Belief in territorial indivisibility and public preferences for dispute resolution

Abstract This study investigates how individuals may develop more or less strong beliefs in the indivisibility of a disputed territory and how such beliefs may influence their policy preferences toward resolving the dispute. Using a survey experiment in Japan, we find that historical ownership strengthens respondents’ beliefs in territorial indivisibility. Furthermore, those who hold the strongest belief in territorial indivisibility are much less likely to support bilateral negotiation and more likely to support contentious policies, including but not limited to military actions. Finally, we explore external validity of the findings by analyzing respondents who had a real dispute in mind during the survey with China, South Korea, and Russia, respectively.


Introduction
In territorial disputes we often observe that countries neither relinquish a sovereign claim nor settle for a compromise resolution. In East Asia, China, Japan, and South Korea have all made sovereignty claims over the entirety of their disputed territories, and there has been no sign of compromise for over half a century. In South America, the Falklands/Malvinas have been under the British control for nearly two centuries, but Argentina has never ceased to assert its claim over the islands and has refused to accept any bilateral bargaining solution short of full Argentine sovereignty (Laver, 2001). Bolivia has repeatedly refused compromise solution to a sovereign access to the Pacific Ocean that it lost to Chile in 1879, despite having no feasible solution to the dispute (Tomasek, 1967;Zwier, 2013, 217-72).
Yet, the basis of such indivisible territorial claims is not obvious. A territory is always physically divisible, and there are often resources in or surrounding a disputed territory that can be divided and shared. Even sovereignty may be divisible. There is a long history of countries sharing sovereignty over territories or proposing feasible arrangements to share sovereignty and resources. 1 Thus, a claim of territorial indivisibility cannot find its justification in some objective attribute of a territory (Hassner, 2003;Goddard, 2006;Kadercan, 2017); rather, "it is human beliefs and actions that give territory meaning" (Knight, 1982, 157). What kinds of relationship of a people to a disputed territory may give rise to a belief in territorial indivisibility?
And what does the belief mean for individuals' policy preferences toward the dispute? We investigate these questions in this study.
Theoretically, we distinguish between two concepts, issue indivisibility and a belief in issue indivisibility, and argue that the essence of territorial indivisibility lies in the beliefs held by individuals. Moreover, we identify a nation's historical ownership of a disputed territory as one of the important sources of such a belief, and develop an explanation for why. Finally, we argue that those who hold the beliefs in territorial indivisibility can support a range of contentious policies, including but not limited to military actions. We test these arguments empirically using a survey experiment in Japan. The case of Japan allows us to isolate the effect of historical ownership, as Japan's existing territorial disputes do not invoke ethnic or religious claims, while making the hypothetical scenarios of historical ownership sufficiently realistic to our respondents.
We highlight three main findings. First, when exposed to the scenario of historical ownership, individuals develop stronger beliefs in territorial indivisibility, resulting in a more skewed preference ranking over the outcomes. In addition, a non-trivial proportion of respondents developed the extreme belief under the treatment, leading them to only accept the indivisible outcome. The other two contextual variables that we included for comparison, the military strength of the opponent and the economic value of the territory, do not have a similar effect. Second, individuals who hold the extreme belief in territorial indivisibility are much less likely to support bilateral negotiation, and more likely to support contentious policy choices, including military actions but also economic sanctions and arbitration by an international organization (IO). Finally, our analysis of the respondents who had a real dispute in mind during our survey suggests the possibility that individuals are more likely to develop a belief in territorial indivisibility in cases where both sides of a dispute claim historical ownership.
These results shed important insights into the domestic conditions under which leaders operate in territorial disputes. The high level of support for the indivisible territorial outcome, especially when historical ownership is invoked, may explain why leaders frequently make all-or-nothing territorial claims. Moreover, the finding that even those who hold the most extreme belief still have a range of policy preferences challenges the conventional assumption that issue indivisibility leads to conflict, but helps explain why countries with no military options resort to other means, such as IO arbitration, while maintaining their all-or-nothing territorial claims. Our study also makes a methodological contribution to a better understanding of the role of issue indivisibility. Most existing research on issue indivisibility has treated a state as a unitary actor who holds a singular view about the nature of a territory (Fearon, 1995;Hensel and Mitchell, 2005;Goddard, 2006;Powell, 2006;Wiegand, 2011;Kadercan, 2017). By distinguishing between territorial indivisibility and a belief in territorial indivisibility, we have reoriented the unit of analysis in this literature to be individuals in a nation who can meaningfully hold different beliefs about a territory.

