Mixed Methods Research in Global Development Studies: State-Sponsored Resettlement Schemes in Ethiopia

This article contributes to mixed-methods research in global development studies. It draws on empirical study of changes in access to livelihood resources ensuing from state-sponsored resettlement schemes in Oromia, Ethiopia. The analysis demonstrates that explaining and understanding processes and outcomes of development interventions can be expanded and deepened by the integration of quantitative and qualitative methods. Quantitative analysis uncovers the magnitude of changes, whereas qualitative analysis reveals political power relations and contextual meanings of the changes as experienced by affected people. The study shows that mixed-methods approach enables addressing the statistical and substantive significance of the variables in question and the diversity of meanings and experiences associated with development interventions. This strengthens the analytical power of research, providing opportunities to enhance evidence-based comprehension of generic patterns and specific conditions and outcomes of development interventions in diversified situations of the global South.


Introduction
In a world bound with crises and hazards, the search for innovative approaches to understand multidimensional changes and elaborate pathways for more just and sustainable development in the global South are highly needed.Recent discussions in the field of global development studies relate to a search for analytical strategies for enhanced understanding of the interlinkages of generic patterns and contextuallygrounded conditions and outcomes of development interventions in diverse circumstances of the global South (Borras et al. 2015;Brockington 2019;Li and Semedi 2021;Lund 2014).This includes initiatives for the use of mixed methods (MM) in a way that compellingly challenges the conventional polarization between quantitative and qualitative research (Ide 2017;McKim 2017), and to reflect on the opportunities and challenges involved in MM approaches to provide evidence-based research results (Baird et al. 2021;Harrison et al. 2020;Ramlo 2016;Wayessa 2020;Zhou and Wu 2022).
Through the analysis of changes in access to livelihood resources associated with state-sponsored resettlement schemes in western Oromia, Ethiopia, since the early 2000s, this article foregrounds the significance of MM approach in global development studies, making a case for careful integration of qualitative and quantitative methods.While researchers in health sciences, education, management, and business studies have increasingly applied MM approach, its use in global development studies is still relatively limited.Although more and more development scholars are using variations of MM approach (Baird et al. 2021;Brockington 2019;Harris 2022;LaRue et al. 2021;Prowse 2010;Randell 2016;Schneider et al. 2020;Shaffer 2013), development research is yet to benefit from its wider application to understand multi-dimensional processes and outcomes of development interventions in the global South.With few exceptions (Guha 2013;Wayessa and Nygren 2016), this is true especially in displacement-resettlement studies, a field dominated by qualitative approaches.
By combining qualitative and quantitative approaches to explain and understand the processes and livelihood outcomes of resettlement schemes in Ethiopia, this study provides conceptual-methodological contributions to displacement and resettlement studies and encourages innovative applications of MM in global development studies, in general.First, the study shows the opportunities and challenges involved in MM for comprehending development interventions in the global South.Second, it analyzes the interventions as arenas of struggle, negotiation, and contestation between different groups of actors, with differentiated opportunities to influence and make decisions related to interventions (Kabra and Mahalwal 2019;Nygren 2021;Ribot 2022).Third, the study shows the state's involvement in resettlement projects as a form of territorialization and strengthening of governmental authority, especially in the contexts of authoritarianism and paternalist politics.
Careful combination of quantitative and qualitative approaches is highly important in circumstances characteristic of the global South, where statistical data on land tenure, access to resources, and livelihood changes are frequently fragmented and out-of-date (Brockington 2019), and where the only opportunity to get contextually-sensitive data on political aspects and social experiences of interventions is often through qualitative fieldwork (Li and Semedi 2021, p. 3).The integration of quantitative and qualitative methods increases the opportunity for evidence-based comprehension of generic patterns and specific conditions and outcomes of development interventions.This is essential under conditions of legal pluralism, common in many parts of the global South, where formal land-tenure regimes and informal resource rights are intrinsically interwoven, and official, unofficial, and shadow forms of governance are tightly entangled (Cleaver and de Koning 2015;Coates and Nygren 2020;Lund 2016;Smith 2014;Sud 2017).
