Experience of discrimination in egalitarian societies: the Sámi and majority populations in Sweden and Norway

ABSTRACT

The Sámi people stand out as the only Indigenous minority in an egalitarian European context, namely the Nordic Countries. Therefore, inequalities that they may face are worth closer inspection. Drawing on the distinction between inequalities among individuals (vertical) and between groups (horizontal), we investigate how different types of inequalities affect the Sámi today. We formulate a series of hypotheses on how social, economic, cultural, and political inequalities are linked with discrimination experience, and test these with original data from a population survey conducted in northern Norway and northern Sweden simultaneously in 2021. The findings show that Sámi ethnic background increases the probability of experiencing discrimination. While individual-level economic inequality is also pertinent, this does not directly materialise as between-group inequality. Instead, minority language use is a strong predictor of discrimination experience, revealing the socio-cultural nature of ethnic inequalities. Cross-country differences are only reflected in the effect of minority language use.

KEYWORDS:

• Discrimination
• inequality
• Indigenous minorities
• Sámi people
• minority languages
• Nordic countries

Introduction

Indigenous peoples differ from other ethnic minorities in that they have claims to specific rights stemming from their inhabitancy from time immemorial, such as rights to land use or self-determination through own institutions. Yet they may face problems comparable to other minorities such as socio-economic inequalities, discrimination, obstacles to effective democratic inclusion, assimilation attempts, and culture and language loss. While former colonies of European states receive the most attention in Indigenous studies, the Sámi stand out as the only Indigenous people in Europe: they are the original inhabitants of a territory that now spans across the mainland of three Nordic countries, which consistently rank highly in terms of the quality of their democratic institutions, their human rights records, and socio-economic equality. Exploring the extent of ethnic inequalities affecting Indigenous peoples in an egalitarian European context constitutes the main motive of this study.

We address the following questions: What kind of inequalities do the Sámi face today? Are these inequalities primarily economic, political, social, or cultural? And how do different types of inequalities overlap? With a focus on cultural inequalities in particular, to what extent does speaking a Sámi language play a role? We also investigate whether there are differences between Nordic countries, which are similar in their social democratic egalitarianism, in terms of how they address ethnic inequalities. Drawing on a theoretical reasoning that distinguishes between vertical inequalities among individuals and horizontal inequalities between groups, we argue that the inequalities that the Sámi face as a group are categorically different from individual-level inequalities, and these should display strong cultural and linguistic characteristics. Moreover, ethnic inequalities would be alleviated more successfully in countries where relevant policies take the horizontal dimension more seriously.

To support these claims empirically, we present original data from a population survey conducted in the northern regions of Norway and Sweden where the relative size of the Sámi population is highest. This dataset, consisting of 5416 responses on core questions and 1402 responses on a larger array of questions, allows us to examine economic, political, social, cultural, and linguistic inequalities, and to compare the Sámi with the majority and other minorities within their country of residence as well as across Norway and Sweden. As an embodiment of the effects of existing inequalities, we focus in this paper on the experience of discrimination and how it is associated with different types of inequality. Findings from statistical analyses lend support to the view that the patterns of discrimination that the Sámi experience go far beyond economic and individual-level inequalities, and these are manifested as between-group inequalities with substantial cultural and linguistic aspects. As a main contribution, we demonstrate the value of a comparative perspective with a simultaneous within-country and cross-country focus, as well as an interdisciplinary approach that examines multiple aspects of ethnic inequalities.

The following section provides background information on the Sámi people and reviews the relevant literature on Sámi politics and policies, as well as inequalities and discrimination. Next, we discuss horizontal inequalities and perceived discrimination to develop our argument and to derive a series of hypotheses. The subsequent section describes the details of data collection as well as the empirical strategy of the paper. This is followed by the presentation and discussion of main findings. The final section summarises and concludes the paper.

Ethnic inequalities and the Sámi

The Sámi people

The Sámi are an Arctic people originally inhabiting the Sápmi area, encompassing the northern parts of Norway, Sweden, Finland and the Kola Peninsula in Russia, and the only Indigenous people in Europe.¹ Due to the unavailability of census data, it is difficult to determine the exact size of the Sámi population. Estimations range from 47,000 to 65,000 in Norway, 20,000 to 40,000 in Sweden, 3500 to 8000 in Finland, and 1800 to 2000 in Russia.² There are nine Sámi languages, all of which are endangered, some even moribund. North Sámi, the Sámi language traditionally spoken in northern Norway, is the most vital and is said to account for around 75% of all Sámi speakers (Aikio-Puoskari 2018). Only a minority of the ethnic Sámi maintains a Sámi language, and many Sámi people today cannot speak a Sámi language at all, or speak it as a second language (Belancic and Lindgren 2020). Therefore, the Sámi constitute a relatively small group inhibiting a large area where they are a minority, and the number of speakers of Sámi languages is even smaller. To achieve the best within- and cross-country comparability, our focus in this study is on northern Norway and Sweden where the Sámi are more concentrated.

From a historical perspective, effective colonisation of the Sámi settlement areas that are today part of Norway and Sweden started in the sixteenth and seventeenth centuries (Kent 2018). Between the mid-nineteenth and mid-twentieth centuries, the oppression of the Sámi lifestyle reached a peak with both Norway and Sweden pursuing openly discriminatory assimilation policies (Henriksen 2008; Lantto 2010; Minde 2003; Trosterud 2008). In Norway, assimilation targeted all Sámi. In Sweden, reindeer herders were recognised as Sámi but segregated from the mainstream society, while other Sámi people were forced to assimilate. Assimilation policies included cultural aspects such as discouraging or banning Sámi languages, as well as economic aspects such as restricting reindeer herding and other forms of traditional Sámi land and natural resource use. As a result of colonisation and assimilation policies, the Sámi are a numerical minority in almost every locality in Sápmi today, with a few exceptions in Norway.