Territorial indivisibility as a belief
There is a widely shared view that indivisible issues or goods are those whose value will be destroyed if they are divided; in this view, issue indivisibility is a binary concept (Fearon, 1995;Brams and Taylor, 1996, 51;Goddard, 2006;Kydd, 2015, 72;Kadercan, 2017;Frieden et al., 2019, 132). Where the scholarship diverges is whether indivisibility is a fundamental nature of an issue, or a social construction emerging entirely from a strategic process (Hassner, 2003;Goddard, 2006Goddard, , 2010Wiegand, 2011). Indivisible issues are of enormous concerns in territorial disputes where leaders frequently make such claims.
Leaders are known to have strategic incentives to publicly represent a territorial issue to be indivisible in order to strengthen their bargaining position, sending signals to opponent, or tie their hands, despite the fact that such indivisibility may not reside in the features of the territory itself (Frieden et al., 2019, 132). Therefore, scholars caution that such claims should be given appropriate scrutiny, with some arguing that indivisible territory is a product of incompatible demands from a bargaining process (Goddard, 2006(Goddard, , 2010Wiegand, 2011). However, as Hassner (2003) notes, for instance, "Sacred places are not plots of land to be partitioned by diplomats according to political priorities…They may, for considerable segments of the population, entail meaning that is absolute, irreplaceable and indivisible." Moreover, deeply held values and knowledge of the relationship of a people to a disputed territory may lead to indivisible territorial claims without a bargaining process. In some of the most contentious territorial disputes in Asia, such as those over the Senkaku/Diaoyu islands and Takeshima/Dokdo islands, one of the disputants (Japan and South Korea, respectively) does not even officially recognize the existence of a dispute, let alone engage in negotiation over the territories. Yet, claims of indivisibility over these islands have existed for many decades. An exclusive focus on politicians' strategies thus overlooks the role of the public when numerous research suggests that they are particularly sensitive to territorial disputes (Vasquez, 1993(Vasquez, , 2009Hensel, 1999;Gibler et al., 2012).
To study public attitudes toward territorial disputes that involve indivisible claims, we propose a new concept, a belief in territorial indivisibility, to be distinguished from the concept of territorial indivisibility: even when a disputed territory is physically divisible, individuals may hold beliefs in its indivisibility. It then follows that indivisible territorial claims are necessarily socially constructed. Moreover, in contrast to territorial indivisibility, which is binary, this new concept is a continuous one: individuals can hold more or less strong beliefs about the indivisibility of a territory. What intangible characteristics of a territory may give rise to such beliefs, and what are their effects on public preferences for dispute resolution?
In the literature on issue indivisibility, scholars have identified homelands, historical ownership, ethnic ties, or sacred places to be sources of the perception of territorial indivisibility (Hassner, 2003;Hensel and Mitchell, 2005;Toft, 2006). A second and larger literature on territorial disputes links disputes over territories with intangible salience to more severe conflict. 2 Our study connects these two literatures in a unified framework by investigating first, whether a belief in indivisibility can arise from an intangible characteristic of a territory, and second, how such beliefs may influence individuals' policy preferences.

Rise and consequence of a belief in territorial indivisibility
Of the intangible characteristics of a territory identified in the literature, this study focuses on the historical ties of a nation to a disputed territory. We do so for several reasons. First, historical ownership has been used to justify territorial claims throughout human history (Burghardt, 1973;Murphy, 1990;Newman, 1999, 4;Abramson and Carter, 2016) and has been arguably the most frequently invoked justification in territorial disputes in the post-World War II era (Murphy, 1990). 3 Second, other prevailing explanations for indivisible territorial claims invoking homelands, sacred sites, or ethnic ties (Hassner, 2003;Hensel and Mitchell, 2005;Toft, 2006;Kadercan, 2017;Zellman, 2018) all implicitly rest on the assumption that a claimant once occupied the territory or drew the national boundary over it, regardless of who controls it in the present day. In this sense, historical ownership is the primordial foundation of ethnicity or religion based claims. Finally, historical ownership can be invoked in territorial disputes over uninhabitable sea features such as those in the East and South China Seas, where other commonly used justifications are not applicable.
A nation's claim of historical ownership of a disputed territory implies that the nation is the rightful owner of the territory. If the nation possesses the territory under the status quo but its 2 For a non-exhaustive sample of the works that arrive at the conclusion, see Hensel (1999), Herb and Kaplan (1999), Newman (1999), Hensel and Mitchell (2005), Gibler et al. (2012), Zellman (2015), and Shelef (2016). 3 Murphy (1990, 531) notes "During the past forty years, territorial claims against neighboring states have almost always been justified as attempts to recover land that had been 'wrongly' taken away. . . Historical arguments have come into ascendancy as claims based strictly on ethnic, strategic, and economic considerations have become less acceptable." ownership is being contested by another state, historical ownership may lead individuals to believe that the other state has no foundation for its claim on any part of their territory. On the other hand, if the territory was taken away by another state in the past, then a claim of historical ownership suggests that the event or the process of losing the territory is viewed as unjust by the nation (Murphy, 1990;Zwier, 2013;Fang and Li, 2020); given the grievance, some individuals of the nation may believe that the injustice can only be undone by recovering the entirety of the lost territory, while others may be willing to accept other outcomes that in their view are commensurate with the lost territory. In both scenarios, then, historical ownership will increase individuals' beliefs in their inalienable right to a disputed territory (Zwier, 2013, 250), and likely substantially raise the size of the reparation demanded to compensate for the lost territory.