This article is organized in the following way.In section two, we present recent conceptualizations related to the MM approach, focusing on the forms of qualitative-quantitative integration relevant for global development studies.Section three explains the socio-political context of resettlement projects in Ethiopia and the methods used in the study.In section four, we examine the resettlement schemes through the notions of resettler voluntarism and host consultation, while in section five, we analyze the outcomes of the resettlement schemes through changes in access to livelihood assets for both the resettlers and hosts.Section six concludes by showing the relevance of MM in explaining and understanding processes and outcomes of development interventions in diversified situations of the global South.

A Multi-phase Qualitative-Quantitative Integration
Mixed-methods research involves the combination of qualitative and quantitative approaches in different phases of the research process.The combination becomes rigorous when it goes beyond mere inclusion of two types of data toward an interactive approach, where one sort of method helps discover new aspects in the data gathered through another type of method (Creamer 2018;Clark 2019).This requires research where qualitative and quantitative procedures are tightly interlinked in order to enable a more holistic understanding of the phenomenon than what is achieved by either alone (Fetters and Molina-Azorin 2017, p. 293;Johnson et al. 2019, p. 302).At best, the integration of quantitative and qualitative approaches starts with the formulation of research questions and/or hypotheses, continues through data collection and data analysis in a way that cross-fertilizers the examination of commonalities and differences between qualitative and quantitative data, and culminates in presenting the results in an interconnected way.
Careful integration of qualitative and quantitative approaches is crucial especially in global development studies, which aims to explain and understand multifaceted development interventions and their socially differentiated impacts on targeted populations in the global South, as well as the discursive and institutional mechanisms involved (Hall et al. 2015;Lund 2016).Qualitative-quantitative integration is fundamental for understanding the processes and outcomes of multifaceted projects of land grabbing and resource extractivism (Borras et al. 2015;Hall et al. 2015;Leifsen 2017;Nygren et al. 2022), people's relocation due to conservation projects, or the construction of different sorts of infrastructure (Büscher and Ramutsindela 2016;Goh 2019;Harris 2012;Schindler et al. 2019), as well as the displacement-resettlement schemes linked to territorialization and strengthening of state authority in diverse contexts of the global South (Rasmussen and Lund 2018;Sud 2021).This is especially crucial under circumstances of legal pluralism and authoritarian governance (Kelly and Peluso 2015;Sekine 2021).
Under such conditions, quantitative methods provide opportunities to analyze the magnitude of changes and statistical significance of the relevant variables, thus enhancing understanding of the general patterns of change (Wayessa 2020).Simultaneously, qualitative methods offer ways to examine the contextualized meanings of changes and the interpretations of involved actors (Li and Semedi 2021).For providing information of the nuances, diversities, and exceptions within the general patterns, qualitative approaches assist capturing the implications of development interventions as experienced by targeted people.They are also important for exploring the embedded power relations, the situated politics, and the attributed values (Harris 2012;Kabra and Mahalwal 2019).Concerning resettlement schemes, an integrated MM approach provides opportunities to analyze the interlinkages between the multifaceted processes and outcomes, and the general patterns and variegated meanings.
The combination of qualitative and quantitative methods is closely related to the concept of triangulation.In a conventional positivistic rationale, triangulation assumed to identify and verify the existing reality through objective use of multiple methods (Blaikie 1991), whereas in constructivist-oriented approaches, it was used to deepen the understanding of a particular version of reality (Seale 1999).In current global development studies, the main purpose of triangulation is to understand the complexity and multidimensionality of the social world (Moran-Ellis et al. 2006).We consider MM as a form of methodological triangulation and a means to advance development research with an intention to explain and understand environmental-social changes in the global South, with multifaceted linkages to the global North (Horner and Hulme 2019;Lund 2010;Oldekop et al. 2020).
