Congress Tweets

In April 2019, the China Data Lab started the “Congress Tweets” project. Since then, we have utilized Twitter’s API to gather 831,331 tweets, including 10,938 tweets related to China, authored by members of Congress from January 1, 2019 to December 31, 2020.

Building off previous blogs in China Data Lab’s Congress Tweets series, we turn our attention to the exploration of topics and information sources within Congressional tweets about China.

Part II: What do Members of Congress Tweet About China?

Bailey Marsheck, Harris Doshay, Lei Guang, Zeyu Li, Molly Roberts, and Young Yang

In our last blog, we described which Members of Congress (MCs) tweeted about China. We found that, while China is an important topic of conversation within MCs’ discussion of foreign affairs, Republicans far out-tweet their Democratic counterparts.

In Part II of our four-part blog series, we ask: how do MCs discuss China in the public sphere?

Specifically, we examine the topic areas revealed by Congressional Twitter posts. What topics are most prominent? How do Democrats and Republicans prioritize different issues? We find that, while issues of human rights and security dominate over discussions of trade, Republicans and Democrats tend to highlight distinct issues and cite vastly different news sources when tweeting about China. Our findings show that America’s general political polarization characterizes MC’s public discussions of China as well.

① For Congress, tweets about security and human rights come first. Then comes trade.

We use a topic model to understand the general themes of Congressional tweets on China. Our model yielded 25 “topics’’ or themes from examining our corpus of China-related tweets. We then label each topic based on the words and tweets most associated with that topic. Having hand-coded each tweet’s sentiment from “Very Negative” to “Very Positive” toward China, we also estimate the relationship between each topic and the sentiment toward China it contains on average.

Figure 1 shows these 25 topics with squares sized in proportion to their prevalence. We find that Congressional expressions on China tend to focus on non-economic issues, even during Trump’s tenure when the “Trade War” was especially salient. We see in Figure 1 that many of the primary topics seized upon by MCs, such as human rights, cybersecurity, Hong Kong, and Taiwan, are non-economic in nature. They are colored red to denote the tweets’ overall negative sentiment. Trade-related issues, on the other hand, are slightly less prevalent.

In contrast to the most popular, highly negative topics surrounding human rights, discussion around U.S. trade with China falls in the “Somewhat Positive” sentiment category. This can be attributed to two factors. First, MCs primarily criticized Congressional opponents, presidential administrations, or even the rules of the international trading system instead of focusing their ire on China. For example, tweets about how Trump was mishandling the U.S.-China trade disputes frequently also emphasized the benefits of U.S. trade with China, making sentiment towards China neutral or positive in such a tweet.

Second, though there were certainly many tweets criticizing China over its trade practices, there were also a large number of Congressional members pointing out their districts' reliance on trade with China. These tweets comprised the “Exports and Trade Deals” topic. Many of these members were mainly concerned with local-level trade ties rather than the overarching US-China dynamic. This topic often involves words like “beef” and “soybeans,” showing the positive influence of China-bound agricultural exports on U.S.-China relations. Trade with China is a multi-layered issue and, as one would expect, it affects local interests in a more varied way than topics like human rights issues or COVID. Bread and butter issues, or perhaps soybean and beef issues for our agricultural states, are the most positive aspects of Twitter-bound China discourse by MCs.

Figure 1. Topics of China Tweets

The topic that occupies the largest proportion of the dataset is labeled “Hong Kong and Tiananmen.” Far from being an amalgamation of two unrelated issues, what they share is an expression of support for movements toward democracy among the Chinese people and condemnation of Chinese state repression. Whether referring to recent events in Hong Kong or highlighting memories of the Tiananmen Massacre, MCs commonly assert that they “stand with the Chinese people” or support efforts to “#FreeHongKong” by calling for an end to “the Chinese Communist Party’s brutal oppression.” This topic, when combined with the related topic that focuses on Human Rights and Uyghurs, takes up around 12% of our observations. Trade related topics, in contrast, only take up less than 8%.

② Republicans highlight security, Democrats focus on domestic issues, both emphasize human rights.

While both Democrats and Republicans talk about human rights, Hong Kong, Taiwan, cybersecurity, and technology frequently, we see several topics to which the two parties give different levels of attention. The top topic for Republicans in our corpus is the origins of COVID, a topic that Democrats rarely discussed. Topics related to research and education in the U.S., such as Confucius Institutes and Research and IP are also more commonly discussed by Republicans than by Democrats. Democrats instead seemed to focus on the trade war and the WTO. Consistent with their focus on Trump’s signature trade initiative, Democrats also sent out more “Trump-specific” tweets. Here, China appears as an issue of political football by the opposing parties. Since the start of the Biden administration, Republicans have played the “China card” in a manner similar to Democrats who focused on criticizing Trump’s policies.

