Minority size and socio-economic inequalities: A case study of Muslim minority in India

Abstract

Spanish

French

The authors investigate the relationship between Muslim minority size and inequality using a new cross-sectional dataset of 599 Indian districts. The authors review existing studies, propose a new population growth inequality theory, and develop three hypotheses. A newly constructed multidimensional index of social well-being is used as a proxy for intergroup inequality. A multi-level mixed effects regression analysis with controls for urbanization and state-level effects is applied. The authors find a U-shaped relationship between the size of the Muslim minority and its absolute and relative well-being. Well-being reaches the lowest point when minority reaches approximately 50% of the population in a district. The average gap in well-being tends to be larger in the districts with lower socio-economic development.

Keywords

India inequality minority size population growth well-being

Introduction

Over the next three decades the demography of the world religions will change markedly. The proportions of various religions in the world population will remain the same or decline except for Islam. Because of comparatively higher fertility and younger age structure the Muslim population is expected to increase from 1.6 billion or 23% to 2.76 billion or 30% of the world population in 2050. One of the most consequential impacts of these trends will be in India. India’s Muslims will grow by 76% from 176 to 310 million in the same period, which will make India the largest Muslim ‘country’ in the world.

The rate of increase of the Muslim population in India is frequently attributed to Islam. However, religious differences do not have a statistically significant effect on fertility if other factors such as access to education, income, local health care services and the degree of urbanization of the community are taken into account. The evidence also shows that two main determinants of fertility –age at marriage and contraceptive choice – are due to differences in socio-economic characteristics and not religion (Iyer, 2002; Iyer and Shrivastava, 2018). Other factors contributing to relatively higher fertility among Indian Muslims include their relatively younger age structure, lower child mortality and greater life expectancy (Pew Research Center, 2015).

The population increase will pose developmental challenges related to provisions of public services and survival goods in India. The economic development over the last three decades should ameliorate intergroup socio-economic inequalities in India. However, the vast majority of Indians have not benefited from this development due to the absence of institutional means to promote inclusive economic development. The dissonance between growth and well-being is especially pronounced among India’s socially marginalized minority groups, including Scheduled Castes, Scheduled Tribes, and Muslims, who face multiple discrimination and exclusion due to systemic bias in their case. Inclusive development in India needs to address the imbalance of political and economic power through greater public engagement with demands of justice and development of socially marginalized groups, which make up more than half of India’s population (Drèze and Sen, 2013).

Indian Muslims have not been equal beneficiaries of its economic growth and even suffered downward mobility since 1950s (Basant and Shariff, 2010; Hasan and Hasan, 2013; Sachar et al., 2006). A growing population of Indian Muslims will further exacerbate their socio-economic conditions which may undermine India’s political stability. A long history of social conflict between Hindus and Muslims in India calls for systematic studies of reproduction of social inequities among India’s growing Muslim minority. Importantly, the trigger of frequently violent Hindu–Muslim conflicts is not religion but a multi-causal phenomenon emerging from a context of social tensions, historical distortions, and state-sponsored actions against ethnic and religious minorities for political and especially electoral benefit (Brass, 2003; Dhattiwala and Biggs, 2012; Varshney, 2003).

The main objective of this article is the empirical examination of the relationship between the minority size and the level of inequality.¹ Minority group in this article is conceptualized not purely as demographic but as a distinct cultural, religious and ethnic group which coexists with a dominant group as a subordinate. The subordinate status invariably affects the social mobility and life chances of the members of minority groups. This relationship represents a critical lens through which a range of social outcomes can be further examined. This relationship has a dynamic nature because the rate of population growth of minorities is different – and often greater – than that of the majority population. As the relative size of the minority population increases, it acts as a catalyst in increasing deprivations, social conflict, and competition over resources.

Existing sociological theories lack the flexibility to fully accommodate the dynamics of the population growth in the relationship between the minority size and inequality. A central question – what happens when the minority becomes the population majority – remains inadequately addressed. Consequently, existing empirical studies have produced models that only partially account for the observed relationships and lack a deeper understanding of the underlying dynamics. This article contributes to the existing theoretical and empirical body of literature by conducting exploratory and explanatory analyses of a new district-level dataset from India. Another important contribution of this article is in applying previous theories on minority size and inequality outside of the developed countries.

India is a unique and important case in examining how this dynamic unfolds. Since its independence in 1947, India has experienced a long social conflict between Hindus and Muslims with a presently high level of religious hostilities (Brass, 2003; Pew Research Center, 2015; Varshney, 2003). There are also current signs of growing inequality and increasing relative deprivation that are developing in some socio-religious communities (SRCs). This development is concerning if we apply the theoretical framework of relative deprivation to make future projections of social outcomes such as social unrest.

As a country with the largest Muslim minority population (and the third largest total Muslim population in the world), India represents a quasi-laboratory for examining intergroup inequality. In addition, the Hindu majority in India is hierarchically stratified along caste lines that have identifiable boundaries. Empirically, these castes can approximate the class structure. A recent surge in studies in the links between economic inequality, religion, and social conflict makes India an ideal place for researching these relationships.

This article is organized by first discussing the broader questions of the relationship between minority size and inequality. Then we describe India and the Indian districts as the sample chosen for this analysis and develop three hypotheses. It is important to note at the onset that even though this article tests hypotheses, the main contribution of this analysis is exploratory rather than explanatory. We then present a new dataset that has not been used previously and we discuss these data and their sources. This section is followed by a discussion of the measurements and methods. Next, we conduct a multi-level mixed effects regression analysis and discuss the results. In conclusion we review the main findings in a broader theoretical context and offer a new theoretical approach that complements existing theories.

