Immigration,ethnic diversity and public goods provisioning: Evidence from rural communities in Uganda

Abstract

A considerable body of literature suggests that immigration undermines voluntary contributions to public goods because it leads to ethnic diversity, which erodes social trust. This article posits that the effect of immigration outweighs that of ethnic diversity, so that immigration may explain why ethnic diversity is negatively associated with social trust and public goods provisioning. I also highlight a need to emphasize the moderating influence of transaction costs when analyzing provisioning problems associated with immigration and ethnic diversity. To examine my hypotheses, I use a mixed-method research design to study public goods management in randomly selected communities in rural Uganda whose rates of immigration and levels of ethnic diversity vary. I analyze community-level attempts at collective action that involve substantially different costs; that is, contributing to toilet construction and participating in litter pickup programs. The findings suggest that socio-political barriers to collective action for public goods provisioning may have less to do with the stock of demographic diversity than the flow rate of demographic change. The same findings suggest a more micro-level explanation that transcends the erosive effects (of immigration and ethnic diversity) on social trust to emphasize the moderating influence of transaction costs.

Keywords

Ethnic diversity immigration public goods institutional analysis community-based governance africa uganda sanitation toilet construction

Introduction

A critical question in social science literature is why some communities are better at generating public goods, such as efficient transport, good schools, clean drinking water, and adequate sanitation (Agrawal, 2001; Ostrom, 2010; Yamagishi, 1986). One prominent argument in recent scholarship is that ethnic diversity explains the variations in outcomes. From African rural communities to American inner cities, studies have repeatedly established the undermining effect of ethnic diversity on public goods provision (Easterly and Levine, 1997; Alesina et al., 1999; Alesina et al., 2004; Miguel and Gugerty, 2005; Kimenyi, 2006; Putnam, 2007; Kumlin and Rothstein, 2007; Letki, 2008; Holtug, 2010). Intriguingly, such studies often examine ethnic diversity in tandem with immigration and, the effects of the two phenomena are often considered equivalent. Also, those studies often present the suggested undermining effect on public goods provision as if it were generalizable across public goods with differing characteristics.

But if we compare communities with varying levels of ethnic diversity and similar experiences of previous immigration, do we inevitably find that ethnic diversity undermines public goods provision? And if we compare the effects of ethnic diversity and immigration on the provisioning of public goods with substantially different costs, do we inevitably find undermining effects for each public good? In this article, I emphasize the collective action problem entailed in contributions towards public goods to posit that the rate of previous immigration may have a stronger negative impact (when compared to ethnic diversity) on the willingness of community members to contribute to public goods. Additionally, I posit that the impacts of either immigration or ethnic diversity on willingness to contribute to public goods may be more noticeable in cases where the costs involved in provisioning processes are higher.

A four-part logic underlies the above hypotheses. First, although public goods usually avail benefits, their provisioning entails costs. Second, individuals are more likely to support collective action for public goods provisioning when the involved costs are appreciably low. Third, if costs are high, individuals are more likely to assist in providing public goods– and incur the internal costs– when they have had a period of interaction with the community that is sufficiently long for them to develop trust that a decision to contribute to the public good will be reciprocated by other community members. Fourth, when such social trust exists, individuals are more likely to support collective action for provisioning a costly public good.

Thus, I posit that the experience of immigration necessarily undermines public goods provisioning since it determines whether community members have interacted for a sufficiently long period to build trust.¹ I expect that the effects of immigration trump any effects of ethnic diversity because, although co-ethnicity can also facilitate trust building either through prior relationships and (or) shared norms, it is not necessarily associated with the length of previous interactions. I also posit that any identified undermining effect (of either immigration or ethnic diversity) should be less apparent when the public good under consideration entails lower costs. To examine these claims, I focus on the communities on Buvuma island, which is situated in Uganda’s portion of Lake Victoria, to study two aspects of public goods provisioning (contributions to toilet construction and participation in litter cleanup programs) that involve substantially different magnitudes (and types) of costs.²

The communities on Buvuma island are appropriate for this study largely because their rates of immigration and levels of ethnic diversity vary sufficiently to enable comparative analyses. A focus on this island can also shed light on communities of other Lake Victoria islands. For, as with Buvuma, most communities in Lake Victoria have several prerequisites for successful public goods provisioning, such as small populations of interdependent individuals, shared languages that foster communication, small territories that enable interpersonal contact, and clearly defined boundaries that ease the exclusion of outsiders.³ Migration and ethnic diversity in these communities are associated with resource misuse and livelihood challenges (Berling et al., 2013; Nunan, 2010; Silsbe and Hecky, 2008). However, the impacts of those two factors on local processes of public goods provisioning have not been analyzed in prior studies.

By problematizing a common assumption that migration-based and other ethnic-based demographic characteristics are interchangeable, the present study contributes to (and builds on) an emerging theory that the effects of migration-based ethnic diversity on social trust are contingent upon interactions among individuals in the group under study (Kokkonen et al., 2014). In addition, by analyzing the moderating influence of transaction costs on the effects of immigration and ethnic diversity, the same study emphasizes a micro-level explanation that is often sidelined. Nevertheless, although this study enables a more micro-grounded analysis than is possible in larger samples that rely on observational data, throughout this paper, I highlight limitations that pose challenges to out-of-sample validity and, in the conclusion, I suggest how to overcome them in follow-up research. Therefore, the evidence presented should be considered suggestive (not conclusive) to motivate further research and analysis in other contexts.

Immigration, ethnic diversity and contributions to public goods

The literature on voluntary contributions to public goods tends to focus on analyzing the collective action problem involved. Some eminent scholars suggest that overcoming such a problem requires a cooperative equilibrium which, in turn, depends on expectations, or trust, that one’s cooperation will be reciprocated, and defection punished (Ostrom et al., 1994; Fehr and Gachter, 2000). Indeed, to explain why ethnic diversity undermines public goods provision, evidence has been produced for a causal mechanism that can best be understood by considering theoretical games with multiple equilibria (Habyarimana et al., 2007). Contributing to a public good is a favored strategy in such games only if players trust counter-players to contribute as well; such that a player's ethnic background may determine the achievement of a cooperative equilibrium if all players trust their own ethnicity and distrust non-identical ones.⁴

This argument may imply that there is no possibility for collective action among non-coethnics. Where the diversity effect is presented as generalizable across public goods (Habyarimana et al., 2007), we can expect non-cooperative behavior among non-coethnics irrespective of the characteristics of the public good under consideration.⁵ And when pre-existing opinions about non-co-ethnics are considered to be based on real differences in preferences, we can expect that those opinions cannot be overcome by interpersonal contact (Miguel and Gugerty, 2005; Leigh, 2006; Khwaja, 2009). Alternatively, the argument may suggest a possibility for collective action among non-co-ethnics; for example, we can expect cooperation among non-co-ethnics when the costs related to the provisioning of public goods are sufficiently low (Jackson, 2013) and, when those non-co-ethnics learn each other’s preferences through interpersonal contact (Taylor, 1998; Bobo and Tuan, 2006).