MM research may take various forms depending on the research problem and the resources available.A research problem can be conceived in a way that entails a dialectic application of deductive and inductive approaches (Creamer 2018).At data collection stage, qualitative interviews can be undertaken in combination with a quantitative questionnaire or a survey.During data analysis, insights from quantitative methods can help systematize the wider patterns in qualitative data, while qualitative methods can help contextualize quantitative results and discern their meanings.Thus, an integrated MM approach provides a dynamic way for expanding the scope and improving the analytic power of development research.
Simultaneously, several challenges are faced in conducting MM research in the global South contexts.First, formulating an appropriate research design can be difficult if adequate statistics on demography, land tenure, livelihood changes, and other variables of interest are lacking.Second, the gathering of sensitive qualitative data often requires time as they require a trust relationship, with a sense of fair co-creation of knowledge between the researchers and the informants (Lesutius 2018), and third, an integrated analysis of diverse types of data is highly demanding.The overall goal of an integrated MM research is to provide abstract generalizations, while offering contextually-rich understanding of specific characteristics of the phenomenon in question (Lund 2014).
The relevance of MM approach in displacement-resettlement studies connects to the complexity and multidimensionality of resettlement as a process and the sociopolitical dynamics involved.Just to consider Cernea's (2000) 'impoverishment risk and reconstruction' model, according to which nine interrelated risks of impoverishment are associated with displacement, including landlessness, joblessness, homelessness, marginalization, increased morbidity and mortality, loss of access to education, food insecurity, loss of common property, and social disarticulation.While some aspects of these components are quantifiable, others necessitate qualitative approach.Likewise, the variegated areas between the voluntary/involuntary dichotomies in resettlement situations can be better understood through MM approach, as we will show in our analysis.While certain dimensions of resettlement-related livelihood changes are quantifiable, people's experiences of them and their attempts at livelihood reconstruction are based on socially sensitive understanding of livelihood resources, and their cultural meanings, which require contextually-grounded, qualitative approach.

Materials and Methods
State-sponsored resettlement programs have decades-long history in Ethiopia, as they were incorporated in government policies in the mid-1960s (Piguet and Pankhurst 2009) and have since been pursued under different political regimes as a form of state territorialization and reinforcing state authority in politically tense regions.The government led by the Ethiopian People's Revolutionary Democratic Front (EPRDF) (1991-2019) formulated relocation of people as part of its Five-Year Strategic Plan in the early 2000s, putting resettlement projects among its top rural development strategies, and one of the main components of the Food Security Program (FDRE 2004).
In this study, the multidimensionality of the research questions, which often is the case in global development research, inspired the use of integrated quantitative-qualitative research methods.This includes the use of random sampling of respondents and purposive 'snowball' sampling of informants (Teddlie and Yu 2007), the formulation of a structured questionnaires and semi-structured, open-ended interview questions for data collection, and the use of statistical and qualitative-interpretative methods in data analysis.While random sampling is important in the case of quantitative questionnaires, in qualitative data gathering, there are always elements of purposive sampling and self-selection, as the informants must consent to be interviewed, which is pertinent especially when carrying out research on sensitive issues in politically fragile circumstances in the global South (Nygren 2004, p. 192).For the purpose of cross-checking, open-ended interviews were further complemented with methods of participant observation and informal conversations.
The study focused on the resettlement schemes conducted in 2003-2004.An intensive fieldwork was carried out in western Oromia from February through August 2009, thus drawing on five to six years of lived experience of the resettlement processes and outcomes by the respondents and interviewees.We chose western Oromia as a representative case of resettlement-displacement dynamics because massive resettlement schemes have been implemented in the region.The state-sponsored resettlements have indicated significant changes in access to land and other livelihood resources both among the resettlers and hosts.People's harsh experiences of displacement and injustice (Hammond 2011;Wayessa and Nygren 2016) have close links to forms of governing and implementing the resettlement programs.
Data were collected from 13 resettlement sites (Fig. 1), through qualitative interviews participant observation, and household survey.1 Furthermore, relevant resettlement policy documents and development reports were subjected to a qualitative content analysis.The sites to be studied were selected from a list of resettlement sites obtained from the Oromia Food Security Disaster Prevention and Preparedness Commission (OFSDPPC), which was the institution responsible for resettlement coordination in Oromia.We sought to include the prevailing variation in people's livelihood status in the site selection, since some sites were better off than others in terms of the quality and quantity of livelihood resources.