Of course, politicians or leaders can make historical claims of disputed territories strategically because such claims have been viewed as more legitimate than other types of claims (Murphy, 1990), or because a historical claim may signal the limit of a state's territorial designs, thus decreases the opposition from international actors (Abramson and Carter, 2016). Such potential for strategic manipulation does not rule out the possibility that a public may genuinely believe in their nation's historical ownership of a disputed territory, and develop preferences for how the dispute should be resolved. Depending on their own personal attributes, some individuals may still find divisible outcomes acceptable even in light of their nation's historical ties to the disputed territory; however, on average, we expect historical ownership to have an effect of increasing the perception of territorial indivisibility. These discussions lead to our first hypothesis: Hypothesis 1: A country's historical ownership of a disputed territory leads its citizens to develop stronger beliefs in territorial indivisibility than otherwise.
Next, we consider how individuals' beliefs in territorial indivisibility influence their policy preferences. Departing from the canonical bargaining model of war that links issue indivisibility with conflict (Fearon, 1995), we argue that those who hold such beliefs may weigh the costs and benefits of a range of contentious policy options, including but not limited to military actions, when they reject a compromise outcome. In practice, a bargaining breakdown, or an inability to initiate a bargaining process at all, does not mean a military conflict will follow automatically. For instance, in the disputes that we discussed earlier, some countries simply do not have military means to pursue their preferred outcomes because military balance strongly favors their opponents; yet they do pursue other options such as appealing to the United Nations (UN) or the International Court of Justice (ICJ) while continuing their uncompromising claims. Moreover, even for those countries that do have military options, the cost of war and the associated risk may make it worthwhile for them to consider other options, such as IO arbitration or economic sanctions. Similarly, between their beliefs and policy preferences, individuals may consider a range of policy options rather than simply supporting the military option. Therefore, our second hypothesis reads: Hypothesis 2: Individuals with stronger beliefs in territorial indivisibility are more likely to support contentious policies for dispute resolution.

Experimental design
To test our hypotheses, we implemented a survey experiment in Japan. The survey began by asking respondents to read a hypothetical scenario of territorial dispute embedded with three randomized features: historical ownership of the territory, the military strength of the potential opponent, and whether or not the territory has economic values. While our theory and first hypothesis were concerned with the effect of historical ownership, we included two additional features of the dispute as likely alternative explanations. Intuitively, if the military power of the opponent is weak, then our respondents might be more emboldened in their support for indivisible outcomes and hardline policies, and vice versa. The economic value of a disputed territory might also promote a preference for possessing the entirety of a disputed territory and support for hardline policies.
Some combination of these features match Japan's three ongoing territorial disputes: the Senkaku/Diaoyu islands dispute with China, the Northern territories/Kurile islands dispute with Russia, and the Takeshima/Dokdo islands dispute with South Korea. Thus, the hypothetical dispute scenarios are sufficiently realistic for respondents to develop informed opinions. Respondents were told that they did not need to think of a particular case in answering the questions, but many did make the connection and later we use this information to analyze real territorial disputes. The hypothetical scenario reads as follows: Japan is involved in a dispute with a [militarily strong/weak] neighboring country over a piece of territory (an island). [This territory has economic value/The economic value of the territory is unknown], and it [historically belonged to Japan/historically belonged to the neighboring country/historically did not belong to any country].
Note that the treatment in which the disputed territory "historically belonged to Japan" does not rule out the possibility that the neighboring country makes a conflicting historical claim. In the real world, countries on both sides of a territorial dispute may claim historical ownership, thus such claims need not be mutually exclusive. 4 As discussed above, other characteristics of a territory that derive from historical ownership may also lead to the formation of a belief of territorial indivisibility. Nevertheless, the case of Japan allows us to isolate the effect of historical ownership without having to control for confounders like the ethnicity of the resident on the disputed territory. Two of the three disputed territories are either uninhabitable or barely inhabited islands (Senkaku/Diaoyu and Takeshima/ Dokdo), and the third one involve residents that are not Japanese nationals (Northern Territories). Therefore, it is highly unlikely that respondents exposed with the historical ownership treatment would base their reasoning on ethnic ties to the disputed territory, or it being a homeland for the Japanese. On the other hand, not many Japanese respondents would have had precise knowledge of the historical ownership of Japan's disputed territories. This in turn increases the mundane realism of our main treatment of historical ownership, also the most manipulable for leaders and analysts alike in the Japanese context.
