Ethnic discrimination and the dynamics of belonging in Africa

Abstract

Using Afrobarometer survey data covering 36 countries (N ≈ 94,000), this study examines the relationship between experiences of ethnic discrimination and patterns of identification among Africans. The regression analysis indicates that as the frequency of ethnic discrimination increases, so does the likelihood that individuals will prioritize their ethnic identity over their national identity. Furthermore, the results demonstrate that the negative association between ethnic discrimination and exclusive national identification is substantially stronger than the positive association between ethnic discrimination and exclusive ethnic identification. These regression results are robust to the use of a binary measure of ethnic discrimination and to the employment of a different estimation method. One possible explanation for these findings is that experiences of ethnic discrimination foster feelings of exclusion from the broader national community, prompting individuals to reaffirm their attachment to their ethnic group as a source of belonging, support, and identity.

Keywords

ethnic discrimination identification national identity ethnicity Africa

Introduction

After gaining independence from colonial rule, many African countries experienced violent conflicts, often with a pronounced ethnic dimension (Ekwe-Ekwe, 1990; Greenland, 1976; Mazrui, 1969; Nnoli, 1974). Reflecting on the fragile unity of newly independent African states, anthropologist Clifford Geertz famously observed that “removing European rule has liberated the nationalisms within nationalisms” (Geertz, 1973: 237), underscoring the resurgence of ethnic identities following the withdrawal of colonial authority. Building on this insight, scholars have argued that postcolonial political dynamics often encourage citizens, political elites, and institutions to prioritize ethnic group interests over those of the broader nation (Ake, 1973; Bates, 1974; Osaghae, 1990). Consequently, ethnic hostilities remain a persistent challenge across many African countries (e.g. Beseng et al., 2023; Eke, 2022; Kefale, 2013; Nyaura, 2018; Onyemechalu and Ejiofor, 2026). Against this backdrop of persistent ethnic tensions, examining the social processes that sustain such divisions remains critical.

This exploratory study shifts the focus away from ethnic violence to ethnic discrimination and its implications for identity formation. Using large-scale Afrobarometer survey data from 36 African countries (N ≈ 94,000), it examines whether experiences of ethnic discrimination reduce the likelihood of individuals identifying more strongly with their national identity than with their ethnic identity. The study further investigates whether disidentification with the nation intensifies as the frequency of ethnic discrimination increases. Understanding how ethnic discrimination shapes identity formation is crucial because ethnic discrimination is far from benign. As Allport (1954) argues, discrimination represents a critical stage in the escalation of prejudice, forming part of a broader trajectory that can culminate in overt hostility and violence toward ethnic out-groups. In The Nature of Prejudice, Allport (1954) outlines five escalating stages of prejudice—antilocution, avoidance, discrimination, physical attack, and ultimately extermination—each more severe than the last.

A growing body of scholarship has examined the determinants of ethnic and national identification. In a cross-country study covering 16 African countries, Robinson (2014) shows that educational attainment, urban residence, and ethnic diversity increase the likelihood of individuals feeling closer to their nationality than to their ethnicity. Focusing on Nigeria, one of Africa’s most ethnically diverse countries, Tuki (2026) finds that experiences of ethnic discrimination weaken national identification while strengthening ethnic identification, with particularly strong effects observed among members of the Igbo ethnic group, who have historically experienced marginalization from central political power. Beyond the African context, Hakim et al. (2018) demonstrate that ethnic discrimination among Arab Americans is associated with lower levels of American identification and stronger ethnic and religious identification. Similarly, research on Latin American immigrants in Spain finds that experiences of discrimination are negatively correlated with feelings of national belonging (Lobera, 2021).

The empirical patterns observed in these studies can be interpreted through Andreas Wimmer’s concept of boundary-making, which is central to identity formation (Wimmer, 2008, 2013). Wimmer challenges the notion that ethnic identity is intrinsic, arguing instead that it is relational. Boundary-making, a cognitive process, creates distinctions between “us” and “them,” clearly identifying those who belong to an in-group and those who do not. These distinctions, in turn, shape individuals’ behavior toward those perceived to be on their side and those on the other side of the boundary. When individuals categorize others and treat them differently on the basis of characteristics such as ethnicity or religious affiliation, they are effectively reinforcing boundaries—that is, boundaries are enacted through differential treatment. Importantly, Wimmer also emphasizes that power asymmetries are crucial to the construction of boundaries: dominant groups may impose exclusionary boundaries on minorities, which can, in turn, trigger mobilization among subordinate groups seeking to defend their interests.

