What We Learned From a Survey Experiment About Global ‘Superpowers’

If survey researchers could choose a superpower, they might opt for telepathy – the ability to read people’s minds. But in a recent experiment, we were interested the geopolitical kind of “superpower” instead.

Pew Research Center has a long history of asking survey respondents which country they think is the world’s top economic or military power.

However, these questions may not capture all the attributes that make a country a superpower. They also ask respondents to choose from a defined list of countries that has remained consistent over time and may not reflect emerging geopolitical trends. Finally, because of their format, these questions don’t allow us to reflect the views of respondents who think multiple countries (or no countries) are global superpowers.

So, we wondered: Could we write a new question that addresses these limits and might be included on our mixed-mode, cross-national Global Attitudes Survey?

In this post, we’ll share what we learned through cognitive interviews and a split-form survey experiment that we fielded on the Center’s nationally representative American Trends Panel in March 2026.

First step: Cognitive interviews

In winter 2024, we conducted cognitive interviews with 12 U.S. adults, asking them the following question:

“Thinking about the world today, would you say it is a unipolar world, meaning one with only one superpower; a bipolar world, meaning a world with two superpowers; or a multipolar world, meaning one with three or more superpowers?”

We learned a few things from these interviews:

• Terms like “unipolar,” “bipolar” and “multipolar” confused some respondents.
• People could more readily name which countries they think are superpowers than how many there are in the world.
• Respondents correctly understood what “superpower” meant in the context of our question. But a couple incorrectly thought we were asking about supernatural abilities. This told us that our final question should include a definition of superpower to avoid similar confusion.

But how should we define superpower in our survey? As a follow-up, we asked interview participants what “superpower” means to them. Collectively, they said it’s a country that is economically and militarily dominant.

Based on these interviews, here’s the question we decided to field in a split-form survey experiment:

“Thinking about the world today, what country or countries are global superpowers, where ‘superpower’ means a country that is both militarily and economically dominant?”

Second step: Asking our new question on a real-world survey

We knew from our cognitive interviews that respondents think more about specific countries when we ask them about global superpowers, rather than the number that exist. But when developing our new survey question, we were also concerned that presenting respondents with a blank text box and letting them write in their answers might lead them to overestimate which countries are superpowers and how many there are. So, we devised two approaches for the survey experiment:

• Unfiltered approach: We asked respondents in a direct, open-ended question which countries are superpowers and included explicit “None” and “Don’t know” options. They could name up to six countries.
• Filtered approach: First, we asked respondents if they think any countries are superpowers. Then, if they said yes, we asked them to name those countries. Respondents could say “Not sure” to both questions.

---

The two ways we asked about ‘superpowers’

Experiments survey questions as they appeared in American Trends Panel Wave 190

Unfiltered approach

Filtered approach

Source: Survey of U.S. adults conducted March 23-29, 2026.

PEW RESEARCH CENTER

---

We wanted to find out whether one approach was more likely than the other to lead respondents to name a particular country. We also wanted to know whether one encouraged a more “unipolar” view of global power dynamics than the other.

What did the experiment find?

Overall, respondents answered both questions similarly.

Under both approaches, respondents most frequently named the United States as a superpower, followed by China and Russia. Similar shares also said there’s one superpower and said there are two.

However, respondents were more likely to express uncertainty when asked the filtered question than the unfiltered one (though demographic differences here were relatively consistent in both versions). In fact, 23% said they were not sure when asked the filter question, and another 5% did so when asked the follow-up question.

Related: Nonresponse rates on open-ended survey questions vary by demographic group, other factors

We also found that respondents named certain countries more frequently in the unfiltered question. For example, 44% of people who received the unfiltered question said China is a superpower, compared with 32% of those who received the filtered question.

Respondents also named a broader range of countries in the unfiltered question. In total, people named 47 different countries in the unfiltered format, compared with 30 in the filtered format.

We suspect these results are tied to the fact that respondents are more likely to say they’re not sure when asked the filtered question.

How did our new question compare with our trend questions?

We conducted this experiment on the American Trends Panel, which is also Pew Research Center’s main vehicle for U.S. surveys. One advantage of survey panels is that they allow us to track attitudes among the same people over time. This gives us another way to assess the results of our survey experiment.

Overall, 1,770 respondents who saw a version of our experimental question also answered our two longstanding questions about global economic and military power in the year prior:

• “Today, which one of the following do you think is the world’s leading economic power – the U.S., China, Japan or the countries of the European Union?”
• “Today, which one of the following do you think is the world’s leading military power – the U.S., China or Russia?”

While our trend questions are not directly comparable to our experimental questions, they are the best measures we have for how people viewed global superpowers prior to our experiment. Here are some notable takeaways:

Some Americans who saw the U.S. as the world’s leading economic and military power still name multiple superpowers when given the chance. Using our unfiltered question, 44% of these Americans see a “multipolar” world. Using the filtered question, the figure is substantially smaller (29%).

Americans who saw a “multipolar” world in 2025 responded differently to the unfiltered and filtered versions of our experimental question in 2026. Using the unfiltered approach, 57% of those who said the U.S. leads on one dimension and China on the other listed multiple superpowers. Using the filtered question, far fewer (30%) said there are multiple superpowers.

Conversely, the filtered question led to larger shares of respondents saying there is only one superpower (32% vs. 18%) – or no superpowers (17% vs. 9%) – than the unfiltered version.

While it is worth noting that views of the global order may have shifted between 2025 and 2026, we think comparing the individual-level results in this way underscores the importance of two features in our open-ended experimental questions.

First, and most importantly, is the chance to name multiple superpowers. Americans often listed more than one country when given the opportunity, regardless of whether they received the unfiltered or filtered version of the new question. Second is the chance to say there are no superpowers or to choose a “Not sure” option. These were key pieces of both experimental approaches.

What’s next?

Given the limited substantive differences in results between the filtered and unfiltered approaches, we’re most likely to use the unfiltered version of our experimental question in future cross-national surveys. This version is also easier to administer for programmers, interviewers and researchers. Results from this experiment also suggest it has the benefit of lower nonresponse rates.