These policies were softened in the post-war era, and gave way to recognition and inclusion policies starting in the 1970s (Lantto and Mörkenstam 2008; Semb 2012). In addition to the political agency of the Sámi people to claim their rights, assimilationist policies have become unacceptable in the changing political climate, as Norway and Sweden assumed leading roles in democracy, human rights, and socio-economic equality. The extent to which the historical injustices against the Sámi are rectified today is, however, an open question. As these injustices have been multidimensional, we endeavour to examine different aspects regarding economic, political, social, cultural and linguistic inequalities.

Sámi politics and policies

Although both countries have made visible efforts to reverse their historical policies, it is generally accepted that Norway has made more progress (Allard 2015). We focus on country differences in four areas: recognition, self-governance, land rights, and language. First, the Sámi in Norway were constitutionally recognised as an Indigenous people in 1988. In contrast, while the Sámi in Sweden trace their first official recognition by the Swedish Parliament (Riksdag) to 1977, their constitutional recognition happened only in 2011. Norway ratified the ILO Convention 169 concerning Indigenous and Tribal Peoples in 1990, whereas Sweden did not; Norway ratified the European Convention on Regional and Minority Languages seven years before Sweden.

Second, arrangements for Sámi self-governance are more favourable in Norway than in Sweden. The non-territorial Sámi Parliaments (Sámediggi in Northern Sámi), established in 1987 (Norway) and 1992 (Sweden), have a dual function as government agencies and representative bodies, and they derive their legitimacy directly from the people and from the national law simultaneously (Lawrence and Mörkenstam 2016; Stępień, Petrétei, and Koivurova 2015). The balance between these roles varies across countries: the competencies, resources, flexibility, and the influence over other state authorities are larger for the Sámi Parliament in Norway as compared to Sweden (Falch, Selle, and Strømsnes 2016; Josefsen, Mörkenstam, and Saglie 2015). Moreover, the empirical literature on Sámi politics, such as voter registration and turnout for Sámi Parliament elections, political behaviour and cleavages, suggests that the Sámi Parliament in Norway is more strongly integrated into mainstream politics and receives more institutional trust (Dahlberg and Mörkenstam 2019; Saglie, Mörkenstam, and Bergh 2020).

Third, regarding policies on land entitlement, Norwegian legislation is more advanced in recognising Sámi territorial rights (Allard 2011). In Sweden, Indigenous land rights are almost invariably limited to reindeer herders (Strömgren 2017), and land right claims are mostly based on Supreme Court decisions rather than explicit political decisions (Torp 2013). In contrast, in Norway, the importance of land use for the entire Sámi community is emphasised in national legislation (e.g. Nature Diversity Act of 2009, Minerals Act of 2009). Furthermore, the Finnmark Act in 2005 transferred ownership of all public land in the former Finnmark County to the Finnmark Estate, a special body where representatives from the Sámi Parliament have considerable influence on land use.

Fourth, conscious efforts are being undertaken to revitalise the Sámi languages on national and individual levels in both countries, but with some differences (Aikio-Puoskari 2018; Lloyd-Smith et al. 2023). Norwegian policy has progressed more in this area as well (Anaya 2011). The Norwegian Sámi Act of 1987 recognises that “Sami and Norwegian are languages of equal worth” with further arrangements for a special administrative area, whereas the Sámi languages in Sweden are considered national minority languages alongside Finnish, Meänkieli, Yiddish and Romani. In Norway, all Sámi pupils have a right to Sámi education in principle (Vangsnes 2021), despite differences in implementation within and outside the special administrative area (Hermansen and Olsen 2020). In contrast, even the few special Sámi schools in Sweden follow bilingual curricula and do not offer Sámi-only instruction, while teaching in comprehensive schools is contingent on the availability of funding and teachers (Cabau 2014).

Reviewing these four policy areas suggests that, despite broad similarities between Norway and Sweden regarding their political and economic institutions, they differ in terms of their national policies concerning the Sámi. Norway appears in every aspect as the country where more favourable policies have been adopted. Such differences attest to the need to study ethnic inequalities from a cross-country comparative perspective for a better understanding of the effects of different policies.

Inequality and discrimination

A large portion of what we know on inequalities and discrimination that the Sámi are facing is based on public health studies in Norway and Sweden. In Norway, two rounds of the SAMINOR study (Broderstad, Hansen, and Melhus 2019; Brustad et al. 2014; Lund et al. 2007) provide representative data that allow us to compare the Sámi with the non-Sámi population. Public health studies in Sweden, in comparison, either rely on small sample sizes or do not systematically compare the Sámi to a non-Sámi control/reference population (e.g. Omma, Jacobsson, and Petersen 2012). Findings from these studies mostly report generally similar levels of health between Sámi and non-Sámi (see the review by Storm Mienna and Axelsson 2019). As for discrimination experience, the general finding is that Sámi report higher rates of discrimination than other population groups, which they clearly link to their ethnic identity (Hansen et al. 2016; Omma, Holmgren, and Jacobsson 2011). Furthermore, there is variance within the Sámi community: Individuals with a stronger affiliation to Sámi identity/culture regarding self-identification, language use, and occupation in reindeer husbandry tend to report poorer health, more discrimination, and lower status according to several socio-economic indicators (e.g. Bals et al. 2010; Hansen, Melhus, and Lund 2010; Nystad, Melhus, and Lund 2008).

Building on these previous works, we aim to provide further theoretical and empirical contributions on inequality and discrimination. First, the above review suggests that we may come to different conclusions depending on whether we focus on between-group inequalities or variation among individuals. Meanwhile, ethnicity-related factors play an important role in both dimensions. In this respect, in the next section, we discuss how we conceptualise ethnic inequalities with regard to groups and individuals, and how we develop our arguments and hypotheses based on this conceptualisation. Second, existing data from previous research have a specific focus on public health indicators and are not suitable for cross-country comparisons. For this reason, we collected original data tailored to address our questions, which we present in the subsequent section.

