Measuring National Identity with Conjoint Experiments

Using the case of Taiwan

Steven Denney · Leiden University
H. Christoph Steinhardt · University of Vienna

The problem

“Nationalism is back” … and it was never “gone” in East Asia

Nationalism and national identity are among the most powerful political ideologies of the modern era. The Economist, 2014

Political identity “suggests first and foremost similarity, and it demands acknowledgment of … a ‘defining community’” (Berezin 1997). Nationalism “locates the source of individual identity within a ‘people’” (Greenfeld, 1992).

Common survey measures for the “content” of national identity are problematic.

How do national identities define who is inside and outside of the national community?

Our approach: a conjoint-based measure that is more realistic, rigorous, and volatile than conventional methods.

The problem

Taiwan as an “outcome case”

Gerring and Cojocaru, 2016

Taiwanese identity changed dramatically, is highly contested, and highly politically consequential.

Given internal polarization and sovereignty conflicts with China, as well as ultra-low fertility and increasing immigration, how contested, democratic and open is Taiwanese national identity after all?

Share of the population born in another country, 1990 to 2024: Taiwan, South Korea, Japan, Mongolia, North Korea, China

Concepts

Dimensions of national identity

What is the content of identity? What determines who is a member of the nation and who is not?

Existing frameworks

Civic vs. ethnic nationalism Kohn, 1944

Five elements Smith, 1991

Democratic “constitutional patriotism” Habermas, 1992

Civic, ethnic, cultural Shulman, 2002

Our approach

Ascriptive criteria

Civic criteria

Voluntarist criteria

Kymlicka, 2001; Shulman, 2002; Wright, Citrin and Wand, 2012

Measurement

Existing approaches and the conjoint solution

Identity choice (e.g. Taiwanese, Chinese, both) or nationalism batteries say nothing or little about identity content. ISSP National Identity module asks for importance of different personal traits to ascertain if someone “is a true [British, German…].”

Problems

Ceiling effects

No information about trade-offs

Treats identity as decomposable into independent dimensions

The conjoint approach provides hypothetical profiles and forces respondents to choose. It eliminates these problems and also mitigates social desirability biases (Markovich and Yamamoto, 2022).

The case of Taiwan

Civicness

National identity is a critical variable in political development since 1947. Three highly significant questions that current research has not satisfactorily resolved:

Is contemporary Taiwanese national identity primarily civic, voluntarist?

Qualitative scholarship claims it (Song, 2009; Wong, 2001); survey research provides some indirect, correlational support (Wang and Liu, 2004; Zhong, 2016).

But discrimination of Mainlanders and South-East Asian immigrants remain common (Cabestan, 2005; Kao and Liu, 2025; Rich et al., 2022).

How do Taiwanese value ascriptive vs. civic and voluntarist criteria?

The case of Taiwan

Democracy and contestation

Democracy

Democratization: “the most important shared living experience” of the island’s population (Wang and Liu, 2004).

Some correlational support (Chen, 2024; Qi and Lin, 2021) but understanding of democracy can be substantive rather than procedural (Shi and Lu 2011).

Is there a Habermasian “constitutional patriotism” in Taiwan?

Can the embrace of democracy make immigrants become accepted as members of the national community?

Contestation

National identity is frequently portrayed as “the single most important political cleavage on the island” (Achen and Wang 2017)?

But scholars have also discerned significant consensus (Schubert, 2004; Shen and Wu 2008; Zhong, 2016).

Is identity content as contested as, or more consensual than, conflict over identity labels and partisanship?

Research design

The survey

Online survey of 2,050 respondents in Taiwan between May and June 2024

ISSP National Identity module

ISSP battery in a conjoint experiment

Taiwan specific conjoint experiment

We opted for the terms “a true national of our country” 我們/ 我國真正的國民 and “more like a national of our country” 更像是我國國民 to describe national identity and the task of evaluating hypothetical profiles.

Research design

The Taiwan conjoint

Table 1: CBC attributes, levels, and design rationale

The conjoint choice task as respondents saw it, in Chinese

Findings

The ISSP national identity battery

Factor analysis confirms that there are three dimensions: ascriptive, civic, voluntarist.

6 out of 9 items show little variation.

What happens when someone born outside of Taiwan feels Taiwanese?

Observational national identity battery: percent rating each criterion very or quite important

Findings

The conjoint experiment

Ascriptive components of ancestry and origin provide a modest advantage, and language proficiency matters.

Civic components matter more.

Voluntarist attributes of self-identification (Taiwanese/both vs. Chinese) and regime support (democracy vs. one-party rule).

AMCEs and marginal means for all nine conjoint attributes

Findings

Can voluntarist attributes overcome disadvantaged ascriptive ones?

They can, by a wide margin.

Conditional marginal means: ascriptive marker alone, plus supports democracy, plus identifies as Taiwanese

People with disadvantageous ascriptive criteria (not Han Chinese, foreign born) can become widely accepted (close to 70%) as co-nationals when they support democracy and identify as Taiwanese.

Findings

How contested is it across the main lines of political conflict?

So, Taiwanese national identity appears quite civic, voluntarist and democratic.

Not that much.

Differences across respondents’ chosen identity label emerge over origin (China), self-identification, and regime support.

But except for origin, the differences are matters of emphasis rather than principle.

Marginal means conditional on respondents' identity label

Conclusion

We have introduced a novel approach to operationalize national identity in surveys

It increases realism by measuring the criteria for national belonging as a combined bundle of a person, instead of artificially decomposing.

It increases rigor and provides theoretically required insights by cleanly disentangling effects of one attribute relative to others and by adjudicating how people trade-off different criteria.

Citizens have internalized “constitutional patriotism.” Support for democracy is a qualifying criterion for national belonging.

The finding

Taiwanese are valuing civic and voluntarist criteria more than ascriptive ones when they define the boundary of the nation.

Conclusion

Voluntarist attributes can overcome disadvantaged ethnic and geographic origins

The criteria for national belonging are more consensual than deep political cleavages over identity labels and partisanship would suggest.

Next questions

National identity is mostly treated as a condition for regime type. Do regimes also have agency to mold identity?

Could democracy in Taiwan have produced a less ascriptive, more voluntarist and democratic identity?

Has the one-party regime in Mainland China bred an ascriptive identity? Has it successfully fused national identity with regime support?

Appendix

By liberal-democratic values

Appendix

By party camp

Marginal means for pan-green vs. pan-blue identifiers, and the difference in marginal means