After reading the scenario, respondents were asked two questions in sequence. The first question aimed to test our first hypothesis regarding the effect of historical ownership as a source of a belief in territorial indivisibility. While we cannot directly measure an individual's belief in territorial indivisibility-a latent variable, we can infer how strongly an individual holds such a belief from her expressed preference over different dispute outcomes. Specifically, we presented the four most realistic outcomes of the dispute, from divisible to indivisible, and then asked whether the respondents found each outcome acceptable or unacceptable, or whether they were "unsure." We expect individuals who hold stronger beliefs to more likely accept indivisible outcomes and less likely to accept divisible outcomes. Giving respondents a range of outcomes has the advantage over asking directly whether they thought that the disputed territory was divisible or not, which might have resulted in respondents overwhelmingly choosing the socially desirable outcome of the territory being indivisible. The four outcomes were: (1) Japan and the neighboring country jointly hold the sovereignty and jointly use the territory. 4 Mutual claims of historical ownership should not pose problem for this study as long as respondents of our survey chose their answers based on the information that they believed to be true.
(2) Japan alone holds sovereignty over the disputed territory, but the two countries jointly use the territory. (3) Japan will independently maintain sovereignty and monopolize the use of the disputed territory. However, Japan will provide financial or political compensation to the neighboring country. Both countries have agreed on the content of compensation. [Also, the agreement will be monitored by international organizations such as the UN and the ICJ./none.] (4) Japan will independently maintain sovereignty and monopolize the use of the disputed territory. Japan does not provide any compensation for the neighboring country.
The four outcomes are all realistic in that they were proposed at different times in the real disputes involving Japan; this would allow the respondents to more easily envision what the options would look like. 5 The outcomes are differentiated by their divisibility: they are different combinations of who owns the sovereignty and/or the right to use the disputed territory. The first two outcomes allow joint sovereignty and/or joint use of the territory, and thus are "divisible" outcomes. The last two outcomes do not allow sharing of either, and thus are "indivisible" outcomes with one allowing side payments to facilitate a compromise solution. For this outcome, we additionally told half of the respondents that the agreement would be monitored by international organizations such as the UN and the ICJ. This design allowed us to explore whether respondents were concerned about the credibility of a bilateral agreement (Fearon, 1995;Powell, 2006). In the second question, respondents were given seven policy options, ranging from propaganda campaign to full military actions, following a statement: "The Japanese government has taken or can take the following policies regarding territorial disputes." The statement was to remind the respondents that all the policies presented were within the realms of possibility, including the more peaceful measures. We believe that this reminder of the peaceful options would provide a hard test of our second hypothesis regarding the link between beliefs in indivisibility and support for hardline policies. We then asked the respondents whether they found each of the options appropriate for the hypothetical dispute scenario that they just read or were "unsure." Figure 1 illustrates these policy options ordered roughly from the most cooperative to the most conflictual. 6 The most cooperative option, "bilateral negotiations," is aimed at reaching a compromise solution between the disputants. The next one, "shelving the dispute," has been proposed by China over the Senkakus/Diaoyu islands dispute, and the proposal involves joint development of the resources in or around the disputed territory while leaving the sovereignty issue unresolved. The rest of the options contain varying degrees of confrontation. "Publicity" involves strengthening external propaganda, stimulating public opinion in Japan, and urging citizens to express dissatisfaction with the neighboring country. "IO arbitration" means taking the dispute to an international organization such as the UN or the ICJ. Such an approach is frequently contentious because only one side of a dispute finds it in its interest to appeal to international arbitration (Merrills, 2005, 272;Fang, 2010;Press Association, 2013). Indeed, Japan has advocated international arbitration for the Takeshima/Dokdo dispute, while South Korea has opposed it (Miller, 2014). Respondents who found international arbitration acceptable received a follow-up question: "Do you think that Japan should abide by the decision made by the UN or the ICJ, whatever the outcome may be?" The possible answers to this question were "yes," "If it is not a ruling that will benefit Japan, it is not necessary to follow," and "unsure." Further down the conflictual side of the continuum, we have "economic sanctions," including stopping official visits and canceling cooperative projects. Finally, we have two most belligerent options that involve Japan's Self-Defense Forces, with the difference being whether the military actions are more limited or full-scale. Taken together, the seven options cover most realistic policies available to Japan 5 The list of outcomes is by no means exhaustive, but contains the most realistic ones from Japan's perspective given its existing territorial disputes. 6 The order of these options in the survey is randomized. See Appendix I for exact wording used for the policy options. on different disputes and at different times, even if these policies may not be the preferred policies of the relevant governments in the present day.