There is an implicit assumption in this literature that being an ethnic minority is correlated with being an immigrant (Lancee and Dronkers, 2007; Huijts et al., 2014; Savelkoul et al., 2015). Immigration is also often assumed to be associated with ethnic diversity in ways that make any revealed effect susceptible to confounding bias (Alesina et al., 2004; Putnam, 2007; Kumlin and Rothstein, 2007; Letki, 2008; Holtug, 2010). Consider Robert Putnam’s influential conclusion, based on dynamics in the US, that immigration and ethnic diversity reduce social capital in the short run while having medium to long-run social benefits. Such arguments seem to assume a priori that immigration necessarily leads to ethnic diversity, and that the effects of the two phenomena are equivalent. But such a relationship is not always deterministic. One can even logically imagine both ethnic demographic change without migration (due to differential birth/death rates) and greater ethnic homogeneity due to the immigration of co-ethnics.

Few existing analyses set out to dissociate the effects of immigration and ethnic diversity on social trust and public good provision. Studies that challenge the “undermining effect of ethnic diversity” often focus on public goods whose provisioning largely relies on third parties/governments (Lee, 2018; Kustov and Pardelli, 2018; Singh and Hau, 2016; Gisselquist et al., 2016; Wimmer, 2016)– rather than on the principle parties through collective action. And, although scholars of migration-based ethnic diversity reveal a negative association with social trust when intergroup contacts are easy to avoid (Kokkonen et al., 2014; Van der Meer and Tolsma, 2014), this argument has not been thoroughly tested empirically; and generally, the interrelationships of immigration, ethnic diversity, social trust, and public goods provisioning remain under-examined.

The above review of existing literature suggests a need for research that is designed to isolate and comparatively analyze the effects of immigration and ethnic diversity on public good provisioning, while simultaneously considering whether any identified effect is generalizable across public goods whose provisioning involves different costs. Otherwise, it could be that due to confounding bias, the suggested effect of ethnic diversity on public good provision does not exist and, that it masks an actual effect of immigration. Also, any identified undermining effects could be, far from being generalizable, only operative when public goods with certain characteristics are considered. In this paper, I examine the causal significance of both ethnic diversity and immigration on public goods provision when the effects of the two phenomena are dissociated and, when public goods provision processes involve substantially differing costs.

I expect the effects of immigration and ethnic diversity on public goods provisioning to be dissimilar such that the effect of immigration trumps the effect of ethnic diversity. This is because the rate of immigration determines whether individuals have interacted for a sufficiently long period, which is a precondition for building social trust. Conversely, although co-ethnicity can also facilitate trust building either through prior relationships and/or shared norms, it is not necessarily associated with the length of previous interactions. Secondly, I expect no undermining impact of migration (on social trust and public goods provision) when the costs of provisioning processes are noticeably low. In other words, the fundamental impact of social trust on institutional processes consists in reducing transaction costs that are involved in such processes so that the undermining effect of immigration should become apparent only when notable gains can be made from minimizing the costs of public good provision.

To assess these theoretical assumptions, I consider public goods whose management involves substantially different costs, among communities with varying levels of immigration and ethnic diversity. Moving away from “Western, Educated, Industrialized, Rich and Democratic” societies that are often considered in studies about the social effects of immigration and ethnic diversity, I focus on Buvuma island, which is situated in Uganda, to analyze communal collective initiatives for cleaning up litter and constructing public toilets.

Litter cleanup and toilet construction on buvuma Island

Buvuma is located a few miles off the northern shores of Lake Victoria (See Figure 1), with roughly 200 square miles of land area, 56 communities, and 40,000 people (Timbuka, 2018). Its communities are ideal for my study because: (1) local rates of immigration and levels of ethnic diversity vary sufficiently to enable comparative analyses; (2) most local immigration results from voluntary individual (or household) decisions, which permits sidelining the effects of immigration of large social groups; (3) most immigrants did not originate from the same district, which implies they could not have developed communal relationships before arriving in the new community, and (4) the communities are small-scale neighborhoods (often with clustered housings) that can facilitate analyses within an individual’s immediate surroundings.

Intriguingly, although the communities on Buvuma island possess several prerequisites for successful collective action, such as small populations and shared languages, they are also riddled with chronic issues of poor sanitation, inadequate health care, and other manifestations of failed public goods management. These puzzling outcomes imply that the communities can facilitate analyses on the determinants of failed public goods provisioning, while at the same time enabling statistical control for certain factors, such as community size. Of the several public goods issue domains, this article focuses on efforts at litter cleanup and shared toilet construction to ensure environments that are free from litter and excreta. I focus on these two largely because their provisioning not only depends on communal collective action but also involves substantially different magnitudes of costs.⁶ Going forward, I first make a distinction that is crucial for this paper, between transit point communities and non-transit point communities.

Transit point communities

Most travelers on Buvuma island transit through landing sites in the two communities of Kirongo and Kikongo. These communities, which I refer to as transit point communities, are notorious for inadequate public health. Their empty spaces are heaped with rubbish, while discarded plastics form huge mounds in their garbage dumps. Furthermore, open defecation is the norm for most residents, and individuals who are transiting through these communities are often forced to forego toilet usage. To be sure, there exists a shared latrine facility in the Kirongo community, which was constructed four decades ago by the central government. But even that latrine facility is unusable and is long abandoned due to lack of cleaning and emptying.

Perhaps because the high mobility of populations is associated with an unwillingness to voluntarily define or support public health regulations, there have been no attempts at collective action to improve garbage collection and human waste disposal in the transit point communities. The few efforts at managing public health in these communities have always been initiated exogenously by government and non-government organizations. The fact that the transit point communities have never attempted collective action goes a long way in highlighting the challenges posed to the ideal of self-organization when members of a group are very mobile.⁷

Nevertheless, although casual littering and open defecation are common practices among communities on Buvuma island, the failure to even attempt collective action for improving public health management is limited to the above-mentioned transit point communities. Elsewhere, successful efforts are ubiquitously undertaken to encourage litter cleanup endeavors, such as sweeping streets with straw brooms and picking up stray bits of litter. Also, albeit with substantially varying levels of success, “non-transit point communities” engage in attempts at collectively contributing finances to the construction of communally shared toilet facilities.

Therefore, even if we agree that the high mobility of populations undermines collective action in the transit point communities, we still need to explain (1) the substantial variation in the levels of success concerning the willingness to contribute towards public toilet construction among non-transit point communities, and (2) why such substantial variation in the levels of success does not apply to the willingness to participate in litter cleanup programs in the same communities. The search for these explanations contextualizes my analysis of the effects of immigration and ethnic diversity on public goods provisioning in local communities.⁸

Litter cleanup

As noted above, littering is a serious menace for the transit point communities of Buvuma island. Besides being an ugly or aesthetic problem, litter is associated with many other challenges. First, it leads to soil and water pollution, especially when hazardous chemicals leach out of the litter to harm nearby soils, plants, and water bodies. Second, it propagates mosquito-borne diseases by serving as a breeding ground for mosquitoes. Third, it attracts dangerous pests and rodents. Fourth, small sparks often start a fire and release unhealthy particulate into the air. Fifth, domesticated animals sometimes die after eating litter they mistake for food.