Furthermore, two complementary theoretical frameworks—the “Rejection–identification” and “Rejection–disidentification” models—provide additional leverage for interpreting these dynamics. The Rejection–identification model posits that experiences of ethnic discrimination strengthen ethnic identification by prompting individuals to rely more heavily on co-ethnics as a source of social support (Branscombe et al., 1999). At the same time, the Rejection–disidentification model suggests that such experiences may weaken identification with the nation, as ethnic discrimination within one’s own country signals exclusion from the national community (Jasinskaja-Lahti et al., 2009; Wu and Finnsdottir, 2021). Taken together, these frameworks offer clear expectations about how ethnic discrimination may simultaneously reshape both ethnic and national attachments. This study tests whether these theories hold in the broader African context.

To situate this contribution within the existing literature, it is important to distinguish the present study from those conducted by Robinson (2014) and Tuki (2026), which also rely on Afrobarometer data. First, this study uses the more recent round 7 and 8 data, covering 36 countries (N ≈ 94,000), unlike Robinson’s, which is based on round 3 covering only 16 countries (n = 22,155), and Tuki’s, which focuses solely on the Nigerian case. Second, Robinson’s work does not specifically examine how experiences of ethnic discrimination shape identification. Although the study by Tuki (2026) does address this issue, its measure of ethnic discrimination is collapsed into a binary variable, thereby obscuring potential variation in the intensity of those experiences. While the present study also employs a binary measure of ethnic discrimination for comparative purposes, it goes a step further by treating ethnic discrimination as a factor variable and explicitly modeling how changes in its frequency influence identification.

Before proceeding to the analysis, it is also important to clarify the concept of ethnic discrimination employed throughout this paper. Fibbi et al. (2021: 19) define discrimination as “the unequal treatment of otherwise similar individuals due to their ascribed membership in a disadvantaged category or group.” They observe that for discrimination to occur, a reference category is necessary, as an individual must be treated unfairly compared to another. They further note that the criterion by which an individual is treated unfairly is their ascribed membership in a group such as race, ethnic origin, or color—traits that cannot easily be changed. These conditions are consistent with the observation of Auer and Ruedin (2019: 222): “For discrimination to occur, we need at least two actors. One of these actors unfairly treats the other based on an irrelevant criterion such as ethnicity, country of origin, or gender.” Allport (1954) observed that discrimination is rooted in prejudice, the latter of which he defined as “an aversive or hostile attitude towards a person or group, simply because they belong to that group, and are therefore presumed to have the objectionable qualities ascribed to the group” (p. 7).

Within the context of the present study, I adopt a subjective, self-reported measure of ethnic discrimination. Specifically, I rely on an item asking respondents how frequently they were treated unfairly based on their ethnicity in the past year, with responses recorded on a Likert-type scale ranging from “never” to “always.” A particular strength of this approach is that it respects the subjective reality of the individual.¹ However, this self-reporting approach also has the limitation that it may underestimate the scale of discrimination. This is because, “self-reports depend on a person being aware of being treated unfairly, but discrimination may be completely unnoticed” (Smith, 2002: 9–10).

Having outlined the study’s contribution and conceptual framework, the remainder of the paper is organized as follows: the next section introduces the data, describes the variables used in the regression analysis, and outlines the analytical technique, alongside a descriptive overview of the data. The third section presents and interprets the regression results, while the final section summarizes the main findings and provides concluding remarks.

Data and methodology

Data

This study draws on data from rounds 7 and 8 of the Afrobarometer surveys, conducted across 36 African countries between 2016 and 2021.² Table 4 in the Appendix lists the countries and the total number of observations drawn from each. The round 7 and 8 surveys include 45,823 and 48,084 observations, respectively, yielding a combined sample of 93,907 respondents. All participants were at least 18 years old, with an equal gender distribution (50:50 male-to-female ratio). Although data from the more recent round 9 survey (conducted between 2021 and 2023) are publicly available, they were excluded from this analysis because the question used to construct the explanatory variable was not included in that round. A limitation of the Afrobarometer dataset is its exclusion of certain countries experiencing conflict—such as Libya, the Central African Republic, and the Democratic Republic of Congo. However, it is notable that some countries with a high incidence of conflict such as Mali, Nigeria, Burkina Faso, and Ethiopia, are included in the sample. Because Afrobarometer employs probabilistic sampling methods, the data are nationally representative for each of the 36 countries included in the survey.³

Measures

Dependent variable

Nationality > Ethnicity measures the degree to which respondents identify with their nationality relative to their ethnicity. The variable was derived from the following question:

Let us suppose that you had to choose between being a [Respondents nationality] and being a [Respondent’s ethnic group]. Which of the following statements best expresses your feelings?

Responses were recorded using a five-point ordinal scale with the following categories:

1 = I feel only [Respondent’s ethnic group]

2 = I feel more [Respondent’s ethnic group] than [Respondents nationality]

3 = I feel equally [Respondent’s nationality] and [Respondent’s ethnic group]

4 = I feel more [Respondent’s nationality] than [Respondent’s ethnicity]

5 = I feel only [Respondent’s nationality]

I treated “Don’t know” and “Refused to answer” responses as missing observations, applying this rule to all variables. Figure 1 depicts the dependent variable using a simple bar chart. A glance at the figure shows that the majority of Africans (45%) identify equally with their ethnicity and nationality. Fourteen percent either feel an exclusive sense of belonging to their ethnicity or feel closer to their ethnicity than to their nationality. Some 9% identify more with their nationality than their ethnicity, while 32% feel an exclusive sense of belonging to their nationality.