After completing the two questions, respondents were asked whether they had envisioned an actual territorial dispute when they were answering the two questions. If the answer was "yes," we asked them to specify the dispute. In the remainder of the survey, we asked typical sociodemographic questions, including instruments that measure an individual's degree of nationalism. 7

Data and main results
The survey was administered in September 2016 by Nikkei Research, an Internet marketing research firm in Japan. The respondents were randomly drawn from Nikkei Research's online subject pool of over 145,000 panelists. 8 Quota sampling was implemented to match the national average in terms of geographical locations, and yielded a sample of 2621 Japanese adults. 9 The percentage of sample respondents drawn from different regions in Japan are comparable to the national averages. 10 Figure 2 summarizes the average level of support for the dispute outcomes over the three historical ownership scenarios. The vertical axis represents the proportion of support, and the horizontal axis lists possible outcomes of the dispute. As explained in the last section, we asked respondents to consider both divisible and indivisible outcomes of the dispute; further, to account for the possibility of third-party enforcement, for the side-payment outcome, we told half of the respondents that the agreement would be monitored by international organizations. The circles, triangles, and squares in the figure are the point estimates for the proportion of respondents who found the outcomes acceptable given different scenarios of historical ownership; the bars are 95 percent confidence intervals. 11 We highlight three findings in Figure 2. First, when respondents were told that Japan historical owned the disputed territory, a significantly higher proportion of respondents found the  See the survey design for specific questions. 8 For recent studies that employed Nikkei Research, see, for example, Tago and Ikeda (2015) and Kohama et al. (2017). 9 We include in the survey two attention checking questions, asking respondents to make a specific choice. The vast majority of the respondents were paying attention; only 62 out of 2621 failed both attention checking questions. The results that we present in the paper are from the full sample. 10 The percentage of samples drawn from the six regions in Japan (Hokkaido-Tohoku, Kanto-Koshinetsu, Chubu, Kinki, Chugoku-Shikoku, and Kyushu-Okinawa) are 11.8, 36.6, 15, 16.7, 8.7, and 11.3 percent, respectively. The national averages (as of 2014) are 11.4, 37.8, 14.2, 16.33, 8.9, and 11.4 percent. See Appendix C for a comparison of our sample with the census data in terms of age, gender, education, and income, showing that our sample is nationally representative on these key demographic variables. Summary statistics for the sample can be found in Appendix B. 11 In this figure, as well as in all of the remaining results, we omit respondents who replied that they were "unsure" (about 20-25 percent of the respondents). See Appendix D for the results that combined the indecisive responses with the "unacceptable" ones. The results were very similar.

Historical ownership and a belief in territorial indivisibility
indivisible outcome acceptable (Japan has sovereignty and the exclusive right to use) compared with the other two scenarios. Conversely, they were much less likely to accept joint sovereignty and joint use of the territory in this scenario (25 versus 42 or 45 percent). The differences are all statistically significant. These results support hypothesis 1, that is, historical ownership made a difference in the respondents' preferences regarding the divisible and indivisible outcomes and in the directions that we hypothesized. Moreover, respondents cared much more about the specific outcome of the dispute if Japan had historical ownership. Specifically, in the scenarios where Japan did not have historical ownership, the support levels for different outcomes did not differ greatly, whether in the case of foreign historical ownership or the case of "owned by neither." But in the case of Japanese historical ownership, there was a wider distribution of preferences; in other words, historical ownership sharpens the preferences of the respondents.
Second, under the treatment of historical ownership, while some compromise outcomes also received significant support, suggesting that respondents' beliefs in indivisibility were not equally strongly held, there is a clear preference ranking over the outcomes in the aggregate, from the most indivisible to the least indivisible in descending order. In particular, the indivisible outcome is by far the most popular one. This may explain why leaders often take the extreme all-or-nothing position. Third, adding IO enforcement made no difference in the respondents' preferences for the indivisible outcome with a side payment: the proportions of respondents who found the outcome acceptable are similar across all scenarios whether or not there is an international organization monitoring the agreement. This result suggests that the enforcement problem is not a particular concern for respondents; the content of the agreement is what matters. Finally, no other characteristics of the dispute, whether the neighboring country's militarily strength or the economic value of the territory, affected respondents' outcome preferences. 12 Taken together, these findings confirm that historical ownership engendered in some respondents a belief in territorial indivisibility, and thus reduced their willingness to accept outcomes that involve compromise.
To more directly gauge the effect of historical ownership on the development of individuals' beliefs in territorial indivisibility, we divided the respondents into two groups: the "hardcore indivisible" group and the "compromise possible" group. We put in the "hardcore indivisible" group those respondents who answered "acceptable" only to the indivisible outcome where Japan has both sovereignty and right to use the territory without making any compromise. The rest of the respondents were relegated to the "compromise possible" group-these respondents found acceptable either some sort of sharing of the territory or no sharing but making side payments to the neighboring country. We dropped the 232 respondents who were unsure about all of the outcomes. Out of the remaining 2389 respondents, 381 belonged to the "hardcore indivisible" group. Figure 3 presents the proportion of "hardcore indivisible" respondents for each value of the three contextual variables. The proportion of individuals in the "hardcore indivisible" group is significantly higher under the hypothetical scenario where Japan historically owned the disputed territory than that in the other two scenarios of historical ownership. In the meantime, there are no discernible differences between the two groups in the military strength and economic value treatments. These results confirm hypothesis 1 from a different angle, showing that historical ownership is more likely to lead to a belief in territory indivisibility, prompting some citizens to reject all but the most demanding indivisible outcome.