Minority size and inequality

The topic of inequality and minorities is well studied in the social sciences. Past studies show that minorities have lower incomes (Kahanec, 2006; Tienda and Lii, 1987), higher poverty rates (Massey and Eggers, 1990), worse health (Smedley et al., 2002; Williams and Jackson, 2005), and increased mortality rates (McLaughlin and Stokes, 2002) compared to the majority population. Some of the reasons why minorities fare worse are attributed to the lower level of social capital of new immigrant minorities (Kurthen and Heisler, 2009). Social capital is also weaker for the domestic minorities, which results in lower earnings (O’Neill, 1990). Larger families and fewer income earners per family also place minorities at a disadvantage (Gradín, 2012).

However, discrimination is the main factor behind the lower social well-being of minorities. Such discrimination manifests in restricted labor markets and discriminatory wages (Becker, 1971), systematic exclusion or under-representation of minorities in political voting (Norris, 2004), limited access to health insurance (Doty and Holmgren, 2006), lower quality of health care (Watson, 1994), and residential segregation (Semyonov and Glikman, 2009). Discrimination limits the minorities’ ability to acquire social capital, employment, and political power. It also inhibits social contacts between the minority and the majority population and preserves the status quo of the limited participation of the minorities in social life. In these and other studies, discrimination is a general mechanism linking minority and inequality.

Identifying the extent of the discrimination, as well as its specific mechanisms, is a central goal of inequality research. However, this analysis alone cannot address a pressing question of how these dynamics change when demographic shifts occur and the minorities grow in absolute and, more importantly, in relative size to the majority population.

Another well-established body of literature looks at the relationship between the size of the minority and inequality (Albrecht et al., 2005; Blalock, 1956; Cohen, 1998; Frisbie and Neidert, 1977; Kahanec, 2006; Quillian, 1995; Rapoport and Weiss, 2003; Tienda and Lii, 1987; Williams, 1947). The general finding is an increase in the level of inequality, grievances, and deprivations when the minority grows in its relative size.

Several theories link the growing minority population to increasing inequalities. The foundational group position theory of prejudice (Blumer, 1958) proposes that there is a resource-based conflict between the majority and the minority groups. As the majority perceives an increasing threat from the minority over the control of resources, the level of prejudice (a form of a defense mechanism) also increases. The majority then proceeds to counter and minimize the perceived threat using the available structural mechanisms that this majority controls.

Similarly, power-threat theory (Blalock, 1967) states that the minority size influences the unobserved level of discrimination that arises because the majority perceives the minority as a threat in the competition for economic and political resources. The majority places various restrictions on the minority group which increases the inequality.

Intergroup threat theory (Stephan et al., 2009) echoes Blalock’s theory and adds that the history of conflict (something Blumer also touched upon) is important in forming the intergroup perceptions of threat. Such perceptions would be the strongest between two groups of similar size, power, and with a long history of conflict.

Other theories such as the integrated threat theory of prejudice (Stephan and Stephan, 2000) advocate that intergroup attitudes are formed by the amount and the quality of the intergroup contact, group status, identification, knowledge, and conflict. These relationships are mediated by realistic and symbolic threats, intergroup anxiety, and negative stereotypes.

Quillian (1995) makes an important expansion to Blumer’s theory by adding Blalock’s theory and developing his own group-threat theory. He argues that the perception of threat by the majority (which in turn determines the level of prejudice) intensifies when the relative size of the minority increases and especially when economic conditions worsen.

Blumer’s theory was further analyzed from the context of a minority (Bobo, 1999; Bobo and Hutchings, 1996) wherein the feeling of oppression and grievances by a racial group increases the perception of threat towards other majority and minority groups. Combining this idea with Quillian’s group-threat theory we can anticipate a continuous downward trend in the extent of social inequalities: as the minority grows, the majority increasingly perceives it as a threat and acts to secure greater control over economic and political resources, thus limiting the minority’s access to these resources. These actions are perceived as hostile by the minority, who begins to see the majority as a threat. That perception begins another iteration of this downward spiral: the majority considers the antagonistic feelings of the minority as an additional confirmation of a threat and further extends its control over the resources and increases deprivations. Ultimately, this downward trend will result in growing social inequalities and the escalation of social conflict. In the end, we see three possible outcomes: broad political reforms and reconciliation, civil war, or genocide.

Figure 1 presents a summative theoretical model of the relationship between minority size and inequality.

Figure 1.

Diagram of the theoretical relationship between minority size and inequality.

A shortcoming of the existing theories is that they consider the relationship between the relative size of a minority and the extent of its deprivations/inequalities as mostly linear. Blalock (1957) considered non-linearity in his study of the percentage of the Black population and inequality, however the results were significant only for very small minorities (less than 5%).² But what happens when the minority surpasses 50% of the population? History shows that it is possible to be a population majority while still being an economic minority (e.g. the Sunni and Shia in Iraq during the Saddam Hussein rule and Black and White populations in South Africa during apartheid). Existing empirical studies apply these theories largely without considering the potential change in the dynamic of this relationship when the minority surpasses 50% of the population. Studies that include minority proportions that surpass 0.5 typically model only a linear effect of the minority size (e.g. Cohen, 1998; McLaughlin and Stokes, 2002).

Why should we expect a continuing widening of the gap as the minority size increases? One could argue that regardless of the minority size, the power and resources are clearly controlled by the former majority group. Just because the former minority has reached 50% population (or has always been the majority) does not mean that it suddenly acquires the resources. South Africa during apartheid would be one such example.