To better assess the extent of ethnic identification among Africans, I calculated the share of respondents in each country who either identify exclusively with their ethnic group or report a stronger attachment to their ethnic identity than to their national identity. Figure 2 presents the 12 countries with the highest levels of ethnic identification. South Africa ranks first, with 26% of respondents exhibiting strong ethnic identification, followed by Ethiopia at 24%. Nigeria (22%) and Mozambique (20%) occupy the third and fourth places, respectively, while Benin and Mali are tied for fifth, each with 19% of respondents expressing strong ethnic identification. Table 4 in the Appendix shows the levels of ethnic identification across all 36 countries in the sample.

Explanatory variable

Ethnic discrimination measures how often respondents experienced ethnic discrimination over the past year. The variable was derived from the question: “In the past year, how often, if at all, have you personally been discriminated against based on any of the following: your ethnicity?” Responses were recorded on a four-point ordinal scale ranging from “0 = Never” to “3 = Many times.” Using “Never” as the reference category, I constructed dummy variables for the remaining three categories: “Once or twice,” “Several times,” and “Many times.” For example, the variable Once or twice is coded as 1 if a respondent experienced ethnic discrimination once or twice during the past year, and 0 if they either did not experience discrimination or experienced it several times or many times. Figure 3 presents the distribution of the explanatory variable using a bar chart. A glance at the figure reveals that the vast majority of Africans (82%) reported no experience of ethnic discrimination, while 18% reported experiencing it at least once or twice. Due to the strong clustering around the “Never” category, I also created a binary version of the explanatory variable—Ethnic discrimination (Binary)—for use in a robustness check. In this version, respondents who reported never experiencing ethnic discrimination are coded as 0, while those who reported any experience of such discrimination, regardless of frequency, are coded as 1.

To assess how widespread experiences of ethnic discrimination are across Africa, I calculated the share of respondents in each country who reported experiencing ethnic discrimination based on ethnicity at least once in the past year. Figure 4 presents the 12 countries with the highest incidence of self-reported ethnic discrimination. Ethiopia, Angola, and Cameroon are jointly ranked first, with 33% of respondents in each country reporting at least one experience of ethnic discrimination during the previous year. Nigeria follows in fourth place at 31%, while Uganda, Namibia, and Gabon are tied for fifth, each with 29% of respondents reporting ethnic discrimination. Table 4 in the Appendix shows the level of ethnic discrimination across the 36 countries in the sample.

Control variables

Trust president is derived from the question, “How much do you trust each of the following, or haven’t you heard enough about them to say? The President.” Responses were measured on a four-point Likert-type scale ranging from “0 = Not at all” to “3 = A lot.” Individuals who have high levels of trust in the central government may identify more strongly with their nationality than with their ethnicity because they perceive the government as legitimate, impartial, and effective. This, in turn, reduces their reliance on ethnic networks and may also weaken the salience of their ethnic identity. Previous research has found a positive correlation between government trust and national identification (e.g. Lenard and Miller, 2018; Yin and Zhang, 2025).

Rural is coded as 1 if a respondent lives in a rural area and 0 if they reside in an urban center. Previous research shows that urban residents are more likely to prioritize their national identity over their ethnic one (Robinson, 2014).

Education level

The original variable is measured on a scale with ten ordinal categories ranging from “0 = No formal schooling” to “9 = Postgraduate education.” I recoded this variable into four categories: No education, Primary, Secondary, and Tertiary. No education is coded as 1 for respondents with no formal schooling or who attended only informal institutions, such as Koranic schools, and 0 otherwise. Primary education is coded as 1 if the respondent attained some primary education or completed primary school, and 0 otherwise. Secondary education is coded as 1 if the respondent attained some secondary education or completed secondary school, and 0 otherwise. Tertiary education is coded as 1 if the respondent has any form of post-secondary education (e.g. a diploma, university degree, or postgraduate training), and 0 otherwise. Using respondents with no education as the reference category, I include the remaining three education categories as dummy variables in the regression models. Consistent with the contact hypothesis (Allport, 1954), education may weaken ethnic identification by increasing contact between individuals and members of diverse ethnic out-groups.

Demographic covariates

Because the dependent and explanatory variables are measured at the individual level, I include basic covariates such as age and gender. Gender is coded as 1 for male and 0 for female, while age is recorded in years.

