The Chart That Made My Stomach Drop

New data says the US experienced the most rapid democratic backsliding in its history in 2025.

As a data scientist and perpetually anxious person, I find that one of the few things that calm me down is a well-researched, rigorous, and thorough dataset on a topic that is weighing on me. From questions like why so many of us feel so miserable (and what we can do about it), to what’s going on with American public opinion, to how much dog poop is actually on New York City streets, if I can find a researcher or team that has put careful thought into turning something I care about into data, then they are my new best friend(s).

This means I have been waiting with bated breath for the 2026 Democracy Report from one of my favorite democracy data sources, the Varieties of Democracy, or V-Dem, Institute. The most recent of its annual reports was released in March of this year and includes their data through Dec. 31, 2025. I couldn’t wait to dig into the trends, and am excited to share them with you. (Well, as excited as one can be about something terribly depressing.)

But first:

Why bother turning “democracy” into data?

There are two good reasons to be skeptical of such an exercise. Isn’t it already painfully obvious that democracy in the US, and probably globally, is going terribly? Why waste a bunch of resources collecting data on something we already know?

Here are two reasons. First, even if we think we know something, it’s helpful to find out for sure. For example, it turns out that a lot of things we assumed were true, such as that misinformation is widespread and Americans are ideologically polarized, are not quite correct once we inspect the data (misinformation is actually quite concentrated, and our polarization is much more about disliking the other side than about disagreeing with them).

Second, even if we find out that trends are exactly as we suspected, it’s useful to confirm this because now we can more confidently develop solutions. You can think of this like going to the doctor when you’re sick. A diagnosis often offers a path toward treatment in a way that “Eh, I feel bad!” does not.

Okay, so it might be worth measuring things even if they seem obvious. But also:

How could someone objectively measure something as complex as “democracy”?

Another fantastic question!

I caution against using the word “objective” when we speak about data. There is certainly data that is reliable, repeatable, and based on countable things. But all data is subjective to some extent.

Yes, we could measure my height and weight with different instruments in different situations, and they’re likely to be about the same every time. But wanting to know someone’s height and weight in the first place is a relatively subjective choice in many cases. There are plenty of other ways we could categorize people: by volume, by how high they can jump, by how fast they read, or by how quickly their hair grows. The very existence of data in the first place is the result of a subjective decision that this thing, person, trait, or quality is worth counting.

But even if nothing is completely objective, there’s surely a difference between measuring how many cars are in a parking lot and measuring how happy, just, safe, or democratic a society is. Right?

Right! It’s absolutely not obvious how to turn something unwieldy and abstract into a number. This brings us to my favorite part of working with data: the art and science of measurement. In order to turn something like “democracy” into a number, we need to follow three steps:

1. Conceptualize: What do we mean by “democracy”? Is it about elections? Freedom? Civil liberties? An independent press? All of the above? Something else?

2. Operationalize: Suppose we choose a concept, say, elections. How do we turn that into a number? Well, we could count up how many elections a country has had. Or how many it has had over some time period.

3. Validate: Did our measure capture what we cared about? To answer this, we look at our results and see if they seem like they map to reality. For example, most countries, including authoritarian ones, have held elections at some point. So counting elections on its own is not enough to “rule in” what we recognize as democracies and “rule out” what we recognize as non-democracies. More criteria are needed to count elections as democratic: they have to be repeated, they have to be competitive, and the candidate who wins has to take office.

That said, we might be interested not just in the rules of the country, but in its culture and adherence to democratic norms. It might have free and fair and repeatable and all kinds of lovely elections, but if no one votes in them, that should be a red flag.

This tension between focusing on structural and institutional features of a country’s government and the actual norms, cultures, and behaviors of its people is at the heart of why there are many different ways to measure democracy. Some researchers argue that the institutional (or “minimalist” or “procedural”) approach is sufficient for teasing out enough distinctions between different regime types. It also has the advantage of relying on data that is close to readily countable instead of trying to capture vibes.

Other researchers argue that if you’re just looking at policies and institutions, you’re missing a ton of important information. Proponents of this more “substantive” approach say that you need to get into the texture, nuance, and details of what’s happening on the ground.

There’s no right answer about which approach is best (though proponents of each would disagree with me). Instead, I recommend consulting multiple datasets anytime we’re trying to understand something. Each will teach us something different, and the places where they depart from one another might be just as informative as their agreements. In that spirit, I encourage you to explore a variety of democracy datasets out there, including the Freedom House Index, the Economist Intelligence Unit’s Democracy Index, and my personal favorite, as I think it balances the institutional and the cultural well, the Polity project (sadly, it lost funding several years ago).