However, a plausible argument can be made that 50% does matter. While the figure of 50% likely carries only a symbolic meaning, in practice a growing minority is more capable of influencing the political process through elections, accumulation of more resources, greater penetration into the better-paying levels of the labor market, and occupying important positions in society at a higher rate than previously. A formation of the minority middle class would allow pooling social and economic resources and influencing the political process. Occupying more government jobs would also allow the minority to influence policy creation and execution. Overall, it is reasonable to expect that a gradual process of political, economic, and social change would bring more resources under the control of the former minority and reduce or altogether eliminate the inequality gap.

But that would mean that there is an inverted U-shaped curve in the relationship between the relative size of a minority and the level of inequality. At the beginning, when the size of the minority is small, the inequality may be small as well. As the minority increases in size and becomes more visible in social and economic life, the majority perceives it as a greater threat and reacts by restricting the resources that go to the minority thus increasing the level of inequality. As the minority becomes the majority (through immigration or higher fertility) it slowly begins to recover the lost resources through political, economic, and social means. To our knowledge no theoretical or empirical study has considered this non-linear relationship. This new theoretical direction might be called population growth inequality theory. We will address this proposition in our hypotheses in the following section.

Indian districts as the sample

Quantitative studies of minorities have largely focused on the USA, with Blacks and Hispanics being the most researched examples of intergroup discrimination. Cross-national analyses of minorities mostly considered European or OECD countries because of the availability of the data. Indeed, Figure 1 shows that there are a number of variables that directly influence or mediate the relationship between the minority size and the extent of the inequalities.

Even in developed countries, however, some variables are hard to measure and the data are spotty at best. For example, Markert (2010) emphasizes that the media coverage is important in shaping the minority–majority relationship; yet he merely adapts the measurement from a previous study of Lee (1998).³ In developing countries much of the relevant data are non-existent or publicly not available. As a result, the theories explaining social inequality of the minorities are inadequately tested outside the developed nations.

In our study we use a new dataset from India to explore the relationship between the size of the minority and the extent of the social inequalities to test several hypotheses based on existing theories, and to suggest a new theoretical approach.

The effects of India’s economic and social development have not been equally distributed among different groups. Specifically, Indian Muslims have benefited relatively less than most other groups due to discrimination as well as different structural position. Consequently, over the past 60 years Indian Muslims have experienced downward mobility and faced economic deprivations, social exclusion, and political under-representation (Basant and Shariff, 2010; Hasan and Hasan, 2013).

The Sachar Commission report (Sachar et al., 2006) as well as the National Household Surveys conducted by the National Sample Survey Organization indicate that on average Muslims have lower literacy rates, lower consumption levels, higher poverty rates, worse access to health care and education, and lower quality of transportation than the general population. Spatially, villages with large Muslim populations tend to cluster in states with poor infrastructure levels. Muslims’ share in public sector employment is significantly lower than their proportion in the population. In 12 states⁴ Muslims comprise about 15.4% of the population but their share in state employment is only 6.3%. Across all important measures, Indian Muslims fared similarly or slightly better off than the lower caste of Hindus (Hindu SC-ST), and worse off than the upper caste of Hindus (Hindu General), Other Backward Castes (Hindu OBC) as well as other minorities.⁵

Examining Indian districts offers several advantages over the past studies. Muslim minorities in India offer an opportunity to test the theories regarding the relationship between minority size and inequality. Many past studies examined societies where minorities had small relative sizes. Other studies often employed the percentage of foreign-born citizens as the measure of a minority, which would artificially elevate the percentage of minorities in countries such as Australia where 28.5% of the overseas-born Australians come from the UK and New Zealand – the countries with similar ethnic composition, religion, and language. Counting minorities based on ethnicity or religion is therefore a more accurate measure of heterogeneity.

There is also an advantage in working with smaller units such as districts. Nearly all cross-national and state-level studies effectively hide the intra-state dynamics that occur on the smaller level. For example, Jefferson county in Mississippi has 86.5% Black population while Mississippi has 37.3% and the US has only 13.2% Black population. While cross-country comparisons can identify useful general patterns, intra-country analysis based on the smallest administrative units can potentially yield a much more detailed picture.

The previously unanalyzed district level data from India combined with a new theoretical approach that considers non-linear dynamics will provide fresh insights into the relationship between minority size and inequality. For the purpose of this analysis, we define inequality as an absolute difference in the level of social well-being between different groups. Based on the theoretical expectations discussed above, we develop three hypotheses:

H1: There is a general negative relationship between the percentage of the Muslim population and its well-being. A higher percentage of Muslims in an Indian district will reduce the value of their well-being.

H2: There is an inverse U-shaped relationship between the percentage of the Muslim population and its well-being. A higher percentage of Muslims in a district will result in the reduction of their well-being followed by an increase after the Muslim population reaches about 50%.

H3: The well-being of Muslim minorities is higher in urban areas.

Although Hypotheses 1 and 2 may appear contradictory, both relationships can be present: a negative linear relationship and a positive quadratic relationship. In effect, Hypothesis 1 tests previously discussed theories and Hypothesis 2 tests a new population growth inequality theory that we propose.⁶ Hypothesis 3 is based on the Sachar Commission findings and other previous studies that suggest Muslims in India are more concentrated in areas with worse infrastructure, services, and employment opportunities, which are rural areas.