Table 1 presents the summary statistics for all variables used in the regression analyses. The dependent variable Nationality > Ethnicity has a relatively large number of missing observations compared to the other variables because the question from which it was derived was not asked in Sudan (n = 3000) and Tunisia (n = 2399), resulting in the loss of 5399 observations. In addition, an earlier filter question asked respondents to identify their ethnic group; those who either refused to answer or did not identify themselves in ethnic terms were not subsequently asked about the strength of their national versus ethnic identification. This pattern is particularly pronounced in countries such as Eswatini, São Tomé and Príncipe, and Cabo Verde, where a substantial share of respondents reported that they do not view themselves in ethnic terms and were therefore excluded from the dependent variable measure. Notably, this exacerbated the issue of listwise deletion in the regression models.

Table 1.

Descriptive statistics.

Variable	Number of observations	Missing values	Mean	Standard deviation	Minimum	Maximum
Nationality > Ethnicity^σ	81,678	12,229	3.52	1.194	1	5
Ethnic discrimination
Never (Ref)	91,934	1973	0.823	0.381	0	1
Once or twice	91,934	1973	0.074	0.263	0	1
Several times	91,934	1973	0.069	0.253	0	1
Many times	91,934	1973	0.034	0.18	0	1
Ethnic discrimination (binary)	91,934	1973	0.177	0.381	0	1
Trust president	90,758	3149	1.657	1.166	0	3
Rural (Ref: Urban)	93,867	40	0.55	0.498	0	1
Education level
No education (Ref)	93,453	454	0.198	0.398	0	1
Primary	93,453	454	0.282	0.45	0	1
Secondary	93,453	454	0.362	0.481	0	1
Tertiary	93,453	454	0.158	0.365	0	1
Male (Ref: Female)	93,907	0	0.5	0.5	0	1
Age	93,849	58	37.105	14.863	18	120

σ indicates the dependent variable. “Ref” is the reference category. The values are based on data from rounds 7 and 8 of the Afrobarometer surveys conducted between 2016 and 2021. “Missing values” denote the number of respondents who were not asked the question of interest, chose “don’t know” responses, or declined to answer the question. Given that the total sample comprises 93,907 observations, the number of missing cases for each variable was calculated by subtracting the number of valid responses from that.

Analytical technique

To examine the relationship between ethnic discrimination and identification, I consider a model of the following general form:

γ_{i j t} = α_{0} + α_{1} E t h n i c d i s c r i m i n a t i o {n^{'}}_{i} + α_{2} {Λ^{'}}_{i} + Π_{j} + τ_{t} + μ_{i}

(1)

In this equation, $γ_{i j t}$ is the dependent variable, which measures the degree to which Respondent $i$ , living in Country $j$ at Year $t$ , identifies with their nationality relative to their ethnicity. $E t h n i c d i s c r i m i n a t i o {n^{'}}_{i}$ is a vector of explanatory variables measuring the frequency with which respondents experienced ethnic discrimination in the previous year; ${Λ^{'}}_{i}$ is a vector of control variables; $α_{0}$ is the constant term; $α_{1}$ is a vector of coefficients for the explanatory variables; $α_{2}$ is a vector of coefficients for the control variables; $Π_{j}$ and $τ_{t}$ denote fixed effects (FE) for the country in which respondents live and the survey year, respectively.

Country FE account for time-invariant characteristics unique to each country—such as colonial history, geography, and cultural norms—that might influence identification patterns. Year FE capture shocks that affect all observations over time, such as global economic trends or regional political developments. Finally, $μ_{i}$ represents the error term.

Because the dependent variable is measured on an ordinal scale with few categories, I estimate the model using an ordered logit regression. This approach is appropriate because it accounts for the ordinal nature of the dependent variable and enables the estimation of associations between the explanatory variables and each category of the outcome. I also conduct a robustness check using ordinary least squares (OLS) regression as an alternative estimation method. To account for potential correlations between observations within the same country, standard errors are clustered at the country level.

Results and discussion

Table 2 presents the regression results. As earlier noted, the reference category for the three explanatory variables is the subsample of respondents who reported no experience of ethnic discrimination in the past year. In model 1, which includes only the explanatory variables, all coefficients are negative and statistically significant at the 1% level. This indicates that experiencing ethnic discrimination—regardless of frequency—reduces the likelihood that respondents identify more strongly with their nationality than with their ethnicity. In other words, exposure to ethnic discrimination shifts identification away from the nation and toward ethnicity. Moreover, a closer inspection of the coefficients reveals that their magnitudes increase with the frequency of ethnic discrimination. This suggests that the more often individuals experience ethnic discrimination, the greater their disidentification with national identity. A plausible explanation for this pattern is that ethnic discrimination fosters feelings of exclusion from the national community. Such exclusion can weaken attachment to national identity while strengthening identification with one’s ethnic group, which may provide emotional safety, belonging, and mutual support (e.g. Branscombe et al., 1999; Lindemann and Stolz, 2021; Sarigil and Fazlioglu, 2014; Tuki, 2026).