The only government initiative for managing litter on Buvuma island was attempted in the communities of Walwanda and Kitamiro. This happened in 2016, when the central government granted a garbage truck to the Health Office at Busamuzi sub-county, intending to make local garbage collection more efficient. However, that truck was operational for only a few months, after which it has stood still due to maintenance challenges. That said, the key point here is that exogenous factors have a negligible influence on local litter and garbage waste management. It follows that to explain why all non-transit point communities are better at cleaning up litter, we need to look at endogenous processes within those communities.

Indeed, litter cleanup programs are organized in all non-transit point communities. During those programs, community members participate in collective efforts to get rid of litter in public spaces.⁹ Participation is voluntary, although one can be socially ostracized for perpetual non-participation. The cleanup programs are spearheaded by local committee leaders,¹⁰ who encourage community members to voluntarily devote 10 to 30 min each week to clean up the litter in public spaces. Such spaces are divided up and each household is allotted a particular area (usually an area adjacent to one's house).¹¹ The cleaning up includes sweeping streets, picking up trash, and taking bulky waste to a dump. The entailed costs, mostly relating to time volunteered, are substantially counter-balanced by the facts that volunteers often use the cleanup time as an opportunity to engage in valuable small talk and ‘juicy’ gossip, and there is some social prestige attached to ensuring a clean area adjacent to one’s house.

Depending on the number of community members who are willing to volunteer, the cleanup can cover a few public spaces, or it might encompass all public spaces in a community. The fact that cleanup programs usually cover most public spaces is a manifestation of a high willingness to participate. Crucially for this paper, such participation raises questions about the contextual conditions that facilitate community members’ willingness to participate.

Toilet construction

The need to construct shared toilet facilities on Buvuma island is urgent because of open defecation– a chronic sanitation problem that is linked to the transmission of cholera and diarrhea. Moreover, locals lose time while trying to access grounds to defecate (Water and Sanitation Program, 2012). The Ugandan Ministry of Water and Environment Performance Report estimated in 2012 that the island had 20%–40% toilet coverage. Nevertheless, that low toilet coverage varies considerably among the communities, which suggests potential explanations for this variation at the community level (See Appendix, Table A2, for illustration of differences in access to privately-owned toilets among the communities). Studies of the causes of persistent open defecation among Lake Victoria’s communities identify the influences of ethnic diversity and immigration (Berling et al., 2013). On Buvuma island, the impacts of migration on sanitation management are captured in the Walwanda community leader’s suggestion that "Most of these are highly migratory communities in which individuals cannot be expected to make long-term infrastructure decisions, such as investing in toilets.”

In addition, the Buvuma District authorities are too operationally weak to implement measures for stopping public health hazards (including those arising due to open defecation) that are stipulated by the National Public Health Act (2000). Not only are there scarcely any monitoring and enforcement officials (focused on averting open defecation), but also, the district lacks the finances to construct communally shared toilet facilities. No wonder, despite the acute awareness of the need for such toilets, the 2019/20 district health budget allocated $16,803 to infrastructure capital development, none of which was directed to toilet construction.¹²

Due to the above limitations, community people are frequently asked to contribute money toward the construction of shared toilet facilities. Although the required construction costs rise for communities closer to lakeshores,¹³ the intended sum for a single toilet is typically around $2,000, therefore a household in a typical town (of around 150 families) needs to give $13.4. The asking for donations is typically started when concerned locals urge their leader(s) to call a public health meeting. Contributions are rarely sufficient, a failure depicted in the following Itojwe community leader’s displeasure: “ for each of the past few years, we have had at least one (failed) attempt at collecting funds for communal toilet construction.”

The few cases of effective collective action, depicted in Table 3, depend to a large extent on having enough individuals who provide the necessary finances for the collective endeavor. The fact that efforts at soliciting finances for toilet construction are regularly repeated (despite several failed experiences) not only reveals a dire need for toilet facilities but also indicates acknowledgment that the efforts can succeed.¹⁴ Nevertheless, because most such efforts fail, the ability to compare data on communally constructed toilets is limited. This paper compares the communities’ willingness to contribute to reveal variations in the potential for collective action.

Sampling methods

Sampling the communities

From May to July (2019), I collected data from heads of households in a target sample of communities on Buvuma island, which forms the basis of my quantitative analysis. Before then I first conducted a preliminary survey in May 2018 – during which I carried out 52 face-to-face interviews with leaders of different communities. This earlier study was crucial because the immigration flows and ethnic compositions on the island had not been previously documented. I used the information gathered from the interviews to classify 40 communities by the perceived differences in migration patterns. With this classification, I randomly selected four communities from each community type to create a target sample for further analysis (see Table 1).¹⁵ Then, I further categorized the selected 12 communities by their levels of ethnic diversity.

Table 1.

Characteristics of the sampled communities.¹⁹

Community^a	Community type by rate of immigration	Main land tenure arrangement	% of permanent houses (estimate)	Number of households²⁰ (estimate)	Communally constructed public toilets^b	Access to landing site^c
Mubale	Camp	“Restricted leasehold”	0	170	0	Yes
Namagombe	Camp	“Restricted leasehold”	0	120	0	Yes
Itojwe	Camp	“Restricted leasehold”	0	130	0	Yes
Wabivu	Camp	“Restricted leasehold”	0	190	0	Yes
Kitamiro	Old village	Freehold & leasehold	80	100	1	Yes
Bukayo	Old village	Freehold & leasehold	70	80	0	No
Magyo	Old village	Freehold & leasehold	90	110	1	No
Butabula	Old village	Freehold & leasehold	85	90	0	No
Kitiko	New village	Leasehold	5	190	0	No
Kabubu	New village	Freehold & leasehold	5	90	0	No
Mayinja	New village	Leasehold	2	100	0	Yes
Bukagali	New village	Freehold & leasehold	8	120	0	No

Notes: Characteristics are based on qualitative interviews with local leaders.

^aSee the location of communities on a map of Buvuma island (Appendix Figure A1).

^bFocus is on only the functional/operating communal toilets.

^c“Access to landing site” indicates a closeness to the lake’s shores, which slightly affects the costs of toilet construction.

Classifying communities by rates of immigration

I first categorized the communities according to their previous rates of immigration. Without available data on local migration patterns, I asked each of the local leaders for descriptive information about migration flows on the island. This information was the basis for the classification that I developed. To be sure, the classification was also aided by a differentiation (among communities) that is widely made locally: when talking about the island's communities, locals usually make a distinction between ‘camps’, ‘old villages’, and ‘new villages.’¹⁶ The community types characterize local rates of immigration, which generally correspond with the longevity of the settlements, as confirmed by interview data (see Table 1).

