Uncovering social assimilation disparities: immigrants’ volunteering in associations

ABSTRACT

This study examines how much immigrants from different country of origin groups assimilate to native-level volunteering in organizations in Denmark. We employ multilevel linear probability models, analyzing a dataset that combines repeated cross-sectional survey data with administrative register information. The dataset encompasses 15,771 respondents, of which 1,918 are immigrants from 127 different countries. Unlike recent European comparative research that has suggested that immigrants of all ethnic backgrounds ultimately reach parity with natives in their levels of civic participation in their destination countries, our findings reveal a different scenario in Denmark. Our results suggest that immigrants from predominantly Protestant and other Christian countries reach parity with natives in their level of volunteering, while immigrants from other countries do not. We show that socioeconomic disparities across immigrant groups partially account for these variations in volunteering assimilation, but even after accounting for socioeconomic factors and many other covariates, significant differences persist.

KEYWORDS:

• Assimilation
• volunteering
• voluntary associations
• immigrants
• socioeconomic disparities

Introduction

Extensive research conducted in Europe and North America has consistently shown that immigrants, particularly those from non-Western countries, are less likely to volunteer for organizations than native populations. In this context, volunteering refers to individuals offering their time and effort to assist organizations without expecting to be paid. Studies have documented disparities in volunteering in various regions, including North America (Wang and Handy 2014) and several European countries, such as Germany (Greenspan, Walk, and Handy 2018), the Netherlands (Carabain and Bekkers 2011), Luxemburg (Valentova and Alieva 2018), Norway (Eimhjellen and Segaard 2010), and Denmark (Qvist 2018).

Given the continuous influx of immigrants and refugees into North America and Europe, gaining a deeper understanding of these disparities in immigrants’ involvement in volunteering is imperative for two main reasons. First, in many Western countries, voluntary organizations are essential democratic infrastructure. Notably, voluntary organizations allow citizens to make their voices heard and participate in political decision-making that shapes their local communities (Ødegård and Fladmoe 2020; Ruiz, Wang, and Handy 2021). Second, voluntary organizations are social arenas where people can form social ties across ethnic and religious boundaries through face-to-face interactions, fostering social integration (Greenspan, Walk, and Handy 2018; Handy and Greenspan 2009).

At present, scholars have primarily attributed the observed native-immigrant gaps in volunteering to socioeconomic disparities (Greenspan, Walk, and Handy 2018; Qvist 2018) and cultural and religious differences (Aksoy and Wiertz 2023; Aleksynska 2011; Carabain and Bekkers 2012; Voicu 2014). However, most of this research has taken a static approach by comparing volunteering rates across immigrant groups and native populations irrespective of the immigrants’ length of residence. Consequently, our understanding of how immigrants assimilate into native-level volunteering over the years spent in their destination societies remains limited.

Nevertheless, a handful of recent European comparative studies have provided insights into how much immigrants adapt to native-level civic participation, with civic participation measured as membership in voluntary organizations (Voicu and Şerban 2012), volunteering for such organizations (Greenspan, Walk, and Handy 2018), or by composite measures of civic participation (Aleksynska 2011). This growing literature suggests that although static comparisons of volunteer rates suggest that immigrants are less likely to volunteer than natives, immigrants from various ethnic backgrounds ultimately attain similar levels of civic participation as native populations in their destination societies. Nevertheless, the studies also highlight varying assimilation timelines among immigrant groups, with some taking longer than others to reach parity (Aleksynska 2011; Voicu and Şerban 2012).

While the literature acknowledges that different immigrant groups have different timelines for assimilation, it is surprising that it indicates immigrants eventually achieve the same level of civic participation as native-born individuals, regardless of their ethnic backgrounds. This finding is surprising because the broader scholarship on immigrant assimilation typically underscores that immigrants from countries with significant religious and cultural differences usually require several generations to fully assimilate (Alba and Nee 2003; Heath and Schneider 2021; Rumbaut 2015).

Against this background, we investigate whether immigrants, during their years in Denmark, reach parity with native residents in their levels of volunteering. We group immigrants based on the dominant religion in their countries of origin, as religious differences are recognized as a significant social and cultural cleavage in Europe (Alba and Foner 2015; Drouhot and Nee 2019). We distinguish between immigrants from predominantly Protestant, non-Protestant Christian, Muslim, and HBO (i.e. Hindu, Buddhist, and other) countries. Methodologically, we use multilevel linear probability models to analyze a random sample of 13,853 natives and 1,918 immigrants nested in 127 origin countries from the three most recent waves of the Danish Volunteer Survey. We merge these survey data with administrative register data, including information about respondents’ countries of origin, length of residence, and socioeconomic factors.

Our primary contribution to the literature is that we demonstrate that immigrants in Denmark assimilate to native-level volunteering at different rates. Specifically, immigrants from predominantly Christian countries, particularly those with Protestant backgrounds, quickly assimilate and reach parity with natives in their levels of volunteering during their time in Denmark. In contrast, our findings reveal persistent disparities between natives and immigrants from non-Christian countries, regardless of their duration of residence in Denmark, even after accounting for their socioeconomic status and many other covariates.

The case of Denmark

Immigrant population composition in Denmark

Historically, Denmark has been culturally homogeneous, with immigration flows primarily coming from neighboring and culturally proximate countries such as Norway, Sweden, and Germany. However, in the 1960s, the Danish government, like many other Western European countries, introduced guest worker programs to secure economic growth during almost full employment. In contrast to earlier waves of migration, the guest workers primarily came from religiously and culturally distant countries such as Turkey, Pakistan, and Yugoslavia (Ejrnæs 2022). However, in response to the oil crisis in 1973, the Danish government abruptly ceased guest worker programs and introduced an immigration ban. Despite this, immigration continued due to family reunifications and influxes of refugees (Ejrnæs 2022), increasing the immigrant population from 3 per cent in 1980 to 14 per cent in 2021.