Data and methods

Much of the past research in minority size and inequality focused on the US states or counties and on a cross-national comparison of developed nations. Indian development studies nearly exclusively focused on the states because of the availability of state-level data. In this article we use district-level data from the dataset developed by the US-India Policy Institute (USIPI) in Washington, DC, and the Center for Research and Debates in Development Policy (CRDDP) in New Delhi.⁷

The dataset is constructed using two sources: the 68th round of the National Sample Survey (NSS) conducted in 2011–2012 and round 3 of the District Level Household and Facility Survey (DLHS) conducted in 2007–2008. Because of the slow yearly change in most measures, the four-year difference between the two surveys will have minimum distortion and the compatibility on the district level is high. The correlations of the different measures are high in value and are strongly significant, thus increasing the external validity of each component. The combined dataset has four indexes that are listed below with their sub-components:

- Economic development index (Monthly per capita expenditure, % population above poverty line, % of the households with a salary-earning member);

- Material well-being index (Household asset index, % of households using liquid propane as the primary source for cooking, % of the households using electricity as the primary source of lighting);

- Educational development index (% female literacy 7+ years old, % adult literacy, % graduated from secondary school, % of annual household expenditure on education);

- Health index (% women married at 18 or older, % women received postnatal care, % fully immunized last two children, % women using contraception, % women aware of HIV/AIDS).

Economic, material well-being, and educational development indexes are derived from the NSS data. The health index is derived from the DLHS data. Out of 640 Indian districts, the DLHS data are limited to 599 districts, thus the complete dataset contains 599 districts. The household-level data from the surveys are aggregated to the district-level. The four components are averaged to produce a combined District Development and Diversity index.⁸ Each measure is recorded for the five SRCs (Muslims, Hindu-General, Hindu OBC, Hindu SC-ST, and other minorities), which allows for a cross-SRC comparison on a district level. For complete methodology see Shariff (2015). The advantage in using the combined index is that it taps into different dimensions of socio-economic well-being theoretically aligning with different aspects of inequality. Methodologically, the distribution of the overall index approximates the normal curve (min. 0.14, max. 0.7, mean 0.38, SD 0.1) which is advantageous in regression modeling.

The dataset is cross-sectional and all conclusions are drawn based on a comparison of the districts, not a comparison over time. Nonetheless, controlling for the district-specific effects and urbanization will allow to some extent to extrapolate the cross-sectional comparison and make an argument on what may occur when the size of the Muslim minority increases. In the following section we apply multi-level mixed effects regression with a control for urbanization to estimate the relationship between the size of the Muslim minority and the level of inequality that it experiences. It is important to note that while our theory is about inequality, the empirical measurement of inequality is the combined index of well-being.

Analysis and discussion

We begin our analysis with the examination of a scatter plot of the Muslim population percentage and the value of the overall well-being index for Muslims presented in Figure 2. The scatter plot features linear, quadratic, and cubic (up to 50% value of the Muslim population) regression lines.

Figure 2.

Scatter plot of the Muslim population percentage and the overall social index for Muslims.

In general, there is a weak negative linear association between the percentage of the Muslim population and the level of the overall index for Muslims. Visually, two trajectories can be identified in the scatter plot. First, as the Muslim population increases, there is a decrease in the overall value of the well-being index of Muslims. That is, when the minority increases in relative size, their average well-being decreases in absolute terms.

This decrease continues until the Muslim population reaches approximately 50%. At that point, the trajectory changes to become a positive association between the Muslim population percentage and their overall index. The trajectory ends with a cluster of seven districts that have above 90% Muslim populations. These districts have above average index values.

Table 1 summarizes the results of a simple regression of the overall well-being index for Muslims on the percentage of the Muslim population. It is important to note that the structure of the data is multi-level – the district observations are clustered within the states. All districts in the states will share some similar state-level characteristics such as the distribution of the SRCs, legislative framework, budgets, political arrangements, social services, educational structure and appropriations, etc. It is therefore reasonable to expect that the district-level observations within a single state would be correlated in some way. Under such conditions the errors need to be corrected for the within-state correlation to reduce the likelihood of false significant results. All models are therefore estimated with a correction for the state-level clustering. This correction also produces the standard errors that are robust against heteroskedasticity.

Table 1.

Regression of the overall well-being index for Muslims on the percentage of the Muslim population.

	(1)	(2)	(3)	(4)	(5)	(6)
	0–100%	0–50%	50–100%	0–15%	15–35%	35–50%
% Muslim	−0.153	−0.42***	0.42***	−0.62***	−0.37	−0.97**
	(0.09)	(0.06)	(0.09)	(0.16)	(0.27)	(0.25)
Constant	41.7***	44.4***	4.7	45.6***	42.8***	70.4
	(2.2)	(1.9)	(8.5)	(1.4)	(7.1)	(10)
N	477	456	21	328	106	22
R²	0.06	0.16	0.51	0.06	0.03	0.29
F	2.7	43.1	20.6	14.46	1.86	15.5

Note: In order to directly correspond to Figure 2, the models in this table do not include a control for urbanization. No overlap occurs at 15%, 35%, and 50% points as no single district has this exact percentage of the Muslim population. The whole percentages in the interval labeling are employed for convenience only. Standard errors in parentheses.

p < .01, ***p < .001, two-tailed tests.

The overall relationship between the percentage of the Muslim population and its index of well-being is weak and non-significant (Model 1 in Table 1). Based on Figure 2 and Models 2–3 in Table 1, it is evident that there is a decrease and an increase in the values of the index as the size of the Muslim population increases. Interestingly, the coefficient value is very similar (the difference is in the rounding). A U-shaped quadratic model should fit the data better than a linear model. However, the relationship between the size of the Muslim population and their overall social index may be more complex. Models 4–6 in Table 1 show that on the interval 0–50% the effect is negative, then non-significant, then again negative. This dynamic of the effect – negative, non-significant, negative – suggests that a more complex curve such as cubic relationship may be more appropriate to fit the data in the districts with less than 50% Muslim population. It is important to note that the three intervals in the range 0–50% in Table 1 were created for illustration purposes only. The percentage of the Muslim population is a continuous variable and other intervals may as well be employed.