Table 2.

Ordered logit models regressing identification on ethnic discrimination in Africa.

Nationality > Ethnicity^σ	(1)	(2)	(3)	(4)
Ethnic discrimination
Never (Ref)
Once or twice	-0.208*** (0.045)	-0.209*** (0.044)
Several times	-0.391*** (0.071)	-0.398*** (0.068)
Many times	-0.438*** (0.092)	-0.434*** (0.09)
Ethnic discrimination (Binary)			-0.319*** (0.056)	-0.321*** (0.053)
Trust president		0.044** (0.02)		0.045** (0.02)
Rural (Ref: Urban)		-0.182*** (0.035)		-0.181*** (0.035)
Education level
No education (Ref)
Primary		0.157*** (0.044)		0.157*** (0.045)
Secondary		0.24*** (0.05)		0.241*** (0.049)
Tertiary		0.306*** (0.065)		0.308*** (0.064)
Male (Ref: female)		0.105*** (0.021)		0.105*** (0.021)
Age		0.002* (0.001)		0.002* (0.001)
Intercept 1	-2.607*** (0.148)	-2.411*** (0.158)	-2.602*** (0.148)	-2.405*** (0.158)
Intercept 2	-1.75*** (0.139)	-1.55*** (0.149)	-1.746*** (0.139)	-1.545*** (0.148)
Intercept 3	0.58*** (0.116)	0.795*** (0.124)	0.583*** (0.117)	0.799*** (0.124)
Intercept 4	1.017*** (0.12)	1.237*** (0.128)	1.021*** (0.121)	1.241*** (0.128)
Year FE	Yes	Yes	Yes	Yes
Country FE	Yes	Yes	Yes	Yes
Observations	81203	78617	81203	78617
Pseudo R ²	0.037	0.04	0.037	0.039
Log pseudolikelihood	-102807.5	-99243.641	-102828.01	-99263.76
AIC statistic	205635	198521.3	205672	198557.5
BIC statistic	205728	198678.9	205746.5	198696.6

σ denotes the dependent variable, which measures the degree to which respondents identify with their nationality relative to their ethnicity. Clustered robust standard errors are in parentheses. All models are estimated using an ordered logit regression. “Ref” denotes the reference category; “FE” denotes fixed effects. AIC = Akaike information criterion; BIC = Bayesian information criterion. The regression models are based on pooled data from rounds 7 and 8 of the Afrobarometer surveys conducted between 2016 and 2021.

***

p < 0.01, **p < 0.05, *p < 0.10.

In model 2, the inclusion of control variables does not alter the direction or the statistical significance of the ethnic discrimination coefficients. All control variables are statistically significant. Trust in the president has a positive coefficient, indicating that higher levels of trust in the central government are associated with stronger national identification relative to ethnic identification. This likely reflects a perception of the state as legitimate, inclusive, impartial, and effective, which reduces the salience of ethnic identity. When individuals view the state as capable of providing public goods and security, they are less reliant on ethnic groups to fulfill these functions, thereby diminishing the instrumental value of ethnicity. This finding is consistent with studies conducted in China (Yin and Zhang 2025) and Nigeria (Tuki, 2024, 2026). Another plausible explanation for this result is that it reflects advantageous treatment by the state toward certain groups, which induces trust in the central government and ultimately fosters stronger national identification.

Rural residency has a negative coefficient. This indicates that, compared to individuals living in urban centers, rural residents are less likely to prioritize their national identity over their ethnicity. A plausible explanation for this finding is that social, economic, and political life in rural areas is often organized around customary institutions rather than national ones, resulting in individuals being strongly embedded in ethnic and kinship networks (Gurgler and Flanagan, 1978; Hao et al., 2025; Wicomb and Smith, 2011). Moreover, individuals in rural areas may have more limited exposure to national institutions such as social services and public goods. In this context, the lower visibility of the state may render national identity less salient than ethnic identity. Notably, this finding is consistent with Robinson (2014).

Educational attainment is also positively associated with national identification. Compared to individuals with no formal education, those with primary, secondary, and tertiary education are increasingly more likely to identify with their nationality than with their ethnicity. The coefficients grow monotonically with educational level, suggesting that national identification strengthens as individuals attain higher levels of education. One plausible mechanism is that education increases cross-ethnic contact, which can weaken ethnic attachment and reinforce national identity, consistent with the contact hypothesis (Allport, 1954; Pettigrew, 1998). Moreover, education may expose individuals to civic education, which in turn strengthens national patriotism.

The gender indicator carries a positive coefficient, suggesting that men are more likely than women to prioritize national over ethnic identity. This pattern may be linked to gendered access to education and participation in the public sphere. In patriarchal contexts where men have greater exposure to education, politics, protest activity, and labor markets, they are more likely to interact across ethnic lines. By contrast, women’s concentration in domestic roles may limit such exposure, reinforcing more localized or ethnic forms of identification. Age is also positively associated with national identification, indicating that older individuals are more likely to prioritize nationality over ethnicity.