Figure 1 shows Denmark’s twenty largest immigrant groups in 2021 based on immigrants’ countries of origin. Aside from immigrants from historically Christian-Protestant neighboring countries such as Sweden, Norway, and Germany, Denmark has received a large immigrant population from predominantly Muslim countries, including Turkey, Syria, Iraq, Lebanon, Pakistan, Somalia, Iran, and Afghanistan (Figure 1). Consequently, as of 2015, Islam has become the largest minority religion, with an estimated 4–5 per cent of the Danish population being Muslim (Jacobsen 2015). Additionally, Denmark has received a relatively small but growing influx of immigrant populations from Christian-Catholic and Christian-Orthodox countries, including Poland, Romania, Lithuania, and Ukraine, mainly due to the Eastern enlargement of the European Union in 2004. However, Protestantism remains the majority religion in Denmark. Estimates from the Pew Research Center (Pew Research Center 2011) suggest that approximately 86 per cent of the Danish adult population identifies with the Protestant denomination. However, it is essential to note that religiosity levels among the native population in Denmark have long been on a declining trend (Voas 2009).

Figure 1. The 20 largest immigrant groups in Denmark in 2021, based on immigrants’ countries of origin. Absolute numbers. Source: Statistics Denmark, FOLK1.

Voluntary sector in Denmark

Denmark and the other Scandinavian countries are well-known for their strong and vibrant civil societies with high participation levels. In 2015, the share of the Danish population reporting to have volunteered within the past year was approximately 38 percent, compared to approximately 19 percent on average across Europe (Henriksen, Strømsnes, and Svedberg 2019). The high level of volunteering has remained largely stable over the past three decades (Henriksen, Strømsnes, and Svedberg 2019; Qvist et al. 2019).

Another characteristic feature of the voluntary sector in Denmark is that a large share of volunteering occurs in cultural and leisure organizations; among these, sports organizations attract the largest share of the volunteering population. In the supplementary material, Table A1 shows natives’ and immigrants’ participation in different areas of volunteering in Denmark. The table highlights that immigrants are only slightly more likely to engage in religious volunteering than natives. This finding contrasts with those from other countries, which show that immigrants have comparatively high volunteer rates in religious organizations (Carabain and Bekkers 2011; Wang and Handy 2014).

Theory and hypotheses

In this section, we discuss how and why immigrants may assimilate to native-level volunteering considering broader theories of these adaptation processes. We also present a set of empirically testable hypotheses based on our theoretical discussion.

Immigrant assimilation and volunteering

Classic assimilation theory, originally formulated during the late nineteenth and early twentieth centuries in the context of European migration to the United States, provides a general framework for understanding how immigrants’ opportunities, beliefs, and behaviors change after resettlement in a new country. The theory’s basic assumption is that immigrants, over their life cycles and across successive generations, converge toward the native population in their destination country on cultural, economic, and social dimensions (Zhou 1997). In the context of civic participation and volunteering in Europe, this assumption has found strong support in the literature. It suggests that immigrants in Europe not only become more likely to engage in civic participation as they spend more time in their destination countries but also that they eventually reach parity with natives in their levels of civic participation in destination societies (Aleksynska 2011; Voicu and Şerban 2012).

New assimilation theory, tailored to contemporary waves of immigrants from non-Western countries to North America and Europe, retains the fundamental assumption that differences between immigrants and natives on various economic, social, and cultural dimensions erode over immigrants’ lifecycles and across successive generations through a multifaceted convergence process (Drouhot and Nee 2019). However, new assimilation theory emphasizes that the duration of these processes varies between immigrant groups as it depends on how quickly boundaries between their members and natives are bridged. The speed at which such bridging processes occur depends on factors such as the religiously and culturally determined social distance between the immigrants and the native residents, the immigrants’ socioeconomic resources, and the degree to which local communities facilitate these processes (Alba and Nee 2003). This argument suggests that immigrants arriving in Western countries from less affluent and religiously and culturally distant Global South countries face more substantial barriers to assimilation than other immigrants.

In their extensive review of immigrant assimilation literature in Western Europe and the United States, Drouhot and Nee (2019) suggest that religious differences between immigrants and native residents pose a significant barrier to the assimilation process, especially for immigrants of Muslim backgrounds in Western Europe. Much empirical research supports that argument. It is, for example, well-documented that native residents in Western European countries are more likely to hold negative views about immigrants from Muslim countries than those from other countries (Alba and Foner 2015; Bansak, Hainmueller, and Hangartner 2016). It is also well documented that immigrants from Muslim countries face discrimination in job-seeking processes (Thijssen et al. 2021). Moreover, a recent cross-national field-experiment found a similar pattern in associational life. The study revealed that individuals with native-sounding names had a better chance of getting a response when requesting to join an amateur football team for a training session than their counterparts with foreign-sounding names. Specifically, in the context of Denmark, the study highlighted that individuals with Turkish-sounding names, which are often associated with a Muslim background, received fewer responses than German- and Polish-sounding names (Gomez-Gonzales, Nesseler, and Dietl 2021).

In addition to the above, substantial evidence indicates that immigrants from religiously and culturally distant countries, especially those from Muslim countries, experience slower assimilation than other immigrants in Western Europe across various aspects (Bisin et al. 2008; Drouhot and Nee 2019). It is, for example, evident that immigrants from Muslim countries in Western Europe have weaker positions in the labor market than immigrants from other countries (Blekesaune 2021; Koopmans 2016; 2020; Statistics Denmark 2021). It is also evident that immigrants from Muslim countries, on average, uphold high levels of religiosity through their life cycles and across generations, despite normative pressure toward secularization from surrounding Western European societies (Drouhot 2021; Simsek, Fleischmann, and van Tubergen 2019). Moreover, evidence suggests that immigrants from Muslim countries are less likely to form friendships or romantic relationships with natives than immigrants from other countries (Leszczensky and Pink 2017; Qvist and Qvist 2023; Simsek, van Tubergen, and Fleischmann 2021).