Figure 2 shows a part of the cubic fit curve for the interval 0–50% of the Muslim population. The cubic curve closely follows the quadratic curve and begins to diverge only after the Muslim population reaches about 40% with the main difference between 40% and 50%. While this difference may not appear substantial, it is worth exploring a cubic relationship in addition to a quadratic in modeling of this dynamic. This cubic curve indicates that when the Muslim minority approaches 50% there appears to be an exponential decrease in their well-being. Perhaps as the minority approaches 50% and is increasingly able to influence policy, resource allocation, and social life the majority perceives it as more of a threat and escalates a discriminatory action against the minority resulting in an accelerated reduction of the well-being of the latter.

Overall, Figure 2 and Models 1–3 in Table 1 show that there are two opposite trajectories: negative in the districts with less than 50% Muslim population and positive in the districts with more than 50% Muslims. The first part is consistent with the expectation of Hypothesis 1: as the Muslim minority population increases, it begins to challenge the majority population for redistribution of economic, political, and social resources. As a result, this struggle further alienates the minority from the majority and increases the perception of the threat by the majority. The potential cubic decrease in the well-being may indicate that this relationship is non-linear: an increase in minority size after a certain point results in the exponential decrease in its well-being.

The second part (Muslim population > 50%) is consistent with the expectations of Hypothesis 2. When the minority becomes the majority population, it begins to produce a greater influence over the redistribution of resources. As the Muslim population moves from 50% to 100%, the well-being of Muslims gradually increases. Importantly, our models do not show why this increase in well-being occurs. Factors such as greater availability of community support are theoretically important.

A basic regression of any social index variable on the percentage of the Muslim population likely suffers from omitted variable bias. To control for the level of development in a district we included a measure of urbanization (percentage of population living in urban areas). Urbanization can also serve as a proxy for different relevant variables that can influence the average socio-economic index values (ease of access to schools and hospitals, level of infrastructure, extent of industry presence, structure of the labor market, class structure, etc.), which reduces the omitted variable bias. Although it may be desirable to include a broader range of control variables, the limitations of the existing data prevent us from including additional variables that may be relevant to this analysis. Some comfort may be taken in looking at the past studies where the authors found that including more than two control variables does little to improve the models because of the high collinearity of socio-economic variables (Blalock, 1957). In addition, many important district-level variables are typically correlated on a higher state level. We model state-level effects by including state intercepts and allowing the slopes to vary between the states. Although this is not a direct substitution for the relevant control variables, we believe that this model specification is fairly robust against the omitted variable bias.⁹

Table 2 shows the regression output from a complete specification of the regression model: y_i = b_o+b₁x₁+b₂x₂+b₃x₁²+b₄x₁³+e, where b_o is intercept, x₁ is the percentage of the Muslim population, x₂ is the percentage of the urban population, and e is the error. The models employ a correction for the clustering of the standard errors of the districts within the states.

Table 2.

Regression of the overall index (Muslims) on the percentage of the Muslim population and urbanization.

	Model group A (Muslim %, 0–100)				Model group B (Muslim %, 0–50)				Model group C (Muslim %, 50–100)
	(7)	(8)	(9)	(10)	(11)	(12)	(13)	(14)	(15)	(16)	(17)	(18)
% Muslims	−0.15	−0.12	−0.65***	−0.9***	−0.42***	−0.39***	−0.81***	−1.25***	0.42**	0.37**	0.68	10.9
	(0.09)	(0.09)	(0.09)	(0.18)	(0.06)	(0.06)	(0.14)	(0.3)	(0.09)	(0.08)	(1.74)	(12.8)
% Urban		0.27***	0.27***	0.28***		0.27***	0.28***	0.28***		0.15***	0.15***	0.17***
		(0.05)	(0.04)	(0.04)		(0.04)	(0.04)	(0.04)		(0.02)	(0.02)	(0.02)
% Muslims^2			0.007***	0.016**			0.011***	0.04*			−0.002	−0.14
			(0.001)	(0.005)			(0.003)	(0.02)			(0.01)	(0.16)
% Muslims^3				−0.00007*				−0.0004*				0.001
				(0.00003)				(0.0002)				(0.001)
Constant	41.7***	34.7***	38.6***	39.5***	44.0***	37.4***	39.1***	40.3***	4.7	6.4	−5.2	−257.1
	(2.2)	(2.6)	(2.0)	(1.8)	(1.9)	(2.3)	(1.9)	(1.5)	(8.5)	(8.1)	(64.0)	(336.8)
N	477	477	477	477	456	456	456	456	21	21	21	21
R²	0.06	0.28	0.403	0.41	0.16	0.38	0.404	0.41	0.51	0.57	0.57	0.59
F	2.7	18.2	34.5	22.6	43.1	38.9	24.5	27.0	20.1	25.5	22.6	22.5

Note: Standard errors in parentheses.

p < .05, **p < .01, ***p < .001, two-tailed tests.

The results are consistent with the expectations based on Hypothesis 2: the average well-being of the Muslim population in a district depends on the percentage of Muslims in the district. This relationship has an evident U-shape – the quadratic coefficient is consistently positive. There is a potential for a more complex relationship, which is indicated by a significant cubed coefficient of the Muslim population percentage, however this relationship is not universal across the entire range of the values of the Muslim population.