Finally, models 3 and 4 show that the main results are robust to the use of a binary operationalization of ethnic discrimination, in which the three frequencies (once or twice, several times, and many times) are collapsed into a single category. Furthermore, the results are also robust to the use of OLS regression as an alternative estimation method (see Table 3 in the Appendix).

To illustrate the strength of the associations reported in Table 2, I plotted the predicted probabilities in Figure 5. A glance at the four panels shows that the association between ethnic discrimination and identification is strongest in the fifth response category of the dependent variable, where respondents report an exclusive sense of belonging to their nationality. By contrast, the association is weakest in the fourth category, where respondents indicate that they identify more strongly with their nationality than with their ethnicity. Notably, as shown at both extremes of the response scale, the negative association between ethnic discrimination and an exclusive national identity is substantially stronger than the positive association between ethnic discrimination and an exclusive ethnic identity.

Panel A of Figure 5 shows that, compared to the reference category (i.e. respondents who did not experience ethnic discrimination), those who experienced ethnic discrimination once or twice are 1.2 percentage points more likely to identify exclusively with their ethnicity and 4.2 percentage points less likely to identify exclusively with their nationality. Panel B indicates that individuals who experienced ethnic discrimination several times are 2.3 percentage points more likely to feel an exclusive sense of ethnic belonging and 7.9 percentage points less likely to feel an exclusive sense of national belonging. Panel C reveals that those who experienced ethnic discrimination many times are 2.6 percentage points more likely to identify exclusively with their ethnicity and 8.8 percentage points less likely to identify exclusively with their nationality. Notably, and consistent with the regression results in models 1 and 2, the magnitude of these associations at both extremes is smallest among those who experienced ethnic discrimination only once or twice and increases steadily with the frequency of ethnic discrimination. Finally, Panel D, which is based on the binary measure of ethnic discrimination, shows that any experience of ethnic discrimination—regardless of frequency—is associated with a 1.9 percentage point increase in exclusive ethnic identification and a 6.4 percentage point decrease in exclusive national identification.

Conclusion

This exploratory study examined the relationship between ethnic discrimination and identification using survey data from 36 African countries. The regression analysis shows that, as the frequency of experiencing ethnic discrimination rises, the likelihood that individuals identify more strongly with their national identity than with their ethnic identity steadily declines. In other words, ethnic discrimination appears to encourage individuals to prioritize their ethnic identity over their national identity. The results further indicate that the negative association between ethnic discrimination and exclusive national identification is substantially stronger than the positive association between ethnic discrimination and exclusive ethnic identification. One plausible explanation for these results is that individuals who experience ethnic discrimination come to feel excluded from the broader national community. In response, they may retreat into their ethnic group as a source of solidarity, support, and belonging, thereby reinforcing their attachment to that identity.

These findings carry important policy implications. They suggest that policymakers should prioritize initiatives that promote an inclusive national identity—one that recognizes and values ethnic diversity—in order to reduce feelings of exclusion among marginalized groups. In addition, strengthening anti-discrimination laws and ensuring their effective enforcement are essential. By addressing ethnic discrimination, governments can help prevent polarization and foster a sense of belonging among all citizens. Finally, improving access to education and promoting interethnic dialogue can engender intergroup trust and mutual understanding, thereby reducing the tendency for individuals to retreat into exclusive ethnic identities in response to perceived marginalization.

It is important to note that this study is correlational and does not make causal claims. Future research should focus on examining these relationships using experimental and quasi-experimental approaches that attenuate the potential problem of confounders and enable causal identification. Moreover, future research could also examine these relationships using qualitative approaches such as in-depth interviews and focus group discussions, as this would provide a deeper understanding of individual experiences, shedding more light on the mechanisms underlying the observed statistical relationships.

Footnotes

Appendix

Table 4.

Countries surveyed and percentages of their populations that identify strongly with their ethnicity and have experienced ethnic discrimination.

Country	Number of observations	Ethnic identification (%)	Experienced ethnic discrimination (%)
Benin	2400	19	20
Botswana	2398	14	14
Burkina Faso	2400	11	8
Cabo Verde	2400	7	9
Cameroon	2402	16	33
Cote d’Ivoire	2400	13	15
eSwatini	2400	14	8
Gabon	2399	12	29
Gambia	2400	10	15
Ghana	4800	10	16
Guinea	2394	9	20
Kenya	3999	8	28
Lesotho	2400	17	5
Liberia	2400	10	23
Madagascar	1200	13	5
Malawi	2400	14	18
Mali	2400	19	8
Mauritius	2400	14	22
Morocco	2400	5	12
Mozambique	3502	20	20
Namibia	2400	15	29
Niger	2399	16	10
Nigeria	3199	22	31
Sao Tome and Principe	1200	10	5
Senegal	2400	12	7
Sierra Leone	2400	12	12
South Africa	3440	26	24
Sudan^†	3000	XX	21
Tanzania	4798	6	4
Togo	2400	12	22
Tunisia^†	2399	XX	6
Uganda	2400	18	29
Zambia	2400	12	20
Zimbabwe	2400	16	15
Angola	2400	17	33
Ethiopia	2378	24	33
Total	93,907