Territorial indivisibility and policy preferences
We now turn to our second hypothesis and investigate whether a belief in territorial indivisibility affected respondents' policy preferences regarding the hypothetical dispute. Since respondents' beliefs in territorial indivisibility are not experimentally manipulated, we use logistic regression to analyze the effect of the perceived territorial indivisibility on policy preferences. The key independent variable, indivisibility, is a binary measure that equals 1 for the "hardcore indivisible" group and 0 for the "compromise possible" group. 13 The results are presented in Table 1. We can see that except for the policy option of IO arbitration, the differences in support for the remaining options are all statistically significant between the two groups. Moreover, the respondents in the hardcore indivisible group are more likely to support belligerent policies, such as economic sanctions, limited and full military actions, but less likely to support bilateral negotiation and shelving the dispute. These results are consistent with our second hypothesis. The results are robust against an alternative, continuous measure of a respondent's degree of belief in indivisibility based on item response theory (Hambleton and Swaminathan, 1985), 14 and controlling for geographic proximity of the respondents to Japan's disputes (Tanaka, 2016). 15 To further explore how a belief in territorial indivisibility affected respondents' preference rankings of different policy choices, in Figure 4 we juxtapose the predicted probabilities of support for each policy for both the "hardcore indivisible" and "compromise possible" groups. For the "compromise possible" group, bilateral negotiation received the highest support at 89 percent and full military action received the lowest support at 18 percent. Generally speaking, this group's level of support decreases as the policy options become more bellicose, with the exception of IO arbitration, a finding that we discuss further below. Surprisingly, there is less support for economic sanctions than for limited military action, while support for full military action is lower than both. Turning to the "hardcore indivisible" group, support for bilateral negotiation dropped 13 The model also includes a range of controls, which we omit here due to space constraint, but the full results are available in Appendix I. 14 See Appendix E for the result using the IRT measure. 15 See Appendix F for details. significantly, while support for IO arbitration remained very high at 87 percent, followed by limited military action (77 percent). Similar to the "compromise possible" group, the support for economic sanctions (59 percent) falls behind limited military action but is at about the same level as full military action (57 percent). It is likely that the respondents considered the costs of different actions, and viewed the costs of economic sanctions as greater than those of limited military action for Japan. Overall, the hardcore group favors more contentious policies. The differences between the two groups' percentages of support for various policies are all statistically significant and the substantive differences are large: the "hardcore indivisible" group is 33, 39, and 39 percent points more likely to support economic sanctions, limited military action, and full military action, respectively, than the "compromise possible" group. These findings provide strong support for our second hypothesis.
The patterns in Figure 4 also yield interesting insights about how country-specific variables may lead to different public attitudes toward a range of policies across countries. We focus on   two of such variables. First, a country's position in the status quo of its disputed territories may matter. If a country is favored in the status quo-for example, having the effective control of the disputed territory-then the public (and the country's leaders) will see little benefit in inviting IO arbitration. This is because the country has little to gain from such litigation, while a ruling against its position will generate reputational costs and embolden its opponent. On the other hand, the country that is disadvantaged in the status quo, for example, having no control of the territory, may view appealing to IO arbitration as an option that could bring legitimacy to their claim. Thus, the fact that support for IO arbitration was close to 90 percent among Japanese respondents for both the "hardcore indivisible" and "compromise possible" groups may be explained by the fact that Japan does not have effective control over two of its three real-world territorial disputes with South Korea and Russia. 16 In a similar survey in China (Fang and Li, 2020), Chinese respondents were found to be much less supportive of IO arbitration (62 percent for the "compromise possible" and 38 percent for the "hardcore indivisible" group), likely reflecting the fact that China has the effective control of some of the disputed islands in the South China Sea, and indeed the country has refused to participate in international arbitration initiated by the Philippines (Thayer, 2021). 17 Second, a country's economic conditions, perhaps in relation to its opponent in a territorial dispute, may explain different levels of support for economic sanctions across countries. Japanese respondents showed much less support for economic sanctions (59 percent for the "hardcore indivisible" group and 26 percent for the "compromise possible" group) than their

16
Consistent with this explanation, in the Takeshima/Dokdo dispute, South Korea has the effective control of the island and has refused Japan's proposal to take the dispute to the ICJ (Miller, 2014). 17 For additional examples, Spain has sought to take its dispute with Britain over Gibraltar to the UN and the ICJ, and Argentina has brought the Falklands/Malvinas dispute to the UN. In both cases, Britain has the effective control of the territories and saw such measures as an escalation of the conflicts (Press Association, 2013).