Theorising ethnic inequalities

As attested above, relative improvements for the Sámi have occurred since the mid-twentieth century. However, the legacy of past policies should not be disregarded (Carlsson 2020), and inequalities between ethnic groups are likely to persist. Despite the leading role that they assumed in democracy, human rights, and socio-economic equality, the social-democratic egalitarianism of Norway and Sweden has traditionally relied on the idea of a relatively homogenous society whereby diversity is primarily considered in terms of social class, i.e. economic inequality among individuals (Bay, Strömblad, and Bengtsson 2010). Yet, both countries arguably represent multi-ethnic societies characterised by two dimensions of inequality: vertical and horizontal. Social democratic egalitarianism primarily targets vertical inequalities, that is, inequalities among individuals, such as uneven distribution of wealth or income. In contrast, horizontal inequalities refer to the inequalities between groups demarcated by shared identities (Stewart 2008). Horizontal inequalities tend to be historically persistent and harder to tackle, due to their economic, political, social, or cultural characteristics which reinforce each other (Stewart 2009; Stewart and Langer 2008). In this respect, policies targeting only vertical inequalities in a group-blind manner, or policies targeting only certain types of inequality in a piecemeal manner may fail to alleviate historically persistent disparities (Stewart, Brown, and Langer 2008).

The literature on horizontal inequalities has a strong focus on violent ethnic conflict as an outcome (for a recent review, see Hillesund et al. 2018) or other forms of political mobilisation (e.g. Hillesund 2022; Stewart and McGauvran 2019). Whereas conflict studies are not directly applicable to the case of the Sámi and political mobilisation is beyond the scope of this paper, this literature still provides valuable insights. The macro-level link between horizontal inequalities and mobilisation is mostly theorised through intermediate micro-processes. For instance, Cederman, Weidmann, and Gleditsch (2011) rely on experimentally verified social-psychological research to suggest social comparison with other ethnic groups and subsequent formation of grievances as a causal mechanism. However, such micro-processes are usually assumed but not directly tested in this literature. Our aim is to take a closer look at such micro-processes, as we expect to find evidence for horizontal inequalities leading to frustrations in the daily experience of individual members of an Indigenous minority.

We conceptualise these frustrations as perceived discrimination, that is, the experience of feeling discriminated as reported by individuals. Perceived discrimination may result from exposure to actual discrimination or from attribution of the meaning of unfavourable experiences to discrimination (Diehl, Liebau, and Mühlau 2021; Schaeffer 2019). Therefore, subjective perceptions of discrimination may not necessarily reflect the objective existence of discrimination, but they must be taken equally seriously as both aspects may have unique consequences for individuals or ethnic groups and their behaviour, especially for members of disadvantaged groups (Paradies 2006; Schmitt and Branscombe 2002). The effects of perceived discrimination on individual wellbeing are substantiated by evidence from public health studies, which show its association with worse outcomes in physical and mental health (see, for systematic reviews, Pearce et al. 2019; Schmitt et al. 2014). Moreover, perceived discrimination, depending on its source, has been found to shape the political behaviour of ethnic minorities in different contexts (e.g. Bilodeau 2017; Oskooii 2020). In short, the subjective aspect of discrimination captures the micro-processes that can link horizontal inequalities with their wider political consequences.

We will examine subjective experience of discrimination as the main outcome variable, which provides a suitable conceptualisation for our purposes, as it embodies adverse experiences that can be linked to membership in an ethnic group. As for its covariates, we turn to types of inequalities. The above-cited literature mostly has a focus narrowed down to economic and political inequalities. However, horizontal inequalities should be conceptualised more broadly to include social and cultural status, such as minority language rights, which may be equally or even more crucial where differences between groups are primarily cultural and identity-based (Langer and Brown 2008; Ridgeway 2014). Identification with an ethnic group may be a stronger predictor of perceived discrimination than structural disadvantages (Lindemann and Stolz 2021).

Against this background, we expect to find a strong effect of socio-cultural inequalities on perceived discrimination, independently of the presence of economic and political inequalities at individual or group level. Overall, our theoretical expectations can be summarised in four hypotheses that will guide our empirical analyses. First, if we take the ethnic background of an individual as a basic indicator to capture inequalities in social status broadly, we can hypothesise that:

H1: Those who have a Sámi ethnic background are more likely to experience discrimination.

The confirmation of H1 will serve as a validation of our basic reasoning that Indigenous ethnic background is fundamentally linked with adverse experiences in daily life. Thus, it will also serve as a prerequisite for the following hypotheses, which will build upon the effect of ethnic background.

For the second hypothesis, we consider economic status, which essentially pertains to inequality among individuals or households. Vertical economic inequality may be related to the prevalence of discrimination; thus, those who have a lower economic status would be more likely to experience discrimination. Moreover, vertical economic inequality may also materialise as between-group inequality; thus, those who have a Sámi background might have a lower economic status on average. To combine these intuitions, we will study the economic dimension of the ethnicity-discrimination nexus by testing the following hypothesis:

H2: The effect of economic status on experiencing discrimination is stronger among those who have a Sámi ethnic background.

Third, we focus on language as a primary identifier of cultural belonging and a major source of cultural inequalities. Minority language use can lead to perceived discrimination through several mechanisms. Regarding actual discrimination, limited opportunities in one’s language, such as lack of schools, books, and other media, can constitute a form of discrimination in itself. Additionally, use of a minority language can make a person easier to identify as a member of an ethnic group by others. Regarding a person’s perceptiveness of discrimination, limited visibility of a minority language may signal low social prestige for its speakers. Furthermore, proficiency in and frequency of using a minority language may be an indicator of the strength of ethnic identification. Therefore, different language-related factors, such as proficiency in the minority language, its use with family members and beyond family context, may simultaneously be associated with discrimination experience. Remaining attentive to these differences, we hypothesise that:

H3: Those who are more proficient in a Sámi language, those who use a Sámi language within their family, and those who use a Sámi language beyond the family context more often are more likely to experience discrimination.

Finally, we consider political inequality, which primarily varies between countries. Based on our review of country differences in the previous section, we assume that central state institutions are more responsive to the Sámi in Norway in general, which would imply a higher likelihood of experiencing discrimination in Sweden. Furthermore, Sweden had adopted, in the past, an occupation-based conception of Sámi ethnicity, which could have economic implications still today. Meanwhile, Norway is more advanced in legislation on land rights, which could be relevant for economic inequalities, and on language and education policies, which could make a difference for cultural inequalities. Therefore, we hypothesise that, among the Sámi:

H4: The effects of economic status and minority language use on discrimination experience are stronger in Sweden than Norway.

