Another terrible chart

The tell is in the axis

The above graph has been making the rounds on social media. It purports to show the vote totals in the last four several elections, and draws attention to the 2020 surge in Democratic votes. The underlying claim is that the large turnout for Joe Biden in 2020 must be somehow fake or otherwise inflated. This is the most egregious chart I’ve seen since I covered this atrocity from the Financial Times. One of those charts that makes me sigh and then wish there was such thing as crimes against data that could be prosecuted in The Hague.

Let’s dive in.

First, this graph should immediately be regarded with suspicion because of the left-hand axis. It starts at 50 million votes. What an odd decision. Generally bar charts begin at zero, especially when analyzing trends, to give the reader a sense of scale. By starting the graph at 50 million rather than zero, it makes the 2020 election bump seem much larger than it actually was. Properly rescaling the graph results in the following. Put into the proper context of tens of millions of votes, the 2020 democratic bump doesn’t look nearly as large:

Just to show how sharp the contrast is with the incorrect scale on the left axis, compare the above graph that starts at 0 with one the one below that starts at 59 million votes:

Scale matters. If you want to make a data point look like an outlier, arbitrarily cutting off most of the data is helpful.

That alone is a venial sin in the world of data presentation. There is, however, an even worse offense here. The original graph started making the rounds on Twitter the week of the election before all the votes were tabulated. So not only is the scale set to make the 2020 bump appear bigger than it was, it also artificially deflates the 2024 count. I suppose that starting the axis at 60 million votes could be plausibly defended, but producing a graph to show that Democrat vote tallies have been almost identical in the 2012, 2016, and 2024 elections, when in fact the 2024 election is significantly higher than the first two, is data malpractice. To be fair, there is something to be said about a country that can't tabulate votes within a few days of them being cast. What is California doing?! Still, the lack of complete data doesn’t mean one should proceed with incomplete information, it means waiting until the data is available. Clearly, whoever made the first graph did not have truth in mind. Today, with over 98% percent of the 2024 votes counted, a more accurate graph can be shown. As suspected, including almost all the votes in the 2024 election makes the 2020 bump look even smaller:

Now you can see that while Joe Biden did do very well in 2020, earning more votes than any presidential candidate before or since, it isn’t much of an outlier. For final comparison, look at the original, misleading graph and the updated, accurate graph side-by-side:

Same information, but the graph on the right has the correct scale and has been updated to fully reflect the vote count from the 2024 election (at least as of this writing, with over 98 percent of votes counted). Looks a bit different, doesn’t it?

Still, one could argue that there’s something funky about the 2020 vote count. How was it possible that Joe Biden got so many votes? It is worth noting that 2020, relative to 2016, did have one of the biggest increases in turnouts in US history. That said, such increases have happened before. Comparing the 2020 election to the 2016 election, voter turnout went from 136,787,187 to 158,427,986, or an increase in turnout percentage of the voting-eligible population from 59.2 percent to 65.8 percent. That’s a big shift - the largest turnout recorded in living memory. However, it is almost identical to the increase from 1988 to 1992, when turnout went from 91,586,725 to 104,600,366, or an increase in turnout percentage of the voting-eligible population from 52.8 percent to 58.2 percent. Thus, the increase in the 2020 turnout has recent historical precedent. As an interesting note, after the 1992 election surge, turnout in 1996 dropped not only below the previous election, but to the lowest turnout since World War II. Turnout wouldn’t exceed the 1992 election until 2004. So single-election bumps have happened before.

On top of that, it’s important to note where turnout changed. If there was widespread vote rigging in favor of Joe Biden in 2020, he should be leading Kamala Harris in the vote count in swing states. That isn’t happening. Harris received the same or more votes than Biden in North Carolina, Georgia, Wisconsin, Pennsylvania, and Nevada. She’s down about 75,000 votes in Michigan and 125,000 votes in Arizona. Instead, it’s in the reliable blue and red states that Democrats underperformed. Clearly, if you wanted to steal an election and pad the Democratic vote, you wouldn’t inflate vote counts in states like California that are guaranteed to go blue, let alone those guaranteed to go red. The change in the swing states matters - they delivered Trump the election. The collapse of Harris's support in the rest of the country, however, delivered Trump the popular vote. Millions of Americans who live in one-party states voted for Biden in 2020 and decided to stay home in 2024.

Plus, and I know many of us have tried to block this out of our memories, but November of 2020 was a bit of an odd time for Americans. It makes total sense that voter turnout was high - there wasn’t anything else to do! Not a lot of Americans were abroad or on vacation or really doing much of anything on November 3rd, 2020. Especially combined with the increase of voting by mail, lots of people were going to vote that year that otherwise wouldn’t have. If turnout could increase by 10 percent between the 1988 and 1992 elections, when the notable change was the highest profile third-party candidate in decades, then a worldwide pandemic that turned American life upside down could certainly increase turnout by 11 percent.

I know this isn’t likely to convince anyone. Those who are convinced that millions of votes were fabricated for Joe Biden in 2020 are not going to be swayed. I’ve even seen some Democrats pointing to the same terrible graph as evidence that some of Harris's votes in 2024 aren’t being counted. It’s all nonsense. Biden won in 2020. Trump won in 2024. But hopefully, this will provide evidence-based arguments for those interested in the truth.