Chinese counterparts (78 and 64 percent for the two groups) in the China study (Fang and Li, 2020). The difference in the willingness to engage in economic punishments might be a function of how respondents perceived their country's economic strength vis-à-vis its opponent, as such actions impose costs on itself and raise the risk of retaliation by the other side. Japanese respondents, on average, perceived such costs to be higher than the Chinese respondents.

Real versus hypothetical disputes
Recall that in the survey, respondents were told that they did not need to think of a particular territorial dispute when they read the hypothetical scenario. However, due to the high salience of Japan's existing territorial disputes and their similarity with some of the hypothetical scenarios, it is entirely possible that respondents were thinking about a particular dispute when they answered the questions. Indeed, we found that in our survey, 1410 out of the 2621 respondents said "yes" when we asked them whether or not they had a real country in mind as the hypothetical neighboring country. We discuss the results from this group of respondents in this section. 18 As described earlier, we asked those respondents who said they had a particular country in mind to further elaborate in a follow-up question which country it was. Out of the 1410 answers, we focus on those that touch on Japan's three main territorial disputes, which are with China, Russia, and South Korea. To facilitate comparison, we removed respondents who mentioned multiple countries. In total, 423 respondents were thinking (only) about the Senkaku/Diaoyu islands dispute with China; 207 were thinking (only) about the Northern territories/Kurile islands dispute with Russia; and 135 were thinking (only) about the Takeshima/Dokdo islands dispute with South Korea.
We first examine the effect of historical ownership on respondents' preferences for the indivisible outcome, that is, Japan having both the sovereignty and the right to use the disputed territory. Figure 5 plots the results for the three groups that had China, Russia, and South Korea in mind when responding to the survey. As a comparison, we also plot the results from those respondents that did not have any real country in mind. Similar to our finding in the full sample, those respondents who were informed that Japan historically owned the territory were the most likely to find the indivisible outcome acceptable compared with the other two treatments.
Additionally, we can see in Figure 5 that, on average, the proportion of respondents who perceived the territory to be indivisible was higher for the groups that had Japan's disputes with China and South Korea in mind than that for the other two groups, regardless of the ownership status. What might explain this result? It is well known to the Japanese public that China claims historical ownership over the Senkakus/Diaoyu islands, and South Korea claims historical ownership over the Takeshima/Dokdo islands; that is, both Japan and its neighbors claim historical ownership in these two disputes. In contrast, in the Northern territories/Kurile islands dispute with Russia, only Japan claims historical ownership while the Russian sovereignty claim began as a result of World War II (Sputnik, 2008;Reuters, 2021). Thus, even though we did not prime respondents with the nature of the claim made by the neighboring country in this treatment, it is possible that respondents thinking about China or South Korea implicitly took into account the fact that these countries also make historical claims, and that strengthened their preferences for the indivisible outcome. The finding shows that conflicting historical claims have the potential to harden public attitudes toward a territorial dispute, perhaps by evoking nationalistic sentiments.
Next, using the same specifications in Table 1, we reanalyzed the effect of the belief in territorial indivisibility on policy preferences in the three groups of respondents. The resulting 18 The results for the respondents who did not have any real dispute in mind are nearly identical to the main findings reported in the previous section. See Appendix G for details. coefficient plots for each of the seven policy outcomes are presented in Figure 6. 19 Similar to the results in Table 1, those who believed the disputed territory to be indivisible (the "hardcore indivisible" group) were less likely to pick bilateral negotiation regardless of which real dispute they had in mind. For the remaining policy options, however, the differences between the "hardcore indivisible" and "compromise possible" groups vary considerably depending on which dispute the respondents had in mind. In the case of the Senkaku/Diaoyu islands dispute with China, the "hardcore indivisible" group was less likely to support shelving the dispute (a policy proposed by China in the past), but did not differ from the "compromise possible" group with respect to the other options. For those thinking about the Northern territories/Kurile islands dispute with Russia, the "hardcore indivisible" group preferred publicity and economic sanctions. In the case of the dispute with South Korea, the "hardcore indivisible" group was more supportive of several belligerent policy options (economic sanctions, limited and full military actions) than the "compromise possible" group, and less supportive of shelving the dispute. It could be that these respondents believed that the belligerent options were better than the cooperative ones in achieving the indivisible outcome because they viewed South Korea as a weaker economic and military power than Japan.