Camps consist make-shift accommodations of mud huts and rickety shacks. Typically, a camp is originally established when a landowner leases out land to new settlers on condition that those new settlers do not erect permanent privately-owned structures on it.¹⁷ This “restricted leasehold” is associated with a migratory tendency among residents; arguably because it is a tenure system that provides little incentive to plan for permanent residency in the local community.¹⁸ In contrast, residents of villages can obtain freehold land tenure and, even when on leased land, can erect permanent privately-owned structures. This possibility of property ownership is associated with a tendency of living in one place for a long time. However, perhaps because wealth accumulation takes time, accommodation structures in villages differ depending on how old a village is: permanent structures (brick-walled and iron-roofed) are more common in villages established more than five decades ago. Such structures are rare in ‘new villages’, most of which got established during the first decade of this century–due to immigration waves triggered by factors such as rising fish prices and population pressures elsewhere.

Classifying communities by levels of ethnic diversity

Buvuma island must be the epitome of ethnic heterogeneity because its range of ethnicities is so diverse that it is often billed as the ‘United States of Uganda’ (Vision Reporter, 2009) – one of the most ethnically diverse countries in the world (Fearon, 2003). This is partly explained by the reputation of the Lake Victoria region as a place where fortunes might be made.²¹ My field study noted 345 individuals who self-identified with 31 distinct ethnic groups.²² Nevertheless, despite the diverse ethnic self-identities, there is a shared language in each community (always Luganda or Lusoga) that everyone uses to communicate outside the household. Additionally, and interestingly, most community leaders suggest that the impact of ethnic diversity on governance mechanisms is rather negligible: Only one leader (of the Bukayo community) pointed to ethnic diversity as a hindrance to institutional processes.

Nevertheless, there are significant differences in the degree of ethnic variety among the island's communities, and such differences are not due to the various immigration rates. Thus, I also asked each local leader to describe ethnic compositions of the island’s communities. Using their descriptive comparisons, I grouped the sampled 12 communities by levels of ethnic diversity, separating the highly diverse from the lowly diverse (See Table 3). All camps are highly diverse; but, while old villages tend to be more homogeneous, some– like Kitamiro and Walwanda– are highly diverse. I learned from my interviews that highly diverse old villages have been home to multiple ethnicities for several decades due to immigration waves (at distant historical junctures) that carried individuals from different ethnic groups. Elsewhere, ethnic diversity in new villages largely depends on whether most immigrants moved from areas of origin that are geographically situated in differing (or similar) ethnic groups.

Sampling the households

I used point sampling to select 30 heads of households from each community (this involved randomly marking houses on a map). When an enumerator physically arrived at a tagged house, he verified if it is a household before sampling it. He then asked to talk to the household head (If absent, he requested a phone contact). The enumerator would then introduce himself to the household head, explain the study purpose and subject recruitment process, and seek his/her consent to participate in the survey. Out of the sampled 360 household heads, 37 refused to participate in the study. Of those willing to participate, 293 actually participated.

To prevent non-response biases from producing a sample that differs significantly from the population, enumerators noted “the presence of a solar panel (an indicator of prosperity in an island not connected to the national grid)” and “the gender of the household head” for each visited household. The severity of the nonresponse issue is illustrated in Table 2. Interestingly, although men were more likely to be nonresponsive, the differences are not significant at standard levels. In general, I had no reason to think that my study’s findings would be affected negatively by nonresponse biases. Finally, to allay concerns about the higher share of men, I interviewed local leaders who agreed that that ‘share’ is reasonable.

Table 2.

Sampling characteristics.

Sample	Overall	Participated	Was willing but did not participate	Refused to participate
Solar (1= solar panel)	0.67 (0.02)	0.66 (0.02)	0.66 (0.09)	0.70 (0.08)
Share of men	0.69 (0.02)	0.67 (0.03)	0.76 (0.08)	0.78 (0.07)
Observations	360	293	30	37

Notes: Standard errors are in parentheses. The share of men is a continuous variable. Solar is measured as a binary variable with the indicated category taking the value 1, and 0 otherwise.

Survey questionnaire

Drafts of face-to-face questionnaires were pre-tested with assistants in a randomly selected community and, the irrelevant or ambiguous questions were removed or corrected to produce the final version. Besides capturing socio-economic and demographic data, the final version included items to measure support for collective action regarding litter cleanup in public spaces and contributions to public toilet construction. Each respondent was asked: (1) “To ensure public sanitation, local administrators periodically organize programs for cleaning up litter in public spaces. How often did you participate in these programs during the last 12 months?” And (2) “To avert open defecation in your community, would you agree to contribute 50,000 Uganda shillings ($13.40) to a joint effort aimed at constructing a communal toilet?”²³

The questionnaire also included questions intended to capture communities’ migration patterns and ethnolinguistic compositions (See Appendix Tables A2), such as: “Were you born in this community?”, “Why did you immigrate to this community?”, “Which ethnic group do you self-identify with?” and, “Which language do you often use when speaking to the other individuals in this community?” For locally relevant responses, interviews were conducted in the Luganda language, which is widely understood, even by non-Ganda people. It also helped that Luganda and Lusoga (the main languages on the island) are mutually intelligible.²⁴

Findings

Descriptive statistics

First, I consider the immigrants’ attributes: 97% moved from places of origin outside the island; 99% moved individually or as part of a household (only 1% moved as part of a group i.e. not a household); 92% moved to seek better economic opportunities (the rest were mostly following relatives or escaping security threats). In addition, as indicated in Table A2, the probability that two immigrants selected at random migrated from the same geographical origin (here defined as a Ugandan district) does not exceed 32.5% for any of the communities.²⁵

To determine whether rates of immigration vary as presumed by the community typology, I consider certain indicators: if a community has experienced a higher rate of immigration, one should reasonably expect that (1) more of its residents immigrated (not natives) and (2), its residents have, on the average, resided locally for a longer duration. While binary data was collected to capture immigrant status, disaggregated primary data was collected to capture ‘length of residency”. In the absence of actual data on local rates of migration, these proxies are relied on as stand-ins for the ideal, even if they are imprecise measures. Both indicators provide suggestive evidence of support for the community typology: 72% of respondents in ‘old villages’ are natives compared to only 4% and 0% for ‘camps’ and ‘new villages’, respectively. And 82% of respondents in ‘old villages’ have resided locally for 10 years or more compared to only 14% and 12% in ‘camps’ and ‘new villages’ respectively (see more data in Appendix, Table A2).²⁶ The average duration of earlier residency (years) is 32.1, 24.1, 4.6, and 4.9 from community groups A, B, C, and D respectively.

Next, to analyze whether the levels of ethnic diversity vary among the sampled communities as previously suggested, I use the Ethnolinguistic fractionalization index (ELF), which measures the probability that two people (randomly drawn from the population) are not from the same ethnic group (Alesina et al., 2003). This measure has the benefit of comparability with previous studies. ELF-based data suggests support for my expectations about the variation in ethnic diversity (see Table 3): communities grouped under (B) and (D) are relatively more heterogeneous than those under (A) and (C), despite some substantive within-group variations.

Table 3.

Ethnic diversity in the communities.

Level of ethnic diversity.