Because of their small values, cubed coefficients serve as mediators for the dominant quadratic curve. In Model 10 the quadratic effect at the maximum value of 100% Muslim population adds 162 points to the index scale, while the cubic effect subtracts 70 points. Therefore the cubic effect mitigates the rate of the increase, but does not reverse the positive trend of the U-shaped curve within the limits of the realistic values. However, for the 0–50% range of Muslim population values, the cubed coefficient is relatively stronger. In Model 14 at the maximum 50% of Muslims the quadratic term adds 99 points while cubic term subtracts 50 points.

It is important to note that within the range of 0–100% of the Muslim population, the combined effect (linear, quadratic, and cubed) of the percentage of Muslims is negative except above 93%. This means that, on average, any size of the Muslim populations except those above 93% results in lower average index values for Muslims as compared to the districts with no Muslim population. The fully saturated Model 10 in Table 2 produces a combined effect of positive 2 (–90 + 162 – 70) at the maximum value of the Muslim population percentage. Therefore, while the overall relationship is non-linear, in the case of Muslims in India virtually any increase in the value of the Muslim population results in lower well-being index values. However, there are relatively few districts with more than 50% of Muslim population and the regression model is dominated by the values below 50%. As the Muslim population grows in relative terms, the U-shaped curve should become more prominent.

To further explore the dynamics of the relationship between the percentage of Muslims in a district and the overall social index for Muslims, we estimate a series of regressions with random state-level intercepts and random state-level slopes. In group D models in Table 3 the intercepts are allowed to vary for each state. In group E in addition to the random intercepts, the slopes for each state are allowed to vary as well. Arguably, model group E in Table 3 represents the most dynamic models that more accurately account for the state-level effects.¹⁰

Table 3.

Random intercept and random slope regression of the overall index (Muslims) on the percentage of the Muslim population and urbanization.

	Model group D(random intercept, Muslim %, 0–100)				Model group E(random slope, Muslim %, 0–100)				Model group F(random slope, Muslim %, 0-50)
	(19)	(20)	(21)	(22)	(23)	(24)	(25)	(26)	(27)
% Muslims	−0.17*	−0.17*	−0.49***	−0.66***	−0.55***	−0.6***	−0.73***	−0.97***	−1.35***
	(0.06)	(0.06)	(0.08)	(0.15)	(0.00)	(0.01)	(0.05)	(0.08)	(0.24)
% Urban		0.15***	0.17***	0.17***		0.16***	0.16***	0.17***	0.18***
		(0.03)	(0.03)	(0.03)		(0.03)	(0.03)	(0.03)	(0.03)
% Muslims^2			0.005***	0.011*			0.004**	0.016***	0.04**
			(0.0008)	(0.004)			(0.0016)	(0.003)	(0.01)
% Muslims^3				−0.00005⁺				−0.0001***	−0.0005*
				(0.00003)				(0.00002)	(0.0002)
Constant	46.96***	42.4***	44.4***	45.1***	50.5***	46.3***	46.8***	47.7***	48.9***
	(0.6)	(1.01)	(1.08)	(1.05)	(0.00)	(0.9)	(0.8)	(0.8)	(0.8)
N	477	477	477	477	477	477	477	477	456
R²	0.57	0.62	0.65	0.65	0.63	0.67	0.68	0.69	0.68

Note: The coefficients of random intercepts and slopes are not shown, standard errors in parentheses.

p < .1, *p < .05, **p < .01, ***p < .001, two-tailed tests.

Models in groups D and E in Table 3 are estimated for the full range of the values of the Muslim population. Model group F is estimated for the interval 0–50%. However, the estimates from model group F should be approached with caution. Estimation for the partial range may produce biased estimates because the state-level intercepts and the slopes will be computed based on the observation of only some of the districts in a state, which will depend on the level of the Muslim population. Therefore, the selected level of the Muslim population may directly influence the coefficients and the errors. Models in Table 2 are free from this potential bias and therefore allow analyzing different ranges of the values of the percentage of the Muslim population. The similarity in the coefficient values between Model 14 in Table 2 and Model 27 in Table 3 suggests that this potential problem is limited. Additionally, only 10 states have districts with more than 50% Muslims (and 9 out of 10 have only a single district in this category).

Accounting for the individual state-level intercepts and slopes improves the precision of the estimates. However, both model groups A and B in Table 3 are generally consistent with the results from group A in Table 2. One notable difference is that the linear effect of the Muslim population percentage is negative, but not significant in the models in group A (1 and 2) in Table 2, while this effect becomes significant and negative in all models in Table 3. However, this difference is marginal because the overall effect is not linear, but quadratic.

Another important difference is in the magnitude of the effect of the Muslim population percentage. The fully saturated Model 10 in Table 2 produces a combined effect of positive 8.8 (–90 + 6.8 + 162 – 70) at the maximum value of the percentage of the Muslim population, keeping urbanization at the mean of 24.4%. At the same time, in Table 3 Model 22 produces a combined effect of positive 1.2 (–66 + 4.2 + 111 – 48) and Model 26 produces a negative 33.8 (–97 + 4.2 + 155 – 96), also keeping urbanization at its mean of 24.4%. This difference indicates that as we incorporate greater state-level effects there is a greater penalty for the Muslim minority when its relative size increases.

Similarly, within the 0–50% of the Muslim population range, Model 14 in Table 2 produces a negative 7.9, while Model 27 in Table 3 produces a negative 26.3 at the maximum 50% of the Muslim population keeping urbanization at the mean of 20.5 for that interval. In this scenario both linear and particularly cubic effects are stronger with the state-level effects fully modeled in the regression. But, as noted above, relatively few districts (and states) have a high percentage of Muslims. As more state-level effects are introduced in the models, the effects of the states with fewer Muslims become more pronounced due to a much greater number of such states and the districts within those states. Under these conditions the quantitative estimates can be best understood with a reference to graphs such as Figure 2.