Completely counter to that spirit, I’m now going to show you some high-level findings from V-Dem, which is a dataset that many consider to be the most detailed, thorough, and exhaustive. Rather than land on one single democracy number for a country, it focuses on five different ways a country might be democratic:

1. Electoral democracy (having robust elections like the ones we discussed above)

2. Liberal democracy (where people enjoy civil and political liberties and freedoms)

3. Deliberative democracy (conversations about politics in the country are productive and respectful)

4. Participatory democracy (people actually run for office, vote, and take part in civil society)

5. Egalitarian democracy (everyone benefits from democracy, including minority groups and marginalized communities)

To get numbers for these five indices, V-Dem collects as much “hard” data as possible (economic data, wealth inequality data, number of elections, voter turnout rates, etc.) and combines it with the results from extensive surveys completed by several experts on each country for each year. These survey results are then weighted according to a complicated algorithm that accounts for the expert’s confidence level in the answers and the extent to which those answers agree with the answers of other experts. Some 200 resulting indicators are then folded into the five indices described above.

It’s a truly heroic effort that involves thousands of people around the world, and the result is a dataset that is incredibly useful. It’s also publicly available and includes a nifty online graphing tool that you should for sure check out. (I promise no one at V-Dem has paid me to say this!)

With that, let’s see what the heck has been going on. Here’s the US since 1900.

The US has experienced a rapid and large decrease in democracy scores along all five V-Dem indicators since 2024

Scores for the five core V-Dem indices for the US, annually since 1900. A score of 0 is the least democratic and a score of 1 is the most democratic.

Source: V-Dem

A few things stand out, especially the breathtaking plummet from 2024 to 2025. I was prepared for a decline, but I was shocked when I saw how steep it was.

The last time our scores in most categories were this low was in the late 1960s and early 1970s. But the “deliberation” score, which started moving considerably in 2015 and 2016, is closer to what it was in the 1920s.

Our highest ever democracy score was a 0.91 for electoral democracy, which briefly put us in the company of countries like Denmark and Norway. And while our scores went as low as 0.18 for egalitarian democracy in the early 1900s, for the better part of the last few decades we’ve hovered pretty reliably in the 0.6 to 0.8 range — until now. In fact, according to V-Dem, in 2025 the US went through the most rapid democratic backsliding in our history, and one of the most rapid ever in the history of the entire world. (USA! USA! USA!)

Speaking of the entire world, let’s take a look at global averages over the same time period.

We are globally going through what many researchers call the “third wave” of autocratization since 1900

Mean scores for the five core V-Dem indices across all countries and politically independent entities, annually since 1900. The same 0–1 scale applies.

Source: V-Dem

While the US is outpacing most of the rest of the world in our democratic backsliding, we are not alone. Globally, we’re seeing a decline in democracy scores on all five indices, with many gains made since the early 2000s effectively neutralized over the past few years. To appreciate the magnitude of this change, let’s zoom in on the last few decades, this time with an eye to the global population as a whole.

Many more people are living in autocracies or autocratizing countries than did 20 years ago

Percentage of the world’s population who live in the following types of countries as of December 31, 2025

Source: V-Dem

To me, the numbers in the above bar chart are truly shocking. In much of the 1990s and early 2000s, many of us were quite hopeful that democracy would spread around the world and an era of international peace, prosperity, and freedom was upon us. I don’t need to tell you that it hasn’t turned out that way, but consider the numbers. We went from just 9% of the world’s population living in countries that were becoming more autocratic to 41% of people, more than a fourfold increase. Today, 74% of people live in full-on autocracies, and only 5% live in countries that are becoming more democratic.

The last cluster of bars, “liberal democracies,” refers to countries that V-Dem considers full expressions of democracy (again they’re our golden children Denmark, Norway, and the like). Depressingly, only 7% of the entire world’s population currently lives in countries that meet that criterion.

Why?!

There’s no easy answer to why we’re seeing these trends. We can all look around and name things that seem to be going wrong — declining press freedom, more state violence against citizens, expansion of executive powers, fewer checks and balances — but why are those things happening?

It’s way too soon to say for sure, and it might be a question as haunting for historians as why we saw World War I (which is to say, we could write 10 million books with different theories in them). But there is research underway to better understand why this is happening and whether we can reverse it (in the spirit of multiple datasets, I should also note that the EIU data suggests the global decline is starting to stabilize, though they note a fair amount of turbulence may yet be ahead).

One source of ideas for why we might be going through this “third wave” that I find compelling comes from a writeup of an expert summit at Harvard’s Ash Center. It suggests a few high-level forces at play. These include suggestions that pro-democracy actors and groups can’t seem to agree with one another enough to build a coalition to counter authoritarianism, that current models of protest and resistance are not enough, that authoritarian networks and legitimacy are expanding, and that pro-democracy movements have lost control of narrative and information.

We may not know the real story for a long time, or at least we may not have a full understanding of all the forces afoot. But I hope that having slightly more concrete information about what we’re dealing with will at least help us craft more effective and hopeful policies and actions to turn things around.