		Low	High
Rate of immigration	Low	(A=Stable Homogeneous)	(B= Stable Diverse)
		Magyo (0.480)	Kitamiro (0.806)
		Bukayo (0.589)
		Butabula (0.389)
	High	(C= Rapid change Homogeneous)	(D= Rapid change Diverse)
		Kitiko (0.480)	Wabivu (0.877)
		Mayinja (0.248)	Itojwe (0.821)
			Namugombe (0.822)
			Mubale (0.805)
			Kabubu (0.715)
			Bukagali (0.760)

Notes: The ELF is used to measure the level of ethnic diversity. Formally, ELF= 1- ∑ (Proportion of group i)².

To start off my analysis of the impacts of immigration and ethnic diversity, I compare attitudes across groups A through D. First, the frequency of participation in litter cleanup is not substantively different across the groups: that is, A (79.7%), B (78.6%), C (79.0%) and D (79.5%).²⁷ Second, the willingness to contribute to toilet construction is higher for communities in A (87.8%) and B (90.2%) than for communities in C (63.2%) and D (63.4%). Third, despite minor differences, the willingness to contribute to toilet construction is remarkably similar for communities in A and B on one hand and, communities in C and D on the other hand. This evidence supports my hypotheses that the effect of immigration on willingness to contribute to public goods trumps that of ethnic diversity, and that the effect of either immigration or ethnic diversity on such willingness is only operational when the involved costs are substantively high.

Also, the data renders support to my prior assumption about the role of trust in causally linking immigration and public goods provisioning. First, the agreement that others are trusted to participate in litter cleanup programs is not substantively different for the four groups: A (99.7%), B (99.3%), C (98.5%), and D (97.9%). Second, the agreement that others can be trusted to contribute to communal toilet construction is comparatively higher for communities in A (88.8%) and B (82.0%) than for communities in C (40.2%) and D (40.0%). Third, despite minor differences, the figures for the agreement that others can be trusted to contribute to toilet construction are remarkably close for communities in A and B on one hand and, C and D.

When other factors (which might explicate provisioning outcomes, besides ethnic diversity and immigration) are taken into account, there are no significant variation across groups, as illustrated by the close to zero p-values (See Table 4).²⁸ The proportions of residents with educated beyond primary school are not very different for communities in A (0.41), B (0.39), C (0.31), and D (0.33). Second, the proportions of residents educated beyond secondary school are not vastly different for communities in A (0.04), B (0.07), C (0.00), and D (0.03). Third, the average number of households in each community within a group is not enormously different for communities in A (93.3), B (100.0), C (145.0), and D (136.7). Fourth, proportions of farmers are not vastly different for communities in A (1.00), B (0.79), C (0.97), and D (0.84).²⁹ Fifth, proportions with access to solar electricity are not vastly different for communities in A (0.79), B (0.86), C (0.51), and D (0.61). Sixth, the average costs of toilet construction for communities in A ($1,810), B ($2,210), C ($2,000), and D ($2,163).

Table 4.

Comparing community groups.

Group	A= stable homogeneous	B= stable diverse	C= rapid change homogeneous	D= rapid change diverse	p-value
Proportion of individuals educated beyond primary school	0.41	0.39	0.31	0.33	0.000
Proportion of individuals educated beyond secondary school	0.04	0.07	0.00	0.03	0.018
Average number of households in each community within a group	93.3	100.0	145.0	136.7	0.000
Proportion of individuals in group that are farmers	1.00	0.79	0.97	0.84	0.000
Proportion of individuals in group with access to solar electricity	0.79	0.86	0.51	0.61	0.000
Average cost of toilet construction for communities in the group ($)	1,810	2,210	2,000	2,163	0.000

Predictors of individual attitudes concerning toilet construction

Given the absence of substantial variation in the above-described attitudes concerning participation in litter cleanup programs, it is intellectually uninteresting to consider the predictors of individual participation in cleanup programs. Therefore, I turn to analyze the predictors of individual attitudes toward toilet construction. In this regard, I focus on the predictors of (1) willingness to contribute to communal toilet construction and (2) trust that other community members can contribute to toilet construction. The analyses utilize ordered logistic regression analysis largely because of the ordinal nature of my Likert scale questions (see Table 9). The control variables in these multivariate analyses are considered in Table A1 (see Appendix).

The data suggest that residency in a group is a strong predictor of attitudes toward toilet construction. Relative to residency in group A, residency in groups C and D are associated with a −92.7% and −86.7% probability of contributing to toilet construction, respectively. And, relative to residency in group A, residency in groups C and D are associated with a −217% and −178% probability of trusting others to contribute to toilet construction, respectively (see Table 5). These significant coefficients indicate that irrespective of variations in the levels of ethnic diversity, communities with higher rates of immigration are both less willing to contribute to toilet construction and less trusting that others can make such contributions.

Table 5.

Predicting individual attitudes.

	Model 1Willingness to contribute to Toilet construction	Model 2Trust that others can contribute to toilet construction
Intercept	4.275*** (0.415)	5.003*** (0.353)
R Resides in a group B community	0.188 (0.218)	0.121 (0.185)
R Resides in a group C community	−0.927*** (0.222)	−2.169*** (0.189)
R Resides in a group D community	−0.867*** (0.243)	−1.782*** (0.207)
R’s community has access to a landing site	0.053 (0.129)	−0.347** (0.109)
Number of households in R’s community	0.001 (0.001)	0.001 (0.001)
Religious fractionalization in R’s community	−0.494 (0.579)	−1.663*** (0.492)
R’s age	0.001 (0.006)	0.002 (0.005)
R is male	−0.030 (0.097)	−0.104 (0.082)
R has spouse or child (ren) outside community	0.116 (0.107)	0.076 (0.091)
R’s length of stay (years) in local community	0.001 (0.001)	0.001 (0.001)
R immigrated to local community.	0.309 (0.225)	0.057 (0.192)
R is educated beyond primary school	−0.047 (0.098)	−0.111 (0.083)
R is educated beyond secondary school	1.220*** (0.271)	0.459 (0.231)
R is a farmer	−0.174 (0.155)	0.129 (0.132)
R has access to solar electricity	−0.218* (0.098)	0.001 (0.083)
Log likelihood (df)	−324.8 (17)	−277.7 (17)
Number of observations	293	293

Notes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1.

R means respondent. Cells contain coefficients from ordered logit regressions with standard errors in parentheses.

The data also suggest that relative to residency in group A, residency in group B is associated with a statistically insignificant 18.8% probability of contributing to toilet construction and a statistically insignificant 12.1% probability of trusting others to contribute to toilet construction. These coefficients suggest that, irrespective of variations in ethnic diversity, communities with analogous rates of immigration do not significantly differ in their willingness to contribute to toilet construction and their trust that others can make such contributions. In addition, although this data is not evidence for a ‘‘diversity dividend” (Gisselquist et al., 2016), it challenges the conventional wisdom about a well-established negative relationship between ethnic heterogeneity and public goods provision.