It is also important to note that urbanization has a consistently positive effect in all models, indicating that the well-being of the Muslim minority is significantly higher in urban than in the rural areas.

The overall consistency of the coefficients across the models indicates that, although the state-level effects add more depth to the overall dynamic, the general relationship between the Muslim population percentage and the overall well-being of the Muslim community is fairly universal across the Indian districts: it is progressively negative, but the penalty on the well-being decreases after the Muslim population reaches a certain point, arguably around 50%.

Conclusion

In this article we explored how the well-being of the Muslim minority in the Indian districts changes based on the relative size of the Muslim population in the district. The main findings are summarized as follows:

There is a U-shaped relationship between the percentage of the Muslim population and its well-being index values. As the percentage of the Muslim population in a district increases, there is a decrease that is followed by an increase in the value of the minority’s index.

This U-shaped relationship is statistically significant and is net of the effect of urbanization and the state-level effects.

As the Muslim minority in Indian districts reaches about 50% of the population and continues to increase, the overall index for Muslims begins to increase as well. However, other ethnic groups do not experience a decline in the values of their indexes. When Muslims reach the majority population they do not seem to reshape the redistribution of the resources in their favor and to the disadvantage of other groups. In other words, while Muslims are catching up, they do not seem to take over.

In the districts with the overall higher well-being, the gap in the well-being appears to reduce, however a separate measure of the district economic development is needed to properly address the hypothesis of intergroup inequality and economic development.

Out of the three hypotheses, H1 received conditional support, while H2 and H3 were clearly supported. What does this mean in terms of current theories about the relative size of minorities and inequality? They are generally correct, but incomplete: they do not explicitly account for the non-linear dynamic. Specifically, we proposed a population growth inequality theory of the relationship between minority size and its well-being. Our theory was tested and received strong empirical support: there is a U-shaped curve in the relationship between the minority size and its well-being. Any future study that incorporates minority size as a predictor variable should also explore non-linear relationships in the models.

The current study has one main empirical limitation: the cross-sectional nature of the data prevents the exploration over time. Because of the cross-district similarity of the relationship between the percentage of the Muslim minority and its well-being, some suggestions to the change in this relationship over time are possible. Considering the higher population growth among Muslims compared to Hindu it is reasonable to expect that the observed cross-district dynamic will be replicated over time in the same districts when the relative size of Muslims increases. However, any such projections are purely logical and cannot be empirically tested with the current data. A longitudinal dataset is necessary to understand the effects of population growth on intergroup inequality. Another set of limitations is theoretical. Although we suggest several reasons, we do not discuss all the possible reasons why the increasing population may result in the U-shaped well-being.

Although not a policy-oriented article, this study raises several important policy questions. How long can the minority tolerate a high level of inequalities? At which point can the minority–majority tension escalate into a conflict and perhaps a violent one? And, what policies need to be developed to prevent the intergroup conflict and reduce the intergroup inequality? Examination of the relationship between deprivation and social mobilization has been one of the central issues in sociology because deprivation was frequently identified as a prerequisite for major social movements or social change. Such deprivation-driven change rarely happens peacefully. This theoretical link between deprivation and mobilization provides a context for the analysis of the relationship between inequality and social unrest. Although there is no uniform agreement on what causes intergroup conflict, the level of minority deprivation is likely one of the main causes. A careful and continuous examination of the causes of this conflict is vital in constructing the social policy that can effectively prevent escalation of the conflict and encourage peaceful resolution of grievances. There is no disagreement that demographic changes will dramatically shift the social landscape in the coming years and decades. Further investigations of the relationship between inequality and minority size may shed light on how society needs to adjust to be more equitable and stable.

Footnotes

Funding

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

Notes

Author biographies

Riaz Hassan (PhD Ohio State University) is the Director of the International Centre for Muslim and Non-Muslim Understanding at the University of South Australia and a Visiting Research Professor at the Institute of South Asian Studies, National University of Singapore. He is also an Emeritus Professor of Sociology at Flinders University, Adelaide, Australia.

Mikhail Balaev (PhD University of Oregon) is a Senior Lecturer of Sociology at Flinders University (Adelaide, Australia). He is a political-economic sociologist who specializes in quantitative methods, cross-national comparative analysis, network analysis, and the study of the power elites.

Abusaleh Shariff (PhD Australian National University) is the Executive and Chief Scholar at the US-India Policy Institute, Washington, DC. Formerly he was chief economist at the National Council of Applied Economic Research, New Delhi. In 2004–2006 as a member secretary of the Sachar Committee he was an advisor to the Indian Prime Minister in the review of inclusive growth policies for India’s Muslim minority.

References

Albrecht

Murguia

(2005) Minority concentration, disadvantage, and inequality in the nonmetropolitan United States. The Sociological Quarterly 46(3): 503–523.

Basant

Shariff

(2010) The state of Muslims in India: An overview. In: Basant

Shariff

(eds) Handbook of Muslims in India: Empirical and Policy Perspectives. New Delhi: Oxford University Press.

Becker

(1971) The Economics of Discrimination. Chicago: University of Chicago Press.

Blalock

(1956) Economic discrimination and negro increase. American Sociological Review 21: 584–588.

Blalock

(1957) Per cent non-white and discrimination in the South. American Sociological Review 22(6): 677–682.