However, differences between countries are not limited to past and present policies concerning the Sámi. Most importantly, there are demographic differences that might affect discrimination experience: not only is the total number of Sámi living in Norway higher, but also their share in the population shows larger variation across localities in Norway, and in a few places, they are numerically the majority ethnic group. Therefore, one should pay attention to these comparability issues in interpreting H4.

To summarise, the theoretical background of our argument is based on the distinction between vertical inequalities among individuals and horizontal inequalities between groups. Drawing on the literature studying this distinction, we identify social frustrations as a potential outcome of horizontal inequalities worthy of attention, which we conceptualise in terms of discrimination experience as perceived by individuals. For its covariates, we specify social, economic, cultural, and political inequalities as distinct sources of discrimination. We argue that the inequalities that the Sámi face collectively are categorically different from individual-level inequalities, and these should display strong cultural and linguistic characteristics in addition to economic ones, leading to a higher likelihood of experiencing discrimination that would be aggravated by less complaisant political contexts.

Empirical strategy

Data collection

Our objective to study multiple aspects of ethnic inequalities from a within- and cross-country comparative perspective required the collection of new data, which we undertook through a full adult population survey (Holzinger et al. 2023). To address the challenge of a rather small Sámi population being spread out across the two countries, we followed a similar strategy as the SAMINOR studies, sampling from the entire population in locations where the share of the Sámi is highest, applying the logic to both Norway and Sweden. We used the electoral rolls for Sámi Parliament elections in 2017 to identify 15 municipalities from Norway and 5 municipalities from Sweden.³ Thus, potential respondents of the survey consisted of more than 30,000 adults in each country, of which approximately 20% in Norway and 9% in Sweden are estimated to be registered for the Sami Parliament electoral roll. Electoral roll registration rates also show larger variation in Norway, reaching levels above 50% in some localities. Further information on these municipalities is available in the online appendix (Table A1).

To maximise the response for key questions, we adopted a multi-stage strategy. In the first stage, the adult population of the selected municipalities was called for a brief telephone interview on core questions, and invited for the second stage, which comprised a more extensive questionnaire that could be filled out online or on paper. This was followed by vocabulary and audio recording tasks, for more objective and detailed linguistic data. Measures to reduce non-response included at least two reminders and material incentives. As a result of the multi-stage strategy, the sample size varies depending on the survey stage. After accounting for dropouts, 5416 respondents participated in the telephone interviews, and 1420 in the questionnaire stage. Both response rates and sample sizes are higher in Sweden, but the share of respondents with a Sámi background is higher in Norway, in line with population estimates; thus, in both countries, there are sufficiently large samples to make inferences about the Sámi and to compare them with the rest of the population. Further details about the sample characteristics are listed in Table 1.

Table 1. Sample sizes in different survey stages.

	Norway			Sweden
	Sámi BG	Other	Total	Sámi BG	Other	Total
Contacted people	–	–	11,153	–	–	6265
Telephone interview	1072 44.7%	1324 55.3%	2396 (21.5%)	847 28.0%	2173 72%	3020 (48.2%)
Questionnaire	254 50.6%	248 49.4%	502 (21.0%)	284 30.9%	634 69.1%	918 (30.4%)
Vocabulary task in majority language	86 40.1%	124 59.0%	210 (41.8%)	149 27.8%	387 72.2%	536 (58.4%)
Vocabulary task in North Sámi	65 55.6%	52 44.4%	117 (23.3%)	82 47.7%	90 52.3%	172 (18.7%)

Note: Percentages in parentheses indicate response rate with respect to the previous stage. “BG” stands for ethnic background.

The first stage questionnaire consisted of a relatively small number of questions on ethnic identification, views on the rights of the Sámi, discrimination experience, economic situation, and basic demographics. An additional set of questions on language use were asked to respondents with a Sámi ethnic background.⁴ The second stage questionnaire consisted of a much higher number of questions on ethnic identification, policy evaluation, political behaviour and views, perceptions of social status, discrimination experience, further demographics, and language use plus self-assessed language proficiency. Respondents who indicated a Sámi background were provided with an extended version of the questionnaire, which included questions specific to Sámi communities and languages. Data from the first stage enable us to analyse certain key questions with a large sample size, and data from the second stage enable us to conduct detailed analyses with a smaller sample size. An important innovation of our study is the integration of detailed linguistic variables and objective measures of proficiency and fluency, through vocabulary and speech recording tasks in the majority language and North Sámi. While these provide valuable information for linguistic studies in their own right, they also validate the accuracy of self-assessed measures in the second-stage questionnaire.⁵ Data from each respondent can be anonymously matched across survey stages and language tasks.

All survey instruments were prepared in English, reviewed internally and externally by experts and local stakeholders, and translated into Norwegian, Swedish, and North Sámi by native speakers. The vast majority of participants completed the web-based questionnaire in the majority language of their respective countries. Fieldwork took place between April and September 2021.

Variables and methods

To test the hypotheses formulated above, the main outcome variable to analyse will be discrimination experience, which has been coded dichotomously based on questions asked in the telephone interview and in the second stage questionnaire, as to whether the respondent experienced discrimination. The most fundamental independent variable is ethnic background, which is required not only for testing H1, but also interacting with other variables in further hypotheses. This is based on ethnic identification of oneself, parents, and/or grandparents⁶, and coded into four categories: Sámi, immigrant, national minority (Finns, Kvens, and Tornedalers), and majority (without any minority identification). Next, we focus on income as the indicator of economic status, and calculate household income adjusted for household size through the OECD equivalence scale. Thus, we test H2 by examining the interaction between household income and ethnic background.