Although assimilation disparities across immigrant groups are well-documented in many respects, few empirical studies have examined how different immigrant groups assimilate to native residents’ levels of civic participation. Notably, the few studies in this area have produced unexpected findings when viewed in the broader immigrant assimilation scholarship context. For example, Aleksynska (2011) conducted a study using data from 19 countries in the European Social Survey. The study suggested that although immigrants with a Muslim background took longer to reach parity with natives in their level of civic participation than other immigrants, they eventually did so after approximately twenty years of residence. The study assessed civic participation through a composite variable encompassing various indicators such as being a trade union member, working for an association, or participate in symbolic actions (e.g. signing petitions or participating in lawful demonstrations). Another example is Voicu and Şerban (2012), who also conducted a cross-national study of memberships in voluntary associations using the European Social Survey. The study likewise found that immigrants reached parity with natives in their voluntary association membership levels after approximately twenty years of residence, even though they were much less likely to be members of voluntary associations than native residents upon arrival. Consequently, based on the literature, we should expect the following hypotheses to receive support:

Hypothesis 1: The probability that immigrants volunteer increases as they spend additional years in Denmark (H1a); however, the time it takes for immigrants to reach parity with natives in their level of volunteering varies across groups. It takes fewer years for immigrants from religiously and culturally proximate Christian, and in particular Protestant, countries to reach parity with natives in their level of volunteering (H1b) and more years for immigrants from religiously and culturally distant non-Christian, and in particular Muslim, countries (H1c).

The role of socioeconomic disparities across immigrant groups

A prominent scholarship emphasizes that volunteering, like paid work, is a productive activity that requires socioeconomic resources such as education, skills, a stable income, and good health (Musick and Wilson 2007; Wilson 2012). In line with this research, socioeconomic disparities are likely to explain, at least partly, the expected differences in the rate at which immigrants assimilate to native-level volunteering in Denmark.

In Denmark, like many European countries, immigrants from non-Western countries, especially those from Muslim countries, have lower levels of educational attainment and lower employment rates than immigrants from Western countries (Statistics Denmark 2021). This pattern suggests that immigrants from Global South countries, especially Muslim countries, might take longer to assimilate to the levels of volunteering among natives, given that participation often hinges on having adequate socioeconomic resources. Previous research supports this argument as studies have documented that observed native-immigrant gaps in volunteering narrow when they control for socioeconomic factors (Greenspan, Walk, and Handy 2018; Qvist 2018).

However, as noted earlier, most of these studies have taken a static approach, comparing the volunteering rates of immigrant groups and native residents without considering how the length of immigrants’ residence influence their assimilation in the context of volunteering. This approach leaves uncertainty regarding how socioeconomic resources shape immigrants’ incorporation into volunteering activities over time. Nevertheless, we hypothesize the following:

Hypothesis 2: When we control for individual-level socioeconomic resources, the magnitude of group-level differences in the years it takes for immigrants to reach parity with natives in their level of volunteering decreases.

Data, measures, and analytical strategy

Data

We analyze pooled cross-sectional data from the Danish Volunteer Survey (DVS) collected in 2012, 2017, and 2020. We pool the data to obtain as large a sample of immigrants as possible. The DVS is a representative survey of the Danish population aged 16–90.

The 2012 wave included an oversample of 960 non-Western immigrants who had resided in Denmark for five years or more and could complete the interview in Danish. Similar oversamples of immigrants were unfortunately not collected in 2017 and 2020, but these waves contained immigrants randomly drawn from the general population who could complete the interview in Danish. Consequently, our sample includes more immigrants from the 2012 wave than the others (see Table 1). The fact that the immigrants in all waves had to complete the interview in Danish means that our study is representative of immigrants who speak Danish at a reasonable level.

Table 1. Characteristics of the pooled dataset by survey year. Absolute numbers.

Survey year	Natives	Immigrants	Total
2012	2,658	1,041	3,699
2017	4,612	358	4,970
2020	6,583	519	7,102
Pooled dataset	13,853	1,918	15,771

In Denmark, all permanent residents hold a unique personal identification number that government authorities use to store information for administrative purposes. This unique personal identification number in anonymized form enables authorized researchers to merge survey data with information from administrative registers hosted on remote access servers at Statistics Denmark (see Qvist 2022, 135-145). This access allows us to obtain precise information about immigrants’ length of residence based on when the respondent first appeared on the Danish administrative registers. Moreover, we also obtain information on immigrants’ countries of origin and socioeconomic factors, including income, educational level, and employment status, from the administrative registers.

In our analyses, we restrict our sample to first-generation immigrants because the second generation in the pooled dataset constitutes only a minor group of respondents (n = 170). In addition, we exclude immigrants for whom information about country of origin is missing (n = 78). The resulting dataset comprises 15,771 respondents, of which 13,853 are natives, and 1,918 are immigrants.

Variables and measurements

Volunteering

The dependent variable is an indicator variable that equals one if the respondent has volunteered for an organization within one or more areas in the past year and zero otherwise. Following the International Classification of Nonprofit Organizations (Salamon and Anheier 1996), the survey asked whether the respondents had volunteered in one or more of the fourteen areas: culture, sports, leisure, education, health, social work, environment, housing, professional work, counseling, religion, politics, international activities, and others. It provided respondents with examples of typical volunteering activities for each area to ensure they knew what activities to include.