Figure 2. Topic Proportions Within Each Party

Using Figure 2 ’s interactive features, you can explore the topic proportions of tweets from both Democrats and Republicans in the dataset. Topic proportion is valuable for thinking about each Party’s relative focus, as it indicates which issues Republicans and Democrats discuss with their precious airtime. Each party prioritizes different topics, as indicated below:

  • Republicans: “Covid Origins and Accountability”; “IP and Research”; “Tech and Medical Decoupling”; “Budgeting Allocation”; “Confucius Institute”

  • Democrats: “Trade War and the WTO”; “Biden-specific”; “Trump-specific”; “POTUS Dealings”

Topic proportions are calculated by looking at the percentage of tweets taken up by a given topic for each of the two parties. Because of this, it’s important to distinguish between proportion and topic volume. The latter refers to the absolute number of tweets about a topic, and it indicates the party’s overall visibility in key debates. Remember that Republican members tweet about China almost four times as much as Democrats, as detailed in our first blog . This means that Democrats are out-tweeted by Republicans on almost all China-related topics, even on those that registered a high proportion among democrats relative to other issues, like “Trade War” and “Trump.” Thus, Republican messaging on China crowds out Democrats’ in topic volume even on issues where Democrats have a higher topic proportion. Democrats and Republicans may emphasize unique issues of interest within their respective parties, but Republican voices speak loudest in public debate on all main aspects of China-related issues.

③ Both parties cite mainstream news sources, but 30% of Republican citations are from sources Democrats won’t touch.

Similar to the topics of their tweets, the information sources cited by MCs are indicative of America’s polarized political landscape. Even when they weigh in on the same issues or events, Republicans and Democrats do so by directing their viewers towards a different array of outlets (Figure 3) .

Democrats primarily cite high-circulation outlets such as The Washington Post, The Wall Street Journal, The New York Times, , and other mainstream news sources. While Republicans also link to these same outlets, often doing so with even greater frequency than Democrats because of the large volume, what is notable about Republican tweets is that they reference a wider variety of news sources, 30% of which Democrats do not cite. 1

Figure 3. What Sources Do MCs Cite in Their China Tweets?

Mainstream conservative news sources, alternative conservative media, and even fringe sites make up the sources cited almost solely by Republicans. Fox News and National Review are cited frequently by the Republican Party; their inclusion is not surprising, though the distinct lack of citations by Democrats points to a lack of partisan dialogue within Twitter discourse on China. Solidly right-leaning outlets which are less mainstream, The Washington Examiner and The Washington Free Beacon , also have a strong presence among Republican tweets.

Most notably, some of the most vocal Republicans also linked to alternative media such as Breitbart and The Epoch Times , which are known to have printed and promoted misinformation. The viewpoints these publications disseminate have become more popular with conservative politicians in recent years, influencing Republican rhetoric on China. 2

One disturbing aspect of all this is the quality of information our Congressional members rely on when communicating with the public. Prominent D.C.-based think tanks such as Brookings were cited less within China-related tweets than both ESPN and The Huffington Post . Respected sources with contributions from leading China scholars, like ChinaFile, are entirely neglected. While the country’s stock of China knowledge has a place in shaping the foreign policy debate, it is ultimately crowded out when politicians draw their familiar domestic battle lines on social media.

In summary , Congressional discussion of China on Twitter is generally dominated by issues of human rights and security rather than topics that could be considered “closer to home” like bilateral trade or Chinese influence in the U.S. Republicans and Democrats tend to seize upon different sets of topics, though Republicans out-tweet Democrats by a large margin on virtually every issue. In these tweets, Republicans reference a much wider range of sources, including some from disreputable outlets.

The coming installments of this blog series will take further steps toward an important aim of our study: understanding how congress tweets about an increasingly polarized China, taking aim at the sentiment used by MCs to discuss China. Are there any bright spots left in U.S.-China relations? We’ll tackle that and more in our next blogs.


Bailey Marsheck, Chinese Language Fellow for the National Bureau of Asian Research, Tsinghua University's IUP Program; holds a bachelor's degree from UC San Diego and a master’s degree from Peking University

Harris Doshay, Assistant Director of Research and Writing, 21st Century China Center, UC San Diego School of Global Policy and Strategy

Lei Guang, So Family Executive Director, 21st Century China Center, UC San Diego School of Global Policy and Strategy

Zeyu Li, Master of Chinese Economics and Political Affairs (MCEPA) candidate, UC San Diego School of Global Policy and Strategy

Molly Roberts, Associate Professor, Department of Political Science and the Halıcıoğlu Data Science Institute, UC San Diego; Co-Director, China Data Lab at the 21st Century China Center, UC San Diego School of Global Policy and Strategy

Young Yang, Research Data Analyst, China Data Lab at the 21st Century China Center, UC San Diego School of Global Policy and Strategy

Methodological Notes

Topic Modeling

The 25-topic model was derived from the Structural Topic Model (STM) framework, with the topic number selected through a trial-and-error process aimed at maximizing coherence, exclusivity, interpretability, etc. Topic labels were generated through manual evaluation of the model’s output via built-in validation tools: FREX terms and indicative posts. The sentiment classifications correlated with each topic are a product of hand-classification, which will be explained in a future blog post.

News Outlet Citation

News outlet citation data is generated through analysis of the website links included within the text of the tweets posted by each MC.