Blalock

(1967) Toward a Theory of Minority-Group Relations. New York: John Wiley.

Blumer

(1958) Race prejudice as a sense of group position. The Pacific Sociological Review 1(1): 3–7.

Bobo

(1999) Prejudice as group position: Microfoundations of a sociological approach to racism and race relations. Journal of Social Issues 55(3): 445–472.

Bobo

Hutchings

(1996) Perceptions of racial group competition: Extending Blumer’s theory of group position to a multiracial social context. American Sociological Review 61(6): 951–972.

10.

Brass

(2003) The Production of Hindu–Muslim Violence in Contemporary India. Seattle: University of Washington Press.

11.

Cohen

(1998) Black concentration effects on black-white and gender inequality: Multilevel analysis for US metropolitan areas. Social Forces 77(1): 207–229.

12.

Dhattiwala

Biggs

(2012) The political logic of ethnic violence: The anti-Muslim pogrom in Gujarat, 2002. Politics and Society 40(4): 483–516.

13.

Doty

Holmgren

(2006) Health care disconnect: Gaps in coverage and care for minority adults. The Commonwealth Fund 21: 1–12.

14.

Drèze

Sen

(2013) An Uncertain Glory: India and Its Contradictions. Princeton, NJ: Princeton University Press.

15.

Frisbie

Neidert

(1977) Inequality and the relative size of minority populations: A comparative analysis. American Journal of Sociology 82(5): 1007–1030.

16.

Gradín

(2012) Poverty among minorities in the United States: Explaining the racial poverty gap for Blacks and Latinos. Applied Economics 44(29): 3793–3804.

17.

Hasan

(2013) Assessing UPA government’s response to Muslim deprivation. In: Hasan

Hasan

(eds) India: Social Development Report 2012: Minorities at the Margins. New Delhi: Oxford University Press, pp. 242–249.

18.

Iyer

(2002) Demography and Religion in India. Oxford: Oxford University Press.

19.

Iyer

Shrivastava

(2018) Religious riots and electoral politics in India. Journal of Development Economics 131: 104–122.

20.

Kahanec

(2006) Ethnic specialization and earnings inequality: Why being a minority hurts but being a big minority hurts more. Institute for the Study of Labor.

21.

Kurthen

Heisler

(2009) Immigrant integration: Comparative evidence from the United States and Germany. Ethnic and Racial Studies 32(1): 139–170.

22.

Lee

(1998) Huddled Masses, Muddled Laws: Why Contemporary Immigration Policy Fails to Reflect Public Opinion. Westport, CT: Greenwood.

23.

McLaughlin

Stokes

(2002) Income inequality and mortality in US counties: Does minority racial concentration matter? American Journal of Public Health 92(1): 99–104.

24.

Markert

(2010) The changing face of racial discrimination: Hispanics as the dominant minority in the USA – a new application of power-threat theory. Critical Sociology 36(2): 307–327.

25.

Massey

Eggers

(1990) The ecology of inequality: Minorities and the concentration of poverty, 1970–1980. American Journal of Sociology 95: 1153–1188.

26.

Norris

(2004) Electoral Engineering: Voting Rules and Political Behavior. Cambridge: Cambridge University Press.

27.

O’Neill

(1990) The role of human capital in earnings differences between Black and White men. The Journal of Economic Perspectives 4(4): 25–45.

28.

Pew Research Center ( 2015) Latest trends in religious restrictions and hostilities. Available at: www.pewforum.org/2015/02/26/situation-as-of-2013/ (accessed 7 October 2016).

29.

Quillian

(1995) Prejudice as a response to perceived group threat: Population composition and anti-immigrant and racial prejudice in Europe. American Sociological Review 60(4): 586–611.

30.

Rapoport

Weiss

(2003) The optimal size for a minority. Journal of Economic Behavior and Organization 52(1): 27–45.

31.

Sachar

Hamid

Oommen

et al . (2006) Social, Economic and Educational Status of the Muslim Community of India. New Delhi: Cabinet Secretariat, Government of India.

32.

Semyonov

Glikman

(2009) Ethnic residential segregation, social contacts, and anti-minority attitudes in European societies. European Sociological Review 25(6): 693–708.

33.

Shariff

(2015) District Development and Diversity Index: Report for India and Major States. Washington, DC: US-India Policy Institute.

34.

Smedley

Stith

Nelson

(2002) Unequal Treatment: Confronting Racial and Ethnic Disparities in Health Care. Washington, DC: National Academies Press.

35.

Stephan

(2000) An integrated threat theory of prejudice. Reducing Prejudice and Discrimination. New York: Psychology Press, pp. 23–45.

36.

Stephan

Ybarra

Morrison

(2009) Intergroup threat theory. In: Nelson

(ed.) Handbook of Prejudice, Stereotyping, and Discrimination. New York: Psychology Press, pp. 43–59.

37.

Tienda

Lii

D-T

(1987) Minority concentration and earnings inequality: Blacks, Hispanics, and Asians compared. American Journal of Sociology 93(1): 141–165.

38.

Varshney

(2003) Ethnic Conflict and Civic Life: Hindus and Muslims in India. New Haven, CT: Yale University Press.

39.

Watson

(1994) Minority access and health reform: A civil right to health care. The Journal of Law, Medicine and Ethics 22(2): 127–137.

40.

Williams

Jackson

(2005) Social sources of racial disparities in health. Health Affairs 24(2): 325–334.

41.

Williams

(1947) The reduction of intergroup tensions: A survey of research on problems of ethnic, racial, and religious group relations. Social Science Research Council Bulletin 57. New York: Social Science Research Council.