Length of residence

We measured length of residence by counting the years since the first year an immigrant appeared on the administrative registers. Since the administrative registers started recording high-quality data in 1985, the variable is right-censored, with the maximum value being 27 years. We use the length of residence both as a numeric measure and as a binary measure to facilitate comparisons of our data with that in previous research. The binary measure categorizes immigrants into two groups: those who have lived in Denmark for less than 20 years and those who have lived in Denmark for more than 20 years.

Dominant religion in immigrants’ countries of origin

Religion plays a pivotal role in shaping a country’s culture. For this reason, we group immigrants into four groups based on the dominant religion in their countries of origin. To construct this country-level variable, we first use administrative register data to identify immigrants’ countries of origin. We then combine this data with information from the Pew Research Center on the religious composition of immigrants’ countries of origin (Pew Research Center 2011, 2012). If at least half of the population in a country is Christian or Muslim, we classify the country of origin accordingly.

Moreover, we split the group of Christian countries into either Protestant or non-Protestant Christian (i.e. Catholic or Orthodox) countries based on which of these denominations has the highest share of affiliates in the respective countries. Finally, we construct a category for immigrants from countries with either Hindu, Buddhist, mixed, or other religious majorities. Going forward, we will refer to this category as HBO countries, an acronym for Hindu, Buddhist, and other countries. Table 2 shows how we grouped the countries.

Table 2. Categorization of immigrants based on the dominant religion in their country of origin.

Dominant religion	Country of origin	N	%
Protestant countries	Germany (n = 83), Sweden (n = 50), Norway (n = 40), Great Britain (n = 30), Iceland (n = 19), USA, Finland, Ghana, Kenya, Uganda, Latvia, Australia, Dem. Rep. Congo, South Africa, Zimbabwe, Barbados, Liberia, Zambia, Guyana, Jamaica, Lesotho, New Zealand, Trinidad & Tobago	305	16
Non-Protestant Christian countries	Poland (n = 51), Philippines (n = 50), Ukraine (n = 31), Russia (n = 24) Netherlands (n = 24), Lithuania, Brazil, France, Italy, Chile, Croatia, Hungary, Spain, Argentina, Colombia, Burundi, Belgium, Canada, Austria, Cuba, Peru, Venezuela, Cameroon, Mexico, Switzerland, Ireland, Portugal, Tanzania, Costa Rica, Dominican Republic, Ecuador, Mozambique, Angola, Bolivia, Grenada, Malta, Nicaragua, Panama, Rwanda, Seychelles, Slovakia, Slovenia, Romania, Macedonia, North Macedonia, Bulgaria, Eritrea, Ethiopia, Greece, Serbia, Belarus, Serbia & Montenegro, Armenia, Georgia, Moldova	436	23
Muslim countries	Turkey (n = 176), Iraq (n = 119), Iran (n = 80), Afghanistan (n = 68), Lebanon (n = 53), Pakistan, Somalia, Syria, Morocco, Egypt, Kosovo, Indonesia, Jordan, Kuwait, Albania, Tunisia, Bangladesh, Algeria, Gambia, Sierra Leone, Uzbekistan, Malaysia, Kazakhstan, Bahrain, Azerbaijan, Bahrain, Guinea-Bissau, Libya, Maldives	727	38
HBO countries	Bosnia-Herzegovina (n = 112), Former Yugoslavia (n = 66), Thailand (n = 57), Vietnam (n = 50), Sri Lanka (n = 44), China, Fed. Rep. Yugoslavia, Nepal, Myanmar, Israel, Japan, Czechoslovakia, Singapore, Estonia, Czech Republic, South Korea, Bhutan, Mongolia	450	23
Total		1,918	100

Note: HBO = Hindu, Buddhist, and other countries. Classifications are based on “The Global Religious Landscape” by the Pew Research Center (Pew Research Center 2011, 2012). The number of respondents in the five largest countries within each category are shown in parentheses. Western countries are written in italics.

Socioeconomic resources

We operationalize socioeconomic resources as education, income, employment status, and self-rated health. Using information from administrative registers, we lag our education, income, and employment status measures by one year to mitigate reverse causality issues.

Education is defined as the highest level of education completed, and it includes “primary school”, “gymnasium/high school”, “vocational education”, “medium-cycle education”, and “long-cycle education”. Additionally, we include a category labeled “no information” to avoid the deletion of those respondents for whom the educational level is unknown. The same strategy applies to the income and employment status measure since complete information on every respondent is unavailable. Income is a categorical variable that captures the respondents’ income in quintiles and a category for respondents with “no information”. Employment status is a categorical variable differentiating between “employed”, “unemployed”, “out of the labor force”, “student”, and we also include a category for respondents for whom there is “no information”. We measure self-rated health by asking respondents, “How is your health in general?” and the values range from 1 (“very bad”) to 5 (“very good”).

Migration-related covariates

To address variations in migration-related factors across immigrant groups, we include citizenship status, age at arrival, and period of arrival as covariates. Citizenship status is a binary variable equal to one if the immigrant has Danish citizenship and zero otherwise. Age at arrival is a binary measure, with one indicating that the immigrant was under 25 when arriving in Denmark and zero otherwise. Migrating at an older age potentially leaves a stronger imprint of one’s country of origin, which can influence proclivities to civically participate in postmigration contexts (Voicu and Rusu 2012). The period of arrival is a categorical variable distinguishing between arrival in the period (1) “before 1985”, (2) “between 1986-1999”, and (3) “2000 and onward”. These periods were selected to reflect different eras with unique integration policies in Denmark (Ejrnæs 2022), which could potentially influence immigrants’ willingness to engage in voluntary associations.

Country-level covariates

To adjust for differences beyond religion across countries of origin, we include covariates related to economic development and democratic and legal culture in our analyses. We include the Human Development Index (HDI) based on the United Nations Development Programme (2020) to capture economic development. HDI is a composite measure that captures life expectancy, years of education, expected years of education, and each country’s gross national income per capita. Specifically, we calculated a mean human development score for each country of origin from 1990, 2000, 2010, and 2020. Our human development variable ranges from 1 (low degree of human development) to 10 (high degree).