For linguistic variables, we use several measurements. First, respondents with a Sámi background were asked in the telephone interview whether a Sámi language is or had previously been spoken at home, and this is coded dichotomously as three variables for oneself, parents, and grandparents. Second, questionnaire respondents were asked to self-assess their proficiency in a Sámi language and in the majority language, and weighted average scores from four dimensions (understanding, speaking, reading, writing) are used for measurement. Third, home use of a Sámi language is measured for respondents with Sámi background as weighted average scores from questions on different family members. Fourth, general use of Sámi is measured similarly as weighted average score from six dimensions based on several types of listening, reading, and writing. Thus, we test H3 by examining and comparing different metrics related to a Sámi language, namely self-assessed proficiency, home use, and general use. For H4, the country variable is coded according to the survey location, and we examine its interaction with household income and general use of a Sámi language.

Finally, we use several control variables. Demographic controls include age, gender, and education level. Furthermore, we use three variables based on respondents’ subjective assessments of their own position in society, of the fairness of wealth distribution, and of their satisfaction with democracy in their countries. These variables are intended to account for the perception-based aspect of discrimination, regarding social, economic, and political inequalities respectively. We provide more detailed information on the operationalisation and measurement of all variables in the online appendix (Section A2). Tables 2 and 3 summarise descriptive statistics for each variable, and show the corresponding survey stage where the responses were collected.

Table 2. Descriptive statistics: categorical variables.

	Norway		Sweden		Source
	N	Ratio	N	Ratio
Discrimination experience	268	11.2%	534	17.7%	TI
Discrimination experience	90	17.9%	243	26.5%	Q
Sámi background	1072	44.7%	847	28.0%	TI
National minority background	242	10.1%	804	26.6%	TI
Immigrant background	192	8.0%	211	7.0%	TI
Majority background	889	37.1%	1142	37.8%	TI
Sami spoken at home – self	233	21.7%	173	20.4%	TI-SO
Sami spoken at home – parent	473	44.1%	260	30.7%	TI-SO
Sami spoken at home – grandparent	787	73.4%	458	54.1%	TI-SO
Gender: male	1119	46.7%	1361	45.1%	TI
Gender: female	1277	53.3%	1636	54.2%	TI
Gender: diverse	0	0.0%	6	0.2%	TI
Education: pre-tertiary	129	25.7%	401	43.7%	Q
Education: advanced vocational	60	12.0%	171	18.6%	Q
Education: University	254	50.6%	257	28.0%	Q

Note: TI: telephone interview, Q: Second-stage questionnaire, SO: Sámi only questions.

Table 3. Descriptive statistics: numeric variables.

	Norway
	Min	Median	Mean	Max	St. Dev.	Source
Adjusted household income (NOK)	84,600	503,704	567,654	1,791,000	258,067	TI
Age	18	55	54.4	97	16.2	TI
Proficiency in a Sámi language	0.00	0.21	0.66	4.00	1.02	Q
Proficiency in majority home	1.75	4.00	3.66	4.00	0.51	Q
Use of a Sámi language at home	1.00	1.00	1.54	5.00	1.16	Q-SO
General use of a Sámi language	0.00	0.45	0.91	3.85	1.05	Q-SO
General use of majority language	0.00	3.74	3.62	4.00	0.47	Q-SO
Self-placement in social ladder	1	6	6.07	10	1.58	Q
Satisfaction with democracy	1	7	6.45	10	2.10	Q
Fairness of wealth distribution	1	6	5.74	10	2.05	Q
	Sweden
	Min	Median	Mean	Max	St. Dev.	Source
Adjusted household income (SEK)	14,471	314,500	335,918	1138,500	209,191	TI
Age	18	55	53.5	101	17.3	TI
Proficiency in a Sámi language	0.00	0.00	0.35	4.00	0.73	Q
Proficiency in majority language	0.25	4.00	3.66	4.00	0.53	Q
Use of a Sámi language at home	1.00	1.00	1.34	5.00	0.77	Q-SO
General use of a Sámi language	0.00	0.15	0.52	3.69	0.80	Q-SO
General use of majority language	0.84	3.61	3.53	4.00	0.49	Q-SO
Self-placement in social ladder	1	6	5.87	10	1.52	Q
Satisfaction with democracy	1	6	5.69	10	2.34	Q
Fairness of wealth distribution	1	4	3.92	10	1.95	Q

Note: TI: telephone interview, Q: Second-stage questionnaire, SO: Sámi only questions. Language proficiency indicators are self-assessed.

An overview of descriptive statistics shows that there is meaningful variation within each variable and between countries to use the data with a view to testing our hypotheses. In the following section, we begin with visual analyses to explore the viability of our hypotheses in terms of the distribution of key variables between Sámi and non-Sámi ethnic backgrounds and between Norway and Sweden. Next, for rigorous hypothesis tests, we move on to multivariate modelling. Since the main outcome variable, discrimination experience, is dichotomously measured, we use logit regression and interpret odds ratios. The overall modelling strategy is shaped, and sample sizes for successive models are determined by the availability of variables in different survey stages, and whether corresponding questions were only asked to respondents with Sámi background. The subset of the data and a script file that can replicate our findings are provided as online attachments.

Results

Exploratory analyses

To begin with visual analyses of the distribution of key variables, the charts below show comparisons of people with and without Sámi background in both countries in terms of their discrimination experience, income, and language use. As seen in Figure 1, more people with Sámi background report discrimination experience than others, supporting our basic hypothesis, but this trend is similar between the two countries (the ratio being 1.8 in Norway and 1.7 in Sweden), in contrast to one of our intuitions behind country differences. In other words, either having a Sámi background or living in Sweden⁷ leads to a higher probability of reporting discrimination experience, but the country of residence does not significantly alter the effect of ethnic background on discrimination experience. However, this does not necessarily preclude the possibility of a significant interaction between country and other posited variables, namely household income and general use of a Sámi language, which will be considered in the next subsection.

Figure 1. Distribution of discrimination experience between ethnic groups in Norway and Sweden.