The other significant predictors of willingness to contribute to toilet construction are education beyond secondary school and access to solar electricity (which serves as an indicator of affluence/poverty): A respondent educated beyond secondary school is 122% more likely to contribute to toilet construction, and a respondent who has access to solar electricity is 21.8% less likely to contribute to toilet construction.³⁰ In addition, the other significant predictors of trust that others can contribute to toilet construction are access to a landing site (which indicates the slightly higher costs of toilet construction) and religious fractionalization. Access to a landing site is associated with a 34.7% less probability of trusting that other community members can contribute to toilet construction, and a 1% increase in the probability that two people (randomly drawn from the population) are from different religious groups is associated with a 166.3% less probability of trusting that others can contribute to toilet construction.

To examine whether these relationships are substantively meaningful, I computed the marginal effects of residency in a group on attitudes towards common property management. I used estimates in models one and two in Table 5, and held other variables at their means, to compute marginal effects (with 95% confidence interval). Movement from a group A community to a group B community is associated with a change of about −0.10 in willingness to contribute to communal toilet construction. Since willingness to contribute to communal toilet construction is measured on the 5-point scale, such movement produces virtually no difference in willingness to contribute to communal toilet construction. By comparison, movement from a group A community to groups C and D communities is associated with a change in willingness to contribute to communal toilet construction of about −1.0 and −0.9, respectively (this is more than one point on the 5-point scale).

The qualitative data acquired from interviews with local leaders supports the evidence of a causal mechanism (social trust) by which immigration lowers desire to contribute to toilet construction: 29 out of 52 local leaders brought up the subject of “trust” without being asked and invariably connected it to issues with collective action. In addition, although those the leaders’ descriptions of their experiences with raising money for toilet construction do not provide any insights that can adequately support explanations of the key findings from quantitative analysis,³¹ it is telling that the existing public toilets– which were constructed through successful collective action– are in communities with relatively lower rates of immigration, despite those communities’ substantially varying levels of ethnic diversity.

Conclusion

Utilizing primary data from one of the poorer parts of one of the poorest countries in the world, this paper explores the differential effects of ethnic diversity and immigration on public goods provision. The findings suggest that socio-political impediments to collective action for public goods provisioning have less to do with the stock of demographic diversity than the flow rate of demographic change (scholars should therefore differentiate between migration-based and other types of ethnic diversity when studying developmental outcomes). The same findings suggest a more micro-level explanation that transcends the erosive effects (of immigration and ethnic diversity) on social trust to emphasize the moderating influence of transaction costs.

To make sense of the data, I presuppose that the 'proportion of immigrants' and 'length of residency' are dependable proxy measures for the rate of immigration. In this way, I assume that there is a strong correlation between the binary data on the proportion of immigrants and the disaggregated data on length of stay. These proxies might not accurately reflect immigration rates (which are often gathered over a period of time), but I took care to ensure that my choice of measurements was not constrained by my data. Since I lacked information on real measurements, I turned to proxies as stand-ins for the ideal. Moreover, there is no need to be preoccupied with the notion that measures for rates of immigration must be precise because their power lies in the patterns of how their values change across localities, not in the individual values themselves.

That said, my analysis has certain limitations. One such limitation is the small sample size: only 12 communities were included in the analysis, and more importantly, one of the community groups (B) included just one village. This constraint complicates the process of measuring the level of interaction required (for the establishment of favorable views toward collective action) following immigration, in addition to casting doubt on the findings' validity outside of the sample. Consequently, follow-up attempts at a more thorough evaluation of the mechanisms in this study should consider larger sample sizes, in addition to 'dynamic evidence' which captures attitudes over time and so offers a more accurate assessment of immigration.

Footnotes

Acknowledgements

I am very grateful to the residents and administrative officials of Buvuma Island for their cooperation and guidance during my field research. I am also indebted to Daniel H. Cole, William K. Winecoff, Eduardo S. Brondizio, and Landon Yoder for their extensive advice and helpful comments. The data and codes (in R format) will be made available upon request from the author

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

Godfreyb Ssekajja

Notes

Data availability statement

The data that support the findings of this study are available online, please visit .

Appendix

Figure A1.

Map of Buvuma island.

Notes: The map is hand-drawn by the author. It is noteworthy that Butabula is only very close but does not touch the lake’s shoreline.

Table A1.

Community and individual-level variables.

Variables	Description	Measure/Indicator
Community-level
Community’s group	The group of a community in which a respondent is residing, as defined in Table 3.	Residency in a group A community is a referent.
Households	The number of households serves as an indicator of the size of a community	Continuous variable
Religious fractionalization	The probability that two people randomly drawn from the population are not from the same religious group	Herfindhal index Fractionalization=1- ∑ (proportion of group i)².
Access to a landing site	This indicates the higher costs of toilet construction.	Binary variable: 1 for yes & 0 for no
Individual-level
Age		Continuous variable (years)
Gender		Binary variable: 1 for men & 0 for women
Farmer	This indicates source of livelihood.	Binary variable: 1 for farmer & 0 otherwise
Access to solar electricity	This serves as an indicator of affluence/poverty	Binary variable: 1 for access to solar electricity & 0 otherwise
Education beyond secondary		Binary variable: 1 for yes & 0 for no
Education beyond primary		Binary variable: 1 for yes & 0 for no
Length of stay	Duration of previous residence in community	Continuous variable (months)
Immigrated to community		Binary variable: 1 for yes & 0 for no
Spouse or child (ren) in community (ies) elsewhere	This serves as an indicator for an individual’s level of community attachment	Binary variable: 1 for yes & 0 otherwise

Table A2.

Selected means.