To capture differences in the country’s democratic and legal culture, we include a composite measure for civil liberties using Freedom House’s annual reports on Freedom in the World. The civil liberty variable is based on 15 items covering freedom of expression and belief, associational and organizational rights, rule of law, personal autonomy, and individual rights. Again, we calculate a mean score of civil liberties for each country of origin across the years 1980, 1990, 2000, and 2010. This variable ranges from 1 (very free society) to 7 (very unfree society).

Demographic covariates

To account for variations in demographic composition among immigrant groups, we include a set of demographic factors as control variables, including survey year, sex, and age. Age is included with its square because participation in voluntary organizations usually follows an inverse U-shaped pattern over lifecycles (van Ingen 2008).

Table 3 provides descriptive statistics for all variables.

Table 3. Descriptive statistics.

	Natives	Immigrants: split by dominant religion in country of origin
	Native Danes	Protestant	Non-Protestant Christian	Muslim	HBO
	(n = 13,858)	(n = 305)	(n = 436)	(n = 727)	(n = 450)
Individual-level variables
Volunteering (%)	38	36	28*	20*	19*
Female (%)	53	54	64*	48*	58*
Age (16–90)	50.7 (17.8)	49.9 (15.9)	42.5 (14.1)*	41.2 (13.4)*	41.2 (13.3)*
Income (0–5)	3.2 (1.4)	3.1 (1.5)	2.8 (1.5) *	2.2 (1.2)*	2.5 (1.4)*
Education (%)
No information	1	19*	25*	11*	14*
Primary school	24	11*	12*	40*	30*
High school	5	7	8*	10*	8*
Vocational	40	29*	22*	24*	31*
Medium length cycle	20	17	15*	9*	10*
Long-cycle	10	17*	18*	6*	7
Employment status (%)
No information	1	9*	11*	5*	7*
Employed	64	59	60	46*	60
Unemployed	5	5	8*	22*	12*
Out of labor force	24	24	15*	18*	15*
Student	7	3*	5	9*	6
Self-rated health (1–5)	4.0 (1.0)	4.0 (1.0)	4.0 (1.0)	3.6 (1.2)*	3.8 (1.1)*
Survey year (%)
2012	19	27*	36*	71*	64*
2017	33	29	24*	13*	16*
2020	48	44	40*	16*	20*
Immigrant-specific variables
Length of residence (0-27) ^a		20	13*	19	17*
Length of residence (%)
Less than 20 years		48	66*	51	60*
20 years or more		52	34*	49	40*
Danish citizenship (%)		28.9	31.2	55.3*	49.8*
Age at arrival (%)
25 years or younger		13	19	41*	38*
Older than 25 years		87	81	59*	62*
Time-period of arrival (%)
1985 or before		8	8	17*	11
1986–1999		53	35*	48	58
2000 or later		39	57*	34	31
Country-level variables
Civil Liberties (1–7)		1.4 (1.1)	2.4 (1.5)*	5.0 (1.2)*	4.0 (1.0)*
HDI (0–10)		8.3 (1.4)	7.5 (1.1)*	6.1 (1.2)*	6.7 (0.8)

Note: HBO = Hindu, Buddhist, and other countries. The reference group is native Danes, except for the immigrant-specific and country-level variables, where immigrants from Protestant countries constitute the reference group. Standard deviations are presented in parentheses. HDI = Human Development Index. ^a For the numeric measure of length of residence, we use the median because the variable is right-censored. * p < 0.05 based on two-tailed t-test for differences in means, medians, or proportions.

Analytical strategy

We use random intercept multilevel linear probability models (multilevel LPMs) with individuals nested in origin countries to model the volunteering behavior of immigrants and natives. We choose LPMs over other binary choice models, such as logistic models, because LPMs provide directly comparable coefficients across models – which is also true when we include interaction terms in the models (Breen, Karlson, and Holm 2018; Mood 2010). We use robust standard errors in all models because heteroskedasticity is always present in LPMs. Our basic multilevel linear probability model takes the following form (1): (1) $Pr(Y_{\mathrm{ij}}=1)={\beta }_0+{\beta }_G{\mathit{G}}_j+{\beta }_Z{\mathit{Z}}_{\mathrm{ij}}+u_j+{\epsilon }_{\mathrm{ij}}$(1) where $\mathrm{Pr}(Y_{\mathrm{ij}}=1)$ is the probability of volunteering within the past year for individual i from origin country j,${\beta }_0$ is the overall model intercept, ${\mathit{G}}_j$ is a vector of dummy variables for immigrant groups (with natives as the reference group), ${\mathit{Z}}_{\mathrm{ij}}$ is a vector of survey-year and sociodemographic covariates that varies across individuals and origin countries, $u_j$ is a random intercept across origin countries. and ${\epsilon }_{\mathrm{ij}}$ is an individual-level error term. The central coefficients of interest are ${\beta }_G$, which provide estimates of the sizes of the native-immigrant gaps in volunteering for different immigrant groups.