Figure 2 considers economic status in terms of income relative to country distributions: the values are standardised and centred on country means.⁸ The distributions of people with Sámi background and the rest in each country are strikingly similar, and any difference is statistically insignificant. This could suggest that the inequalities that the Sámi are facing are not fundamentally economic in nature, and other inequalities are not strongly reflected in their economic status. In other words, vertical economic inequality does not materialise as a form of between-group inequality. Yet, we will still consider in the next subsection if income has an effect on discrimination experience, and if such an effect varies between people with and without Sámi background and across countries.

Figure 2. Distribution of household income between ethnic groups in Norway and Sweden.

Finally, Figure 3 focuses on general language use scores among respondents with Sámi background. The use of the majority languages is quite high in both countries, only slightly and insignificantly lower in Sweden. The difference between countries in terms of the use of a Sámi language is much more significant, whereby people in Norway display higher scores. This can be taken to reflect the differences between countries regarding language and education policies. It remains to be seen if there is an association between language use and discrimination, and if a similar cross-country variation can be observed in this association as well.

Figure 3. Distribution of the general use of a Sámi language and the majority language among the Sámi in Norway and Sweden.

Multivariate models

In the following, we model the potential effects of the independent variables across two sets of multivariate models. Table 4 summarises the results from the first set of models, which are based on the larger sample of all respondents, i.e. those with and without a Sámi background. Table 5 summarises the results from the second set of models that focus on people with a Sámi background only. In each table, the first three models are from the larger telephone interview data, and the next two models are from the second stage questionnaire data with additional questions. We discuss below the implications of these findings for each hypothesis.

Table 4. Logit regression models comparing people with and without Sámi background. Outcome variable: discrimination experienced.

	Odds ratios
	Model 1.1	Model 1.2	Model 1.3	Model 1.4	Model 1.5
Ethnic background (ref: majority)
Sámi	2.31***	2.30***	2.68***	1.62***	1.57**
Immigrant	1.92***	1.92***	3.05***	0.87	0.92
National minority	1.39**	1.39**	1.92**	1.20	1.28
Country (ref: Norway)
Sweden	1.82***	1.83***	2.30***	1.93***	1.53**
Adjusted household income	0.85***	0.97	0.85***	0.82***	0.89
Ethnic BG × Income (ref: majority)
Sámi: Income		0.82*
Country × Ethnic BG
Sweden: Sámi			0.81
Proficiency in Sámi				1.39***	1.35***
Self-placement on social ladder					0.86***
Satisfaction with democracy					0.87***
Perceived fairness of wealth distribution					0.92**
Demographics included:	Age, age-squared, gender	Age, age-squared, gender	Age, age-squared, gender	Age, age-squared, gender, education	Age, age-squared, gender, education
N	4420	4420	4420	1248	1248
Null deviance	3709 (4419 df)	3709 (4419 df)	3709 (4419 df)	1363 (1247 df)	1363 (1247 df)
Residual deviance	3393 (4410 df)	3390 (4407 df)	3388 (4407 df)	1261 (1235 df)	1208 (1232 df)
AIC	3413	3416	3414	1287	1240
Log-likelihood	−1697	−1695	−1694	−630	−604

Significance levels: ***p < 0.01 < **p < 0.05 < *p < 0.1.

Table 5. Logit regression models focusing on people with Sámi background. Outcome variable: discrimination experienced.

	Odds ratios
	Model 2.1	Model 2.2	Model 2.3	Model 2.4	Model 2.5
Ethnic background (ref: Sámi only)
Majority and Sámi	0.55***	0.54***	0.55***	0.77	0.81
Country (ref: Norway)
Sweden	1.89***	1.87***	1.65***	1.73**	1.05
Adjusted household income	0.82***	0.91	0.82***	0.77**	0.79*
Sámi spoken at home:
Oneself	2.97***	2.99***	2.43***
parents	1.12	1.11	1.10
Country × Income (ref: Norway)
Sweden: Income		0.82
Country × Sámi at home (ref: Norway)
Sweden: oneself			1.56
Home use of Sámi				1.03
General use of Sámi				1.65***	1.31*
Country × Sámi gen. use (ref: Norway)
Sweden: General use of Sámi					2.04***
Demographics included:	Age, age-squared, gender	Age, age-squared, gender	Age, age-squared, gender	Age, age-squared, gender, education	Age, age-squared, gender, education
N	1570	1570	1570	439	446
Null deviance	1526 (1569 df)	1526 (1569 df)	1526 (1569 df)	530 (438 df)	542 (445 df)
Residual deviance	1323 (1560 df)	1321 (1559 df)	1321 (1559 df)	463 (427 df)	466 (434 df)
AIC	1343	1343	1343	487	490
Log-likelihood	−661	−660	−660	−231	−233

Significance levels: ***p < 0.01 < **p < 0.05 < *p < 0.1.

To begin with H1, we check whether Sámi ethnic background increases the probability of experiencing discrimination. In Table 4, we consistently observe odds ratios that are statistically significant across models. Results from the first three models suggest that a Sámi ethnic background is associated with a more than two-fold increase in the probability of experiencing discrimination, compared to majority ethnic background. In the fourth and fifth models, where proficiency in a Sámi language is considered, the independent effect of Sámi ethnic background approximates a 50% increase. In Table 5, we can test this relationship by comparing those who only have a Sámi ethnic background with those who have both Sámi and majority backgrounds. Odds ratios from the first three models show that, among the Sámi, the probability of experiencing discrimination is almost halved for those who additionally have a majority ethnic background. Results from the fourth and fifth models, where Sámi language use is taken into account, are not statistically significant. Overall, these findings confirm H1.

Regarding H2, which expects that Sámi ethnic background aggravates the effect of economic status on discrimination, we initially consider whether a relationship exists between income and discrimination experience. Model 1.1 shows that a standard deviation increase in household income is linked with a 15% decrease in the probability of experiencing discrimination. To test H2, we introduce an interaction term between Sámi background and household income in Model 1.2. Its results suggest that the effect of income on discrimination among the Sámi is slightly stronger than the average, and almost non-existent for the majority. However, the statistical significance of the interaction term is quite low (the p-value is larger than the 5% threshold). While the empirical evidence suggests a variation between the Sámi and the majority, it is not strong enough to decisively confirm H2.