	Bukayo	Kitamiro	Magyo	Butabula	Kitiko	Kabubu	Mayinja	Bukagali	Mubale	Namugombe	Itojwe	Wabivu
Age	37.69 (2.69)	36.20 (2.30)	36.48 (2.60)	35.00 (2.56)	29.20 (2.26)	31.67 (1.74)	28.82 (1.69)	29.44 (1.60)	29.92 (1.51)	31.20 (2.19)	29.85 (1.43)	28.67 (1.34)
Gender (1 = male)	0.69 (0.10)	0.65 (0.09)	0.61 (0.10)	0.68 (0.10)	0.65 (0.11)	0.54 (0.10)	0.77 (0.09)	0.76 (0.09)	0.77 (0.08)	0.58 (0.08)	0.70 (0.09)	0.70 (0.09)
Marital status (1 = married)	1.00 (0.00)	0.97 (0.03)	1.00 (0.00)	1.00 (0.00)	0.95 (0.05)	1.00 (0.00)	1.00 (0.00)	0.96 (0.04)	0.96 (0.04)	1.00 (0.00)	1.00 (0.00)	0.96 (0.01)
Beyond primary school (1 = yes)	0.39 (0.10)	0.39 (0.09)	0.48 (0.10)	0.36 (0.10)	0.30 (0.10)	0.50 (0.10)	0.33 (0.10)	0.32 (0.10)	0.35 (0.10)	0.28 (0.09)	0.26 (0.09)	0.26 (0.09)
Beyond secondary school (1 = yes)	0.00 (0.00)	0.07 (0.05)	0.13 (0.07)	0.00 (0.00)	0.00 (0.00)	0.04 (0.04)	0.00 (0.00)	0.04 (0.04)	0.00 (0.00)	0.04 (0.04)	0.00 (0.00)	0.04 (0.04)
Source of lighting (1 = solar)	0.70 (0.10)	0.86 (0.06)	0.83 (0.08)	0.86 (0.07)	0.60 (0.11)	0.54 (0.10)	0.41 (0.10)	0.56 (0.10)	0.54 (0.10)	0.80 (0.08)	0.59 (0.10)	0.63 (0.09)
Farmer (1 = Yes)	1.00 (0.00)	0.79 (0.07)	1.00 (0.00)	1.00 (0.00)	0.95 (0.05)	1.00 (0.00)	1.00 (0.00)	1.00 (0.00)	0.65 (0.09)	0.76 (0.09)	0.89 (0.06)	0.74 (0.08)
Born in-community (1 = yes)	0.74 (0.09)	0.55 (0.09)	0.83 (0.08)	0.82 (0.08)	0.00 (0.00)	0.00 (0.00)	0.00 (0.00)	0.00 (0.00)	0.07 (0.05)	0.00 (0.00)	0.04 (0.04)	0.04 (0.04)
Length of residency (years)	30.39 (3.58)	24.12 (3.61)	33.13 (3.42)	32.77 (3.17)	4.38 (0.50)	4.78 (0.80)	4.81 (0.85)	4.16 (0.60)	4.99 (1.11)	6.74 (1.76)	5.69 (1.09)	2.96 (0.47)
Length of residency-3 years + (1 = yes)	1.00 (0.00)	0.93 (0.05)	1.00 (0.00)	1.00 (0.00)	0.75 (0.10)	0.62 (0.10)	0.64 (0.10)	0.68 (0.10)	0.65 (0.10)	0.68 (0.10)	0.74 (0.09)	0.44 (0.10)
Length of residency -5 years + (1 = yes)	0.96 (0.04)	0.86 (0.06)	0.96 (0.04)	1.00 (0.00)	0.55 (0.11)	0.38 (0.10)	0.36 (0.10)	0.32 (0.09)	0.27 (0.09)	0.44 (0.10)	0.44 (0.10)	0.30 (0.09)
Length of residency- 10 years + (1 = yes)	0.87 (0.07)	0.69 (0.09)	0.87 (0.07)	0.91 (0.06)	0.00 (0.00)	0.13 (0.07)	0.23 (0.09)	0.08 (0.05)	0.11 (0.06)	0.20 (0.08)	0.22 (0.08)	0.04 (0.04)
Plan to emigrate (1 = yes)	0.04 (0.04)	0.17 (0.07)	0.04 (0.04)	0.00 (0.00)	0.15 (0.08)	0.00 (0.00)	0.04 (0.04)	0.20 (0.08)	0.77 (0.08)	0.84 (0.07)	0.89 (0.06)	0.85 (0.07)
Plan to emigrate (1 = no)	0.87 (0.08)	0.83 (0.07)	0.91 (0.07)	0.86 (0.07)	0.75 (0.10)	0.88 (0.07)	0.68 (0.10)	0.72 (0.09)	0.15 (0.07)	0.08 (0.05)	0.04 (0.04)	0.04 (0.04)
Plan to emigrate (1 = uncertain)	0.09 (0.06)	0.00 (0.00)	0.04 (0.04)	0.14 (0.07)	0.10 (0.07)	0.12 (0.07)	0.28 (0.10)	0.08 (0.06)	0.08 (0.05)	0.08 (0.05)	0.07 (0.05)	0.11 (0.06)
Willingness to contribute to toilet	4.34 (0.16)	4.51 (0.14)	4.52 (0.14)	4.45 (0.13)	3.05 (0.21)	2.83 (0.21)	3.27 (0.18)	3.28 (0.19)	3.15 (0.12)	3.04 (0.13)	3.29 (0.14)	3.41 (0.13)
Participation in cleanup programs	4.00 (0.00)	3.93 (0.05)	3.95 (0.04)	4.00 (0.00)	3.90 (0.10)	4.00 (0.00)	4.00 (0.00)	4.00 (0.00)	4.00 (0.00)	4.00 (0.00)	3.96 (0.04)	3.89 (0.06)
Trust that others can contribute to toilet	4.26 (0.14)	4.10 0.15)	4.65 (0.10)	4.41 (0.15)	1.85 (0.15)	1.96 (0.09)	2.18 (0.14)	2.52 (0.08)	2.00 (0.01)	1.80 (0.11)	1.85 (0.10)	1.89 (0.11)
Trust that others participate in cleanup	5.00 (0.00)	4.96 (0.03)	5.00 (0.00)	5.00 (0.00)	4.90 (0.01)	4.96 (0.04)	4.95 (0.05)	4.84 (0.09)	4.92 (0.05)	4.84 (0.09)	4.92 (0.05)	4.89 (0.06)
Ethnic fractionalization	0.589	0.806	0.480	0.389	0.480	0.715	0.248	0.760	0.805	0.822	0.821	0.877
Religious fractionalization	0.643	0.569	0.355	0.351	0.715	0.740	0.318	0.742	0.707	0.730	0.746	0.710
Forest area (hectares)	353	460	151	170	700	700	283	260	380	380	271	980
Probably two immigrants moved from same district	0.139	0.136	0.000	0.000	0.230	0.159	0.325	0.186	0.137	0.136	0.097	0.117
Observations	23	29	23	22	20	24	22	25	26	25	27	27

Notes: Standard errors are in parentheses. Age, previous residency, and forest area (hectares) are measured as continuous variables. Support for forestry rules, willingness to contribute to communal toilets, and trust towards others are measured as categorical variables on a 5-point Likert scale. The Herfindhal index is used to measure both ethnic and religious fractionalization. All other variables are measured as binary variables with the indicated category taking the value 1, and 0 otherwise.

References

Agrawal

(2001). Common property institutions and sustainable governance of resources. World Development 29(10): 1649–1672.

Alesina

Baqir

Easterly

(1999) Public goods and ethnic divisions. Quarterly Journal of Economics 114(4): 1243–1284.

Alesina

Devleeschauwer

Easterly

, et al. (2003). Fractionalization. Journal of Economic growth 8(2): 155–194.

Alesina

Glaeser

(2004) Fighting Poverty in the US and Europe: A World of Difference. Oxford, UK: Oxford University Press.

Allison

E. H.

Ellis

(2001) The livelihoods approach and management of small-scale fisheries. Marine policy 25(5): 377–388. doi: 10.1016/S0308-597X(01)00023-9.

Anderson

Hill

(1975) The evolution of property rights: a study of the American West. The Journal of Law and Economics 18(1): 163–179.

Berling

Mgabo

Salu

(2013) The persistence of open defecation in fishing communities of Lake Victoria; A reflection on inconsistent use of toilets in Ukerewe Island, Tanzania. International Journal of Research in Social Sciences 3(2): 293.

Bobo

Tuan

(2006) Prejudice in Politics: Group Position, Public Opinion, and the Wisconsin Treaty Rights Dispute. Cambridge, MA: Harvard University Press.