The first and second columns list the countries in the sample and the total number of observations collected from each. The third column shows the percentage of the population in the respective countries that identifies exclusively with their ethnic group or more strongly with their ethnic group than with their nationality. The fourth column shows the percentage of the population in each country that experienced discrimination based on their ethnicity at least once during the previous year.

†

Countries for which the question on national versus ethnic identification was not asked. The table is based on data from rounds 7 and 8 of the Afrobarometer survey, conducted across 36 countries between 2021 and 2023.

Acknowledgements

I thank the handling editor, Prof. David Norman, two anonymous reviewers, Jeffery Conroy-Krutz and Brian Howard, for their helpful comments.

ORCID iD

Daniel Tuki

Funding

The author disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Financial support from the state of Hessen via the DynaRel project is gratefully acknowledged.

Declaration of conflicting interests

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

Data availability statement

The replication files underlying the regression results are available in the Harvard Dataverse:

Notes

Author biography

Daniel Tuki’s research focuses on conflict studies and development economics. His work has appeared in International Studies Quarterly, Democratization, Terrorism and Political Violence, Environment and Development Economics, and Defence and Peace Economics among others.

References

Ake

(1973) Explaining political instability in new states. Journal of Modern African Studies 11(3): 347–359. https://doi.org/10.1017/S0022278X00008478

Allport

(1954) The Nature of Prejudice. The Beacon Press.

Auer

Ruedin

(2019) Who feels disadvantaged? Reporting discrimination in surveys. In: Steiner

Wanner

(eds) Migrants and Expats: The Swiss Migration and Mobility Nexus. Springer Nature, pp.221–242.

Bates

(1974) Ethnic competition and modernization in contemporary Africa. Comparative Political Studies 6(4): 457–484. https://doi.org/10.1177/001041407400600403

Beseng

Crawford

Annan

(2023) From “Anglophone problem” to “Anglophone conflict” in Cameroon: Assessing prospects for peace. Africa Spectrum 58(1): 89–105. https://doi.org/10.1177/00020397231155244

Branscombe

Schmitt

Harvey

(1999) Perceiving pervasive discrimination among African Americans: Implications for group identification and well-being. Journal of Personality and Social Psychology 77(1): 135–149. https://doi.org/10.1037/0022-3514.77.1.135

Dilmaghani

(2019) Perceived religious discrimination and confidence in Canadian institutions. Ethnic and Racial Studies 42(15): 2603–2622. https://doi.org/10.1080/01419870.2018.1539505

Eke

(2022) Why does segregation prevent conflict in some regions but not others? Interrogating social distance amid ethnic conflicts in Jos, Nigeria. Third World Quarterly 43(9): 2209–2224. https://doi.org/10.1080/01436597.2022.2085548

Ekwe-Ekwe

(1990) Conflict and Intervention in Africa: Nigeria, Angola, Zaire. Palgrave Macmillan.

10.

Fibbi

Midtbøen

Simon

(2021) Migration and Discrimination: IMISCOE Short Reader. Springer Nature.

11.

Geertz

(1973) The Interpretation of Cultures: Selected Essays. Basic Books.

12.

Greenland

(1976) Ethnic discrimination in Rwanda and Burundi. In: Veenhoven

(ed.) Case Studies on Human Rights and Fundamental Freedoms. Brill, pp.95–133.

13.

Gurgler

Flanagan

(1978) Urbanisation and Social Change in West Africa. Cambridge University Press.

14.

Hakim

Molina

Branscombe

(2018) How discrimination shapes social identification processes and well-being among Arab Americans. Social Psychological and Personality Science 9(3): 328–337. https://doi.org/10.1177/1948550617742192

15.

Hao

Chen

Wang

(2025) Changing customs: The practice of cultivating civilized rural customs through co-creation. In: Wang

Hao

Chen

(eds) Co-creation: Modernizing grassroots governance in China. Springer. https://doi.org/10.1007/978-981-96-9467-9_2

16.

Jasinskaja-Lahti

Liebkind

Solheim

(2009) To identify or not to identify? National disidentification as an alternative reaction to perceived ethnic discrimination. Applied Psychology 58(1): 105–128. https://doi.org/10.1111/j.1464-0597.2008.00384.x

17.

Kefale

(2013) Federalism and Ethnic Conflict in Ethiopia: A Comparative Regional Study. Routledge.

18.