To examine hypothesis 1 regarding the extent to which immigrants reach parity with natives in their level of volunteering over the years they spend in Denmark, we run two different multilevel linear probability models, as stated in Equations (2a) and (2b). Both Equations (2a) and (2b) are extensions of the basic model in Equation (1): (2a) $Pr(Y_{\mathrm{ij}}=1)={\beta }_0+{\beta }_G{\mathit{G}}_j+{\beta }_{\mathrm{LR}}L{R}_{\mathrm{ij}}+{\beta }_{G\times \mathrm{LR}}{\mathit{G}}_j\times L{R}_{\mathrm{ij}}+{\beta }_Z{\mathit{Z}}_{\mathrm{ij}}+u_j+{\epsilon }_{\mathrm{ij}}$(2a) (2b) $Pr(Y_{\mathrm{ij}}=1|L{R}_{\mathrm{ij}}^b=k)={\beta }_0+{\beta }_G{\mathit{G}}_j+{\beta }_Z{\mathit{Z}}_{\mathrm{ij}}+u_j+{\epsilon }_{\mathrm{ij}}$(2b)

In Equation (2a), we interact immigrant groups with the numeric measure of the length of residence ${\mathit{G}}_j\times L{R}_{\mathrm{ij}}$, allowing us to investigate whether the likelihood that immigrants will volunteer increases with the years spent in Denmark. Because the length of residence is an immigrant-specific variable, we exclude natives from these analyses. We also run subgroup regressions, stratifying on a binary measure of the length of residence ($L{R}_{\mathrm{ij}}^b=k$): one regression for k = short-term immigrant residents and another for k = long-term immigrant residents. The central coefficients in Equation (2b) are ${\beta }_G$, which estimate the gap between how much natives and immigrants volunteer, stratified on how long immigrants have resided in Denmark.

Finally, to test hypothesis 2, we carry out a series of regressions that allow us to investigate the gaps in volunteering when we adjust for socioeconomic resources and the other covariates. We again exclude native Danes and instead use immigrants from Protestant countries as the reference group. Our model is as follows: (3) $Pr(Y_{\mathrm{ij}}=1)={\beta }_0+{\beta }_G{\mathit{G}}_j+{\beta }_{L{R}^b}L{R}_{\mathrm{ij}}^b+{\beta }_{G\times L{R}^b}{\mathit{G}}_j\times L{R}_{\mathrm{ij}}^b+{\beta }_Z{\mathit{Z}}_{\mathrm{ij}}+{\beta }_M{\mathit{M}}_{\mathrm{ij}}+{\beta }_C{\mathit{C}}_j+{\beta }_R{\mathit{R}}_{\mathrm{ij}}+u_j+{\epsilon }_i$(3) where ${\mathit{G}}_j\times L{R}_{\mathrm{ij}}^b$ is an interaction term between immigrant groups and the binary measure of length of residence, ${\mathit{M}}_{\mathrm{ij}}$ is a vector of migration-related covariates (i.e. citizenship status, age at arrival, and time-period of arrival), ${\mathit{C}}_j$ is a vector of country-level covariates (i.e. civil liberties and human development), and ${\mathit{R}}_{\mathrm{ij}}$ is a vector of individual-level socioeconomic resources (i.e. education, income, employment status, and health). The central part is to assess whether the coefficients ${\beta }_G$ change with the stepwise inclusion of ${\mathit{M}}_{\mathrm{ij}}$, ${\mathit{C}}_j$ and ${\mathit{R}}_{\mathrm{ij}}$ because that would indicate that differences in the probabilities of volunteering between long-term immigrant residents from different immigrant groups are related to individual-level socioeconomic factors, as posited in hypothesis 2.

Results

Disparities in volunteer rates between immigrants and natives

Figure 2 shows how much natives and immigrants differ in their likelihood of volunteering, as indicated by predicted probabilities of volunteering based on a multilevel linear probability model that only controls for the survey year. The figure shows substantial variations in the likelihood of volunteering between native individuals and immigrants from different immigrant groups. In the case of immigrants from Protestant countries, the probability of volunteering is approximately 36 percent (95 CI: [30.9, 40.8]). This probability is close to the probability that natives volunteer, approximately 38 per cent (95 CI: [37.9, 38.0]). The slight difference between the predicted probability that immigrants from Protestant countries and natives volunteer is statistically insignificant.

Figure 2. Predicted probabilities that natives and immigrants volunteer, controlling for survey year. N = 15,771. Note: HBO = Hindu, Buddhist, and other countries. The dashed line indicates native Danes’ level of volunteering (≈38%). Probabilities are presented in percentages. Error bars present 95% confidence intervals calculated with robust standard errors. The complete results appear in Table B1 in the supplementary material.

For all other immigrant groups, we find relatively large and statistically significant differences between natives and immigrants, which means that immigrants from non-Protestant Christian, Muslim, and HBO countries are less likely to volunteer than natives. For immigrants from non-Protestant Christian countries, the predicted probability of volunteering is approximately 29 per cent (95 CI: [22.1, 35.1]). The probability is approximately 22 per cent for immigrants from Muslim countries (95 CI: [17.6, 26.0]) and approximately 20 per cent for immigrants from HBO countries (95 CI: [17.6, 22.3]). These estimates remain largely the same when we control for demographics such as sex and age (see Table B1 in the supplementary material).

Do immigrants reach parity with natives in their level of volunteering?

Figure 3 plots the predicted probabilities that immigrants will volunteer during their years in Denmark. The figure suggests that while immigrants from Christian countries reach parity with natives, this is not the case for other immigrant groups. For immigrants from Protestant countries, we see that their likelihood of volunteering is relatively high upon arrival, and for those who have been long-time residents, it is on par with that of native Danes. This pattern indicates that immigrants from these countries quickly assimilate to native-level volunteering until they reach parity with natives. We see a similar pattern for immigrants from other Christian countries, although the point estimates suggest that the predicted probability of volunteering increases less steeply with length of residence than it does for immigrants from Protestant countries.

Figure 3. Predicted probabilities that different immigrant groups volunteer based on their length of residence. Immigrant sample. N = 1,918. Note: HBO = Hindu, Buddhist, and other countries. AME = average marginal effects. Gray areas around the prediction lines show 95% confidence intervals calculated with robust standard errors. The dashed line indicates the probability that native Danes volunteer (≈ 38%) as a point of reference, although they are not part of the regression. The predicted probabilities are derived from multilevel linear probability models controlling for survey year, sex, and age (squared). The full sets of estimates can be found in Table B2 in the supplementary material. * p ≤ 0.05.