Buvuma District Hazard (2016) Risk and Vulnerability Profile. Buvuma: Buvuma District Hazard. Available at: https://www.necoc-opm.go.ug/HzCentral2/Buvuma_District_HRV_Profile.pdf

10.

Cole

Grossman

(2010) Institutions matter! why the herder problem is not a prisoner’s dilemma. Theory and Decision 69(2): 219–231.

11.

Easterly

Levine

(1997) Africa's growth tragedy: policies and ethnic divisions. Quarterly Journal of Economics 112(4): 1203–1250.

12.

Fearon

(2003) Ethnic and cultural diversity: keys to power. Journal of Economic Growth 8(2): 195–222.

13.

Fehr

Gachter

(2000) Cooperation and punishment in public goods experiments. The American Economic Review 90(4): 980–994.

14.

Gisselquist

Leiderer

Niño-Zarazúa

(2016) Ethnic heterogeneity and public goods provision in Zambia: evidence of a subnational “diversity dividend”. World Development 78: 308–323.

15.

Green

(2010) Patronage, district creation, and reform in Uganda. Studies in Comparative International Development 45: 83–103.

16.

Habyarimana

Humphreys

Posner

, et al. (2007) Why does ethnic diversity undermine public goods provision? American Political Science Review 101: 709–725.

17.

Holtug

(2010). Immigration and the politics of social cohesion. Ethnicities 10(4): 435–451.

18.

Huijts

Kraaykamp

Scheepers

(2014) Ethnic diversity and informal intra-and inter-ethnic contacts with neighbours in The Netherlands: a comparison of natives and ethnic minorities. Acta sociologica 57(1): 41–57.

19.

Jackson

(2013) Diversity and the distribution of public goods in sub-Saharan Africa. Journal of African Economies 22(3): 437–462.

20.

Khwaja

(2009) Can good projects succeed in bad communities? Journal of Public Economics 93(7-8): 899–916.

21.

Kimenyi

(2006) Ethnicity, governance and the provision of public goods. Journal of African Economies 15: 62–99

22.

Kokkonen

Esaiasson

Gilljam

(2014) Migration‐based ethnic diversity and social trust: a multilevel analysis of how country, neighbourhood and workplace diversity affects social trust in 22 countries. Scandinavian Political Studies 37(3): 263–300.

23.

Kumlin

Rothstein

(2007) Minorities and mistrust: the cushioning impact of informal social contacts and political-institutional fairness. European Consortium for Political Research Joint Session of Workshops. Helsinki: European Consortium for Political Research

24.

Kustov

Pardelli

(2018) Ethnoracial homogeneity and public outcomes: the (non) effects of diversity. American Political Science Review 112(4): 1096–1103.

25.

Lancee

Dronkers

. Ethnic diversity in neighborhoods and individual trust of immigrants and natives: a replication of Putnam (2007) in a West-European country. In: International Conference on Theoretical Perspectives on Social Cohesion and Social Capital. Brussels: Royal Flemish Academy of Belgium for Science and the Arts.

26.

Lee

(2018) Ethnic diversity and ethnic discrimination: explaining local public goods provision. Comparative Political Studies 51(10): 1351–1383.

27.

Leigh

(2006) “Trust, inequality and ethnic heterogeneity.” The Economic Record 82(258): 268–280.

28.

Letki

(2008) Does diversity erode social cohesion? Social capital and race in British neighbourhoods. Political Studies 56(1): 99–126.

29.

Miguel

Gugerty

(2005) Ethnic diversity, social sanctions, and public goods in Kenya. Journal of Public Economics 89(11-12): 2325–2368.

30.

Newman

(2013) Acculturating contexts and Anglo opposition to immigration in the United States. American Journal of Political Science 57(2): 374–390.

31.

Nunan

(2010) Mobility and fisherfolk livelihoods on Lake Victoria: implications for vulnerability and risk. Geoforum 41(5): 776–785.

32.

Olson

(1965) The Logic of Collective Action. Cambridge, MA: Harvard University Press.

33.

Ostrom

(2010) Analyzing collective action. Agricultural Economics 41: 155–166.

34.

Ostrom

Gardner

Walker

, et al. (1994) Walker, and Jimmy Walker. Rules, Games, and Common-Pool Resources. Ann Arbor, MI: University of Michigan Press.

35.

Putnam

(2007) E pluribus unum: diversity and community in the twenty‐first century the 2006 Johan Skytte Prize Lecture. Scandinavian Political Studies 30(2): 137–174.

36.

Rafa

Moyer

Wang

, et al. (2017) Estimating District GDP in Uganda. Denver, CO: Josef Korbel School of International Studies, University of Denver

37.

Vision Reporter (2009) Buvuma the United States of Uganda. Kampala: New Vision. Available at: https://www.newvision.co.ug/new_vision/news/1221327/buvuma-united-Uganda

38.

Savelkoul

Gesthuizen

Scheepers

(2015) Associational involvement in Dutch municipalities and neighbourhoods: does ethnic diversity influence bonding and bridging involvement? Urban Studies Research 2015: 1–15.

39.

Silsbe

Hecky

(2008) Are the Lake Victoria fisheries threatened by exploitation or eutrophication? Towards an ecosystem-based approach to management. The Ecosystem Approach to Fisheries. Rome, Italy: FAO, p. 309.

40.

Singh

Vom Hau

(2016) Ethnicity in time: politics, history, and the relationship between ethnic diversity and public goods provision. Comparative Political Studies 49(10): 1303–1340.

41.

Ssekajja

(2021) Commons management in migrant communities. International Journal of the Commons 15: 132–153.

42.

Ssekajja

(2023) Can immigration explain why ethnic diversity is negatively associated with common property management? Journal of Development Studies 59(5): 653–672.

43.

Taylor

(1998) How white attitudes vary with the racial composition of local populations: numbers count. American Sociological Review 63: 512–535.

44.

Timbuka

(2018) Assessing the Impacts of Oil Palm Expansion on Ecosystem Services Under Different Scenarios, a Case Study of Buvuma Island, Uganda. Master’s Thesis. Enschede, Netherlands: University of Twente.

45.

Uganda National Household Survey (2017) Uganda Bureau of Statistics. Uganda: Uganda National Household Survey. Available at: https://www.ubos.org/onlinefiles/uploads/ubos/pdf_documents/UNHS_VI_2017_Version_I__27th_September_2017.pdf

46.

Meer

Tolsma

(2014) Ethnic diversity and its effects on social cohesion. Annual Review of Sociology 40(1): 459–478.

47.

Water and Sanitation Program (World Bank) Africa region (2012). Economic Impacts of Poor Sanitation in Africa. Montreal: WSP, Water and Sanitation Program. Available at: https://www.zaragoza.es/contenidos/medioambiente/onu/825-eng-v14.pdf

48.

Wimmer

(2016) Andreas. “Is diversity detrimental? Ethnic fractionalization, public goods provision, and the historical legacies of stateness. Comparative Political Studies 49(11): 1407–1445.

49.

Yamagishi

(1986) The provision of a sanctioning system as a public good. Journal of Personality and Social Psychology 51(1): 110–116.