, because that is not a practical exercise when done by hand. They do, however, take a random sample of ballots cast: large enough to get representative results, but small enough that it can be done by hand. 

At the end of the day, by conducting elections with hand counts only, states will have to spend more money and more time to get election results that are less accurate than if they used machine tabulation. Data confirms this, which is why it was heartening to see the officials in Mohave decide against a hand count for the 2024 election. 

There are still some people out there who believe this is a viable means for conducting elections, but they’re wrong. No matter what conspiracy theories may abound, all of the available information points to a single conclusion: hand count-only elections just don’t work. If we want accurate, secure election results, we have to work towards further modernizing our elections to do better. 

Sam Oliker-Friedland is the executive director of the Institute for Responsive Government.