For immigrants from Muslim countries, on the other hand, we find no evidence that their probability of volunteering increases with the number of years they spend in Denmark (Figure 3). Instead, they are nearly half as likely to volunteer as natives upon arrival, and long-time residents are not more likely to volunteer than short-time residents. This pattern indicates that immigrants from Muslim countries do not assimilate to native-level volunteering. For immigrants from HBO countries, we find a different pattern. They have a low probability of volunteering upon arrival, but it is higher for long-time residents than for short-time residents. Still, they do not reach parity with native-level volunteering.

While the estimates presented in Figure 3 make complete use of the variation in the length of residence, which we include as a continuous variable, we cannot compare the immigrants’ volunteering levels to natives. Consequently, we run a supplementary subgroup analysis with the length of residence coded as a binary variable that distinguishes short-time residence (i.e. less than 20 years of residence) and long-time residence (i.e. 20 years of residence or more). Here, we plot differentials in the predicted probabilities of volunteering between the immigrant groups and natives, and we present the results in Figure 4. Overall, the results from this analysis lead to conclusions similar to those above. For immigrants from Protestant countries with long-term residence, we see that their predicted probability of volunteering is slightly higher than that for natives, although not statistically significantly different. For immigrants from other Christian countries with long-term residence, we find that the predicted probability of volunteering is approximately seven percentage points lower than that for natives. However, the estimated confidence interval overlaps the level of volunteering among natives (95 CI: [−17.5, 6.0]).

Figure 4. Differences in the predicted probabilities that natives and immigrants’ volunteer. Full sample. N = 15,771. Note: HBO = Hindu, Buddhist, and other countries. The dashed line indicates native Danes as the reference group. Error bars present 95% confidence intervals calculated with robust standard errors. The results are derived from subgroup multilevel linear probability models controlling for survey year, sex, and age (squared). The complete results appear in Table B2 in the supplementary material.

For immigrants from non-Christian countries, this additional analysis that distinguishes between short- and long-term residing immigrants illuminates the magnitude of gaps in volunteering between natives and immigrants. When comparing immigrants from Muslim countries who have been in the country for a long time to natives, we find that they are almost 19 percentage points less likely to volunteer than natives (18.9 per cent-points, 95 CI: [−24.1, −13.6]). Likewise, immigrants from HBO countries are 17 per cent less likely to volunteer than natives (−17.0 per cent-points, 95 CI: [−21.1, −12.8]).

Are socioeconomic disparities related to assimilation differentials in volunteering?

In this section, we examine the extent to which socioeconomic disparities, migration-related factors, and country-level factors are related to the differences in assimilation rates among immigrant groups. We present the results from these multivariable analyses in Figure 5. The figure presents the predicted probabilities of volunteering among the different immigrant groups. The results suggest that the assimilation differentials in volunteering among immigrant groups remain largely unchanged after considering migration-related and country-level factors.

Figure 5. Differences in the predicted probabilities that immigrants volunteer. Immigrant sample. N = 1,918. Note: HBO = Hindu, Buddhist, and other countries. The dashed line indicates immigrants from Protestant countries who have resided in the country for a long time (i.e. 20 years or more) as the reference group. All models control for the survey year. Migration-related factors include citizenship status, age at arrival, and period of arrival. Country-level factors include civil liberties and human development. Individual-level socioeconomic resources include income, education, employment status, and self-rated health. The complete results appear in Table B3 in the supplementary material.

In contrast, our results suggest that assimilation disparities in volunteering are partly related to socioeconomic disparities measured by education, income, employment status, and self-rated health. For example, the difference between the predicted probabilities of volunteering for long-time residing immigrants from Protestant and Muslim countries decreases by about seven percentage points from approximately 18 percentage points in Model 1 to approximately 11 percentage points in Model 4. The difference between the predicted probability of volunteering for long-time residing immigrants from Protestant countries and HBO countries likewise decreases. Nevertheless, the differences in the predicted probabilities remain present and statistically significant even when we control for socioeconomic, migration-related, and country-level factors.

Supplementary analyses and robustness checks

We conducted additional analyses and robustness checks to supplement our main results. First, we reran our analysis with ten years or more as our cutoff value to examine the sensitivity of our binary measure of length of residence. The results of these analyses are essentially like those we present in our primary analysis (see Table C1 in the online supplementary materials).

Second, because immigrants from Protestant countries in Denmark mainly came from neighboring countries such as Germany, Sweden, and Norway (see Table 2), we conducted additional analyses to include only immigrants from non-Western countries. This additional analysis allowed us to assess whether immigrants from non-Western Protestant countries are as likely to volunteer as their Western counterparts. Notably, these analyses suggest that immigrants from non-Western Protestant countries are almost as likely to volunteer as immigrants from Western Protestant countries are (see Table C2 in the online supplementary material). Additionally, immigrants from non-Western Protestant countries, like their Western counterparts, reach parity with natives in their level of volunteering during the years they spend in Denmark (see Table C2 and Figure C1 in the supplementary material). Interestingly, the estimates also indicate that the probability that immigrants from non-Western, non-Protestant Christian countries will volunteer is very high. Our results suggest that after 20 years of residence, non-Western immigrants from non-Protestant Christian countries also reach parity with natives in their level of volunteering (see Table C2, Table C3 and Figure C4 in the supplementary material). While we consider these results intriguing, they should be considered provisional because our dataset only includes a modest sample of immigrants from non-Western Protestant countries (n = 55) and non-Western non-Protestant Christian countries (n = 230).

Last, we replicated our results with a modified dependent variable that only measures volunteering in secular organizations. Excluding religion-based volunteering from our dependent variable makes the differences between the immigrant groups slightly less pronounced (see results in Tables C4 and C5 in supplementary material). However, these modest changes do not alter the overall conclusions.