Next, H3 is concerned with the relationship between language and discrimination. To examine multiple aspects of language knowledge and use, we consider different variables. First, Model 1.4 shows that one standard deviation increase in the self-assessed proficiency in a Sámi language⁹ is associated with an approximately 40% increase in the probability of experiencing discrimination. Second, among the Sámi respondents, Model 2.1 shows that the probability of experiencing discrimination increases almost three-fold as a function of whether one speaks a Sámi language at home, controlling for their parents’ language use. Third, as shown by Model 2.4, the frequency of using a Sámi language in situations outside the family context (termed “general use”) is associated with a 65% increase in the probability of experiencing discrimination, controlling for its use with family members (termed “home use”). Therefore, empirical findings confirm H3 with respect to all three aspects.

Finally, H4 posits that the effects of income and language use should vary across the two countries. We test this by introducing interaction terms between country and the relevant variables. First, we consider whether this pattern can be observed more generally for ethnic background, but Model 1.3 does not return a statistically significant interaction term. Second, Model 2.2 tests the cross-country variation of the effect of income among the Sámi, but does not return a statistically significant interaction term either. Third, we examine two aspects of language use. We cannot find a statistically significant difference between the two countries regarding the use of a Sámi language at home (Model 2.3), but we find a positive result for the general use beyond the family context (Model 2.5). According to the latter, the general use of a Sámi language is associated with a two-fold increase in the probability of experiencing discrimination in Sweden, while we observe only a 30% increase for Norway with low statistical significance. Therefore, H4 can only be confirmed with respect to the general language use.

To account for potential confounding effects of demographic differences between countries, we ran additional analyses: we grouped Norwegian and Swedish municipalities into comparable categories¹⁰, and allowed the coefficient of general language use to vary across these categories instead of countries. As seen in Table 6, in the municipalities of Norway where the share of the Sámi is relatively high, the effect of general language use is indeed almost non-existent, suggesting that our finding is valid only when the Sámi are a clear minority. However, when we compare similar municipalities between the two countries, we still see that the effect of general language use is stronger in Sweden, supporting H4.

Table 6. Odds ratio estimates for general language use, from a mixed-effect random-slope model based on Model 2.4, varying across municipalities.

Norway			Sweden
Sámi share	N	Odds ratio	Sámi share	N	Odds ratio
8.50% (Alta)	130	1.50	7.50% (Kiruna)	142	2.04
9.40%–17.50%	41	1.70	8.30%–15.50%	108	2.16
20.70%–68.60%	83	1.08
			Unknown	34	1.71

Note: Sámi share is based on Sámi Parliament electoral roll registration rates, probably lower than true percentages (see also note 3). The “unknown” category in Sweden is due to postcodes that correspond to localities outside the sampling frame, invalid postcodes, and missing data points. In Norway, all data are accurate.

In addition to hypothesis tests, in Model 1.5, we introduce three variables measuring the respondents’ subjective assessments about society, economy, and politics to account for the subjective aspect of the outcome variable, which may result in an endogenous relationship with predictors. We can see that the statistical significance of income disappears with this model specification, while the odds ratios for ethnic background and minority language proficiency are minimally affected. In this respect, we can be more confident about the validity of our findings, especially concerning ethnic background and language. With these results, we confirm that the probability of experiencing discrimination is higher for those who have a Sámi ethnic background, those who are more proficient in a Sámi language, and those who use a Sámi language within or beyond the family context, while the last relationship is stronger in Sweden than Norway.

Discussion

Our first and most fundamental finding is that the Sámi are indeed more likely to experience discrimination than the majority population, which is confirmed by both exploratory analyses and multivariate models. While it is hardly surprising that ethnic minorities are more likely to experience discrimination, which can also be seen here for immigrant and national minority groups, the effect of Sámi background seems clearly stronger. Therefore, it is worth making the distinction between different types of ethnic minorities and paying particular attention to Indigenous peoples.

We expected that income inequality could act as a mediating factor between ethnic background and discrimination experience. However, in our exploratory analyses, we failed to find a significant difference in the income distribution between the Sámi and the rest of the society, at least in the areas in which the survey was conducted.¹¹ This may be associated with Nordic countries being relatively successful in tackling economic inequalities. Meanwhile, income is still a strong predictor of discrimination experience at the individual level. When we considered the interaction between the two, we could only find weak empirical evidence that Sámi ethnic background aggravates the effect of income. Although this finding should not be taken as definitive, it reaffirms the existing literature in that there is not a big material gap between the Sámi and the rest of the society in the same localities, but that the variation within community is significant.

As the inequalities that the Sámi face are not fundamentally economic, we turn to cultural aspects. To this end, we examined different types of minority language use. Self-assessed proficiency in a Sámi language, use of a Sámi language at home, and use of a Sámi language outside the family context are all associated with a higher probability of experiencing discrimination, supporting the view that the cultural dimension is particularly relevant for horizontal inequalities. However, when home use and general use are tested in the same model, we saw that the latter has the dominant effect. This seems to suggest that the effect of language use is not simply a reflection of individual factors such as the strength of ethnic identification, but is also linked with societal factors such as limited opportunities to participate in social or educational activities using a minority language or low prestige ascribed to it. We suspect that speakers of Sámi languages who use their languages in the wider community context are more acutely aware of these problems. Although both Norway and Sweden have come a long way in improving the situation of the Sámi languages, work is apparently still needed in increasing opportunities for language use, and in raising the prestige and visibility of the Sámi languages in the wider community.

Finally, we consider the differences between Norway and Sweden. In contrast to our expectations based on Norway’s more complaisant policy environment, the effects of ethnic background, income, and home use of a Sámi language do not vary across countries. However, those who frequently use a Sámi language beyond the family context are more likely to experience discrimination in Sweden than Norway, and this is not merely due to demographic differences. In this respect, language and education policies come forth as an area where cross-country differences are more consequential and the need to address ethnic inequalities is more pressing.