The variation across the three cases is informative. It suggests that respondents were likely taking into account the potential costs when evaluating the policy options in each real dispute, and were reluctant to resort to military actions against Russia (a powerful military opponent), and economic sanctions against China (Japan's largest trading partner). In other words, while respondents in the "hardcore indivisible" group all believed in territorial indivisibility, they expressed different policy preferences when confronted with different opponents, some involving the risk of military conflict, while others involving belligerent but less aggressive policies. The For a clear visual presentation, we omitted the contextual variables and sociodemographic controls, but the full results can be found in Appendix H. phenomenon provides further empirical support for our broader conceptualization of the relationship between beliefs in territorial indivisibility and policy preferences.

Conclusion
Compared with the informational problem and the commitment problem, issue indivisibilitythe third of the three causes of war identified in the bargaining theory of war-has attracted significantly less attention by way of rigorous theoretical and empirical research. Yet in many highly salient territorial disputes in the world today, a common feature is that disputants claim the whole of a disputed territory, leaving no room for compromise, and have done so consistently for decades, or even for over a century. Such a phenomenon calls for a better understanding of the nature and the effects of indivisible claims in territorial disputes, and their effects on policy preferences. This study seeks to address these questions theoretically, empirically, as well as methodologically.
Theoretically, we distinguish between the concepts of issue indivisibility and a belief in issue indivisibility, and argue that the essence of territorial indivisibility lies in individuals' beliefs rather than the physical characteristics of a territory. Moreover, we argue individuals may develop beliefs in territorial indivisibility from their nation's historical ownership of a disputed territory. Using a survey experiment conducted in Japan, we empirically examine how historical ownership may give rise to respondents' beliefs in territorial indivisibility and the subsequent effect of such beliefs on respondents' policy preferences toward resolving the dispute.
The empirical results show that historical ownership led to stronger beliefs in territorial indivisibility, making compromise outcomes less acceptable and the indivisible outcome more acceptable to Japanese respondents. Furthermore, those who held the extreme beliefs and only accepted the indivisible outcome supported a range of more conflictual policies, including but not limited to military actions. Finally, there is suggestive evidence that respondents were more likely to Fig. 6. Coefficient estimate of "hardcore indivisible" on respondents' policy preferences by dispute. Point estimates with 95 percent confidence intervals for the "hardcore indivisible" variable were obtained from logistic regressions using the same specifications as in Table 1. Other variables were omitted for brevity. develop a belief in territorial indivisibility in cases where both sides of a dispute claim historical ownership.
Our study also makes a methodological contribution by addressing an empirical challenge in the study of issue indivisibility in international relations, namely, measuring individuals' beliefs in issue indivisibility. Measuring such beliefs is difficult because if researchers ask respondents directly what they think of the divisibility of an issue, such as a disputed territory, they are unlikely to receive highly reliable answers because of social desirability bias. The survey design we developed addresses this challenge and allows us to measure respondents' latent beliefs through their revealed preferences over different outcomes of a disputed issue. Our research design can be modified to apply to other intangible characteristics of territories to further investigate the rise and consequences of a belief in territorial indivisibility in world politics.
How generalizable are our findings from Japan to other countries? The key feature of the territorial claims that we investigate in this study-historical ownership without ethnic or religious ties -can be found in many other real-world disputes. For example, China's territorial disputes with the Philippines in the South China Sea involve claims of historical ownership over many uninhabited islands. In their territorial disputes with Britain, Argentina and Spain claim historical ownership of Falklands/Malvinas and Gibraltar, respectively, but the residents of these islands are British. To study these cases, our survey design for Japan requires only slight modifications to adapt to local contexts. Results from a similar survey conducted in China show that the main findings regarding historical ownership are robust across these two countries (Fang and Li, 2020). A particularly interesting finding from the two studies is that among those who believed in territorial indivisibility, economic sanctions received as much as or more support than full military actions.
Where we may observe different patterns in public attitudes toward territorial disputes across countries, or even across disputes within a country, are their policy preferences stemming from country (or even dispute)-specific contexts. In general, we expect public support for different policy options to be influenced by country-specific contexts; at the same time, we believe such variation can be understood and predicted by carefully developed theories. One of the most important country-specific contexts in territorial disputes that we highlight in this study is a country's relative position in the status quo of a disputed territory; this position can influence respondents' support for a policy option based on whether the option will bring a better outcome than the status quo. We conjecture that countries who are disadvantaged in the status quo in a territorial dispute are more likely to seek IO arbitration because a favorable international ruling can rally international support and strengthen the country's claim. On the other hand, if a country has an effective control of a disputed territory, it would likely oppose an IO arbitration because the country has nothing to gain from an international ruling, but something to lose vis-á-vis the status quo. Seen in this light, our finding that a very high proportion of Japanese respondents support IO arbitration likely reflects the fact that in two of the three territorial disputes that Japan is involved in, the other countries (South Korea and Russia) have the effective control of the territories. Further investigation of the effect of such country-specific contexts on public support for different policy options will be a natural extension of this study.
Supplementary material. The supplementary material for this article can be found at https://doi.org/10.7910/DVN/ I5JA7L.