Lenard

Miller

(2018) Trust and national identity. In: Uslaner

(ed.) Oxford Handbook of Social and Political Trust. Oxford University Press, pp.57–74. https://doi.org/10.1093/oxfordhb/9780190274801.013.36

19.

Levitt

(2015) Discrimination and the distrust of democratic institutions in Latin America. Politics, Groups, and Identities 3(3): 417–437. https://doi.org/10.1080/21565503.2015.1050410

20.

Lindemann

Stolz

(2021) Perceived discrimination among Muslims and its correlates. A comparative analysis. Ethnic and Racial Studies 44(2): 173–194. https://doi.org/10.1080/01419870.2020.1729389

21.

Lobera

(2021) Postcolonial bonds? Latin American origins, discrimination, and sense of belonging to Spain. American Behavioral Scientist 65(9): 1222–1233. https://doi.org/10.1177/0002764221996757

22.

Mazrui

(1969) Violent contiguity and the politics of retribalization in Africa. Journal of International Affairs 23(1): 89–105. https://www-jstor-org-443.web.bisu.edu.cn/stable/24356594

23.

Nguyen

Ngo

Lee

(2025) Perceived discrimination and sense of community belonging among Canadian young adults: A population-based study. Canadian Ethnic Studies 57(1): 1–27. https://doi.org/10.1353/ces.2025.a960309

24.

Nnoli

(1974) Socio-economic insecurity and ethnic politics in Africa. African Review 4(1): 1–23.

25.

Nyaura

(2018) Devolved ethnicity in the Kenya: Social, economic and political perspective. European Review of Applied Sociology 11(16): 17–26. https://doi.org/10.1515/eras-2018-0002

26.

Onyemechalu

Ejiofor

(2026) Igbophobia as racism: The making and unmasking of racial discrimination in Nigeria. Nationalism and Ethnic Politics. Epub ahead of print 22 January. https://doi.org/10.1080/13537113.2026.2616953

27.

Osaghae

(1990) The problems of citizenship in Nigeria. Rivista trimestrale di studi e documentazione dell’Istituto italiano perl’Africa e l’Oriente 45(4): 593–611. https://www-jstor-org-443.web.bisu.edu.cn/stable/

28.

Pettigrew

(1998) Intergroup contact theory. Annual Review of Psychology 49(1): 65–85. https://doi.org/10.1146/annurev.psych.49.1.65

29.

Robinson

(2014) National versus ethnic identification in Africa: Modernization, colonial legacy, and the origins of territorial nationalism. World Politics 66(4): 709–746. https://doi.org/10.1017/S0043887114000239

30.

Sarigil

Fazlioglu

(2014) Exploring the roots and dynamics of Kurdish ethno-nationalism in Turkey. Nations and Nationalism 20(3): 436–458. https://doi.org/10.1111/nana.12058

31.

Smith

(2002) Measuring Racial and Ethnic Discrimination. Report no. 96. General Social Survey Project.

32.

Straiton

Aambø

Johansen

(2019) Perceived discrimination, health and mental health among immigrants in Norway: The role of moderating factors. BMC Public Health 19(1): 1–13. https://doi.org/10.1186/s12889-019-6649-9

33.

Tuki

(2024) Undead past: What drives support for the secessionist goal of the Indigenous People of Biafra (IPOB) in Nigeria? Journal of Race, Ethnicity and Politics 9(1): 26–54. https://doi.org/10.1017/rep.2023.36

34.

Tuki

(2026) You’re not like us! Ethnic discrimination and national belonging in Nigeria. Self and Identity 25(2): 368–392. https://doi.org/10.1080/15298868.2025.2603518

35.

Wicomb

Smith

(2011) Customary communities as ‘peoples’ and their customary tenure as ‘culture’: What we can do with the Endorois decision. African Human Rights Law Journal 11(2): 422–446. https://hdl.handle.net/10520/EJC51956

36.

Wilkes

(2019) Immigration, discrimination, and trust: A simply complex relationship. Frontiers in Sociology 4(32): 1–13. https://doi.org/10.3389/fsoc.2019.00032

37.

Wimmer

(2008) The making and unmaking of ethnic boundaries: A multilevel process theory. American Journal of Sociology 113(4): 970–1022. https://doi.org/10.1086/522803

38.

Wimmer

(2013) Ethnic Boundary Making: Institutions, Power, Networks. Oxford University Press.

39.

Finnsdottir

(2021) Perceived racial and cultural discrimination and sense of belonging in Canadian society. Canadian Review of Sociology 58(2): 229–249. https://doi.org/10.1111/cars.12339

40.

Schimmele

(2021) Perceived religious discrimination and mental health. Ethnicity & Health 26(7): 963–980. https://doi.org/10.1080/13557858.2019.1620176

41.

Yin

Zhang

(2025) The role of government trust in shaping national identity: Empirical evidence from China. Sage Open 15(4): 1–16. https://doi.org/10.1177/21582440251399785