Conclusion

We have examined different types of ethnic inequalities with a focus on the Sámi living in the northern regions of Norway and Sweden. Drawing on a theoretical background differentiating inequalities among individuals and between groups (vertical and horizontal respectively), we have argued that the members of the Sámi communities face mainly horizontal inequalities that display strong cultural and linguistic characteristics. Conceptualising the main outcome variable, discrimination experience, as an embodiment of social grievances, we found that both vertical and horizontal inequalities, as well as a combination of social, economic, cultural, and political inequalities matter. We can see persistent effects of horizontal inequalities, even if these are not manifested as economic differences between ethnic groups. In this sense, horizontal inequalities are not a mere reflection of vertical inequalities clustered around group identities. Moreover, the use of a Sámi language, especially in the public sphere, comes to the fore as a strong predictor of discrimination experience, and the only aspect that reflects the less favourable policies adopted in Sweden. We can thereby confirm that between-group inequalities display strong cultural and linguistic characteristics, and they can be more effectively alleviated when political priorities are not limited to economic and individual-level inequalities.

This paper represents a contribution to Indigenous studies examining ethnic inequalities in a European context marked by social-democratic egalitarianism, demonstrating the value of a multidisciplinary approach, especially with regard to the use of linguistic variables. We have shown that persistent effects of historical and ongoing inequalities cannot be addressed by targeting economic status and disparities among individuals alone. As a theoretical contribution, the application of the concept of vertical and horizontal inequalities broadens the scope of this line of research, by focusing on discrimination experience as an immediate effect of ethnic inequalities, which existing literature assumed as a linkage with further economic and political outcomes. As the main empirical contribution, we provide original representative data, with which both cross-country and within-country comparisons can be conducted between the Sámi and other members of the society living in the same regions. This study used a subset of the collected data; the full dataset will enable researchers to offer further contributions.

Future research can shed light on the mechanisms through which different types of inequality are linked with discrimination, as this study remained focused on observing the existence of such links. Moreover, our empirical findings on the effect of economic inequality are ambivalent; for a more detailed understanding, future research can concentrate on other economic aspects, such as the perceptions of economic status among members of Indigenous minorities. While we found a strong association between language use and discrimination experience, we have not verified the underlying intuitions empirically; future research could pose additional questions, such as: How do the Sámi assess the lack of opportunities in their languages? What are the implications of using a minority language in public? How are Sámi languages esteemed in the majority population? What is the relationship between language use and ethnic identification? Finally, we examined political inequality from a macro perspective and could only partially observe cross-country differences; to complement and deepen this perspective, future research could examine how the Sámi appraise their political status and their inclusion in national political processes, whether there are differences between ethnic groups and countries in political participation, attitudes towards government performance, and policy preferences.

Ethics statement

Norwegian National Research Ethics Committees required internal ethics review by the university of the authors, which was fulfilled through the approval by Institutional Review Board, Ethics Commission, University of Konstanz (IRB statement 21/2020). The study was also approved by the Swedish Ethical Review Authority with the decision Dnr 2020-06853. Special attention has been paid to the principles of voluntary participation, informed consent, non-intrusive invitations, clear communication, open contact channels, and the right to withdraw. Interviews and questionnaires began with consent confirmation accompanied by information on the survey. For data protection, confidentiality and privacy, personally identifiable information has been immediately separated from survey data, and the current dataset is fully anonymised.

Acknowledgements

The authors are grateful to Johanna Al-Gbori, Vincent Björkquist, Anna Bühler, and Stefan Götz for their assistance in the preparation of the survey and data processing. They are also grateful to Jo Saglie (Institute for Social Research, Norway) and Øystein Vangsnes (UiT The Arctic University of Norway) for their support and feedback.

Disclosure statement

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

Additional information

Funding

This paper reports results from a research project that was funded by the Deutsche Forschungsgemeinschaft (DFG – German Research Foundation) under Germany’s Excellence Strategy – EXC-2035/1-390681379.

Notes

1 According to ILO Convention 169 (1989), Article 1, the definition of Indigenous peoples is based on self-identification as Indigenous, inhabitancy at the time of colonisation, and retention of own institutions.

2 Numbers are compiled from various sources: The World Factbook (CIA, Washington, DC, 2021), Ethnologue (Eberhard, Simons, and Fennig 2023), Sámi Information Centre in Sweden, Young and Bjerregaard (2019). These numbers mostly refer to the Sámi living in the Sápmi area.

3 Electoral roll figures are not ideal predictors of the distribution of Sámi, since a considerable number are probably not registered (Bergh et al. 2018). However, in the absence of more reliable information, it served as the best indicator to identify the municipalities where we would be most successful in achieving a higher share of Sámi respondents.

4 For the purposes of this study, a Sámi background is defined in terms of ancestral ties with the Sámi community, or self-identification as Sámi. The definition of Sámi ethnicity based on these two dimensions has been shown to be stable over time (Pettersen and Brustad 2015).

5 Results from the language tasks are not presented in this paper. However, self-reported linguistic variables, which are analysed here, have been validated using the objective data from language tasks.

6 In line with note 4 above, this operationalisation was inspired by the registration criteria of Sámi Parliament electoral rolls.

7 People living in Sweden may actually face more discriminatory practices, or their perception of what constitutes discrimination may be broader. It is not possible to distinguish between the two based on the data, and our approach remains agnostic in this respect. But the finding suggests that the country difference must be considered in further analyses.

8 Meaningful comparisons between countries could not be directly made due to local differences in purchasing power and living standards.

9 All respondents were asked to self-assess their proficiency, but only a few non-Sámi respondents reported any proficiency in a Sámi language.

10 Since Alta (Norway) and Kiruna (Sweden) are the largest municipalities in the sampling frame with the lowest Sámi share in each country, and constitute approximately half of the sample, we consider them comparable with each other. For the rest, while 15.50% is the maximum in Sweden, we created a comparable group in Norway by setting the cut-off point at below 20%.

11 Due to the sampling strategy, this comparison is valid for the localities where the Sámi are concentrated. Any generalisation beyond this area, e.g. to the national level, should be made with caution.