Conclusion and discussion

The influx of immigrants from the Global South in North America and Europe has spurred extensive research on how quickly they become incorporated into various spheres of their destination societies. While much of this research has focused on educational and labor market assimilation, we have focused on a less explored aspect: how quickly immigrants assimilate to native-level volunteering in organizations.

Our findings reveal that immigrants from different groups exhibit varying levels of volunteering upon arriving in Denmark and experience different assimilation rates into volunteering. Immigrants from Protestant and other Christian countries exhibit relatively high levels of volunteering upon arrival, and those who have stayed in Denmark for many years are on par with natives. However, this pattern does not apply to other immigrant groups.

For immigrants from Muslim countries, we find that they are almost half as likely as natives to volunteer upon arrival, and we find no evidence that those who have stayed for many years are more likely to volunteer than those who have stayed fewer years. This pattern suggests that immigrants from these countries do not assimilate to native-level volunteering.

For immigrants from Hindu, Buddhist, and other countries, we find that they are much less likely to volunteer than natives upon arrival. However, those who have stayed for many years are significantly more likely to volunteer than those who have recently arrived. This pattern indicates that immigrants from these countries gradually assimilate to native-level volunteering over the years they spend in Denmark, but they never reach parity with natives.

Our central finding that only immigrants from Christian countries reach parity with natives in their level of volunteering in Denmark runs counter to previous cross-national European research. This research has suggested that immigrants of all ethnic backgrounds ultimately reach parity with natives in their level of civic participation, as determined by membership in voluntary associations or composite measures (Aleksynska 2011; Voicu and Şerban 2012).

There are different possible explanations for the discrepancies between the results of our study and those obtained in previous cross-national research in Europe. First, we rely on different measures for civic participation. Unlike our study, which focuses on volunteering, Aleksynska (2011), for example, used a composite variable that lumped together various types of civic participation. This difference in approach makes it difficult to compare our findings with that particular study directly. On the other hand, Voicu and Şerban (2012) focused on membership in voluntary organizations, which is closer to the measure we use. However, a significant difference is that we only consider volunteering for such organizations – not mere membership. This difference may appear subtle, but it might be crucial because recent longitudinal evidence suggests that active participation in civil society organizations, more so than passive membership, is related to the social integration of immigrants (Wang and Morav 2021).

Another possible reason our results from Denmark deviate from those obtained in previous cross-national European research could be that Denmark is an exceptional case. After all, the rates at which people volunteer in Denmark are high when considering Europe as a whole (Henriksen, Strømsnes, and Svedberg 2019). This situation implies that immigrants must volunteer at a high level to reach parity with natives. Moreover, a smaller share of the total amount of volunteering activities in Denmark occurs in religious organizations than in most other European countries, implying that immigrants must volunteer through secular organizations to reach parity with natives.

On the other hand, other arguments speak against Denmark as an exceptional case. First, although the extent to which people volunteer in Denmark is high, people in other Western European countries, such as the Netherlands, Sweden, and Norway, volunteer as much or more (Qvist et al. 2019). Moreover, Denmark does not appear to be an exceptional European case regarding anti-immigrant or anti-muslim attitudes (Strabac and Listhaug 2008).

Despite the valuable insights provided by our study, it is essential to acknowledge that it has limitations. One significant limitation is that although we demonstrate that immigrants from different groups assimilate to native-level volunteering at different speeds, we still do not know precisely why these variations exist. Although we point to socioeconomic differences as one reason for these assimilation disparities, further research is required to discover other influencing factors.

Another limitation is that we rely on cross-sectional rather than longitudinal data. Consequently, we cannot observe how the probability of individual immigrants volunteering changes during their time in Denmark. Instead, we base our analysis on comparing the volunteering behaviors of immigrants from different groups who have lived in Denmark for varying lengths of time. This approach has drawbacks because when analyzing variation across individuals and time, there are likely unobserved differences between the individuals beyond the factors we adjust for statistically. This issue is particularly pertinent given that immigrants are usually selected subpopulations of their origin countries’ populations (Ichou 2014). Therefore, although our findings reveal notable patterns and trends, different data types and analyses will be required to pinpoint their causes.

Despite these notable limitations, our study offers valuable insights with policy implications. Most importantly, it indicates that immigrant groups are unlikely to achieve equal participation in volunteer organizations merely over time without active efforts from public interventions. Effective inclusion will require deliberate strategies by policymakers, community leaders, and volunteer organizations. Researchers are critical to this effort, as they can identify the barriers preventing immigrant involvement in voluntary organizations, allowing for developing strategies specifically designed to address these barriers.

Statement of ethics

This study was approved by Statistics Denmark (Project No. 707945). Further, ethical approval from Aalborg University was not pursued, as it was not a prerequisite for this kind of research when the study was carried out.

Preprint

Please note that a pre-print version of this paper is available. The preprint has the following Doi: https://doi.org/10.31235/osf.io/n4ae2.

Acknowledgements

We owe special thanks to Dingeman Wiertz, Lars Skov Henriksen, Karen Nielsen Breidahl, Jeevitha Yogachandiran Qvist, Christian Albrekt Larsen, and Kasimir Dederichs for their helpful advice and comments on earlier drafts of this article.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability

The survey data we use in this paper is publicly available on request from the Danish Data Archive. However, the register data we use are not publicly available. These data are stored on double password protected remote access servers located at Statistics Denmark. A subset of these data can be made available to researchers that have been approved by Statistics Denmark for a specific research project that also has to be approved. Researchers who want to replicate our results, will have to apply for and be granted access to the data in question by Statistics Denmark.

Additional information

Funding

This project was supported by the Independent Research Fund Denmark [grant number: 0133-00041B].