Eight resettlement sites were purposively selected for quantitative survey.The households of these sites were classified into resettlers and hosts.2From each stratum, households were selected randomly, indicating a stratified random sampling for the quantitative strand.Participants for qualitative interviews were identified purposively, in consultation with residents and based on the ethical principle of informed consent.The goal of the sampling design was to improve the precision of the quantitative estimates and their extrapolation to the target population and enhance the relevance of the findings beyond the studied cases (Teddlie and Yu 2007).
For the collection of qualitative data, we used interview checklists with semistructured questions and open-ended answers, whereas for quantitative data, we employed structured questionnaires.This procedure resonated with what Teddlie and Tashakkori (2009) call 'between-strategy' data collection, where two data collection instruments are used-one for qualitative and another for quantitative-for purposes of cross-checking and integrated analysis.Quantitative data was planned to provide information of the general trends in the distribution of land and other livelihood resources, while qualitative data was expected to give nuanced information of the formal and informal politics involved, and people's interpretations of them.Regarding data collection, the study adopted a concurrent MM design, where the collection of quantitative and qualitative data was undertaken simultaneously (Fig. 2).We surveyed 630 households and conducted 70 qualitative interviews.
The data collection was designed in a manner that it facilitated comparisons between the former and current settlements for the resettlers, and before and after the arrival of resettlers for the hosts.The questionnaires were pre-tested to make sure that the questions are contextually relevant.Quantitative analysis involved descriptive statistics and inferential statistics (α = 5%).For the analysis of the interview data, we used interpretative analysis, with a search for insights and aspects that complemented, questioned, and/or shaded light on quantitative findings.
Overall, quantitative research was carried out to explore the patterns of changes associated with the resettlement schemes, including the distribution of livelihood resources, whereas qualitative research helped uncover the how and why aspects of the changes.With qualitative analysis, it was possible to fill the gaps in explanation of the quantitative indicators of change and strengthen the analysis of the transformations occurred.Quantitative changes attributable to resettlement schemes were relevant also for policy debate, concerning the forms of decision-making in development interventions.However, what these indicators meant as perceived and experienced by affected people could only be explored through qualitative approach, with active engagement by local residents.Qualitative methods were crucial for capturing contextual meanings of interventions and understanding the socio-political circumstances under which people make their living and their livelihood decisions in Ethiopia, known for its authoritarian regimes of governance.The results of the study are presented in the following sections.

Resettler Voluntarism
Depending on the willingness of people involved, resettlement schemes are often categorized as voluntary or involuntary (Yntiso 2002).In a voluntary resettlement, people can choose between resisting displacement or risking resettlement, whereas the only option they have in involuntary situations is to take the risk of resettlement, sometimes at gunpoint.However, according to our analysis, this kind of distinction

InterpretaƟon and discussion of results
•Combined qualitaƟve-quanƟtaƟve analysis, crosschecking, and synthesis Fig. 2 The design for qualitative-quantitative integration in fieldwork-based data collection and analysis is often too simplistic.Concerning 'voluntary' resettlement in Oromia, two issues became worth considering.First, did people literally volunteer, by expressing willingness to move, and second, did they really volunteer through an informed consent?Answering the second question entails analyzing the existing and perceived conditions from which people were displaced and into which they were resettled.The authoritarian procedures by the Ethiopian government created highly restrictive political conditions.There is little space for people to claim resource justice, respect for human rights, and right for self-expression, all entailing meticulous methodological approach.An integrated MM approach provided an opportunity to analyze the official promises communicated and the prospects left uncommunicated, the political practices of implementation, and people's lived experiences.Through our analysis of qualitative and quantitative data, it became clear that the 'voluntary' resettlement involved substantial coercion.Our survey results showed that an overwhelming majority of the resettlers 'volunteered' to be relocated.This result came from descriptive statistics of responses to a question seeking 'yes' if they had volunteered and 'no' if not.However, it did not tell us about the differences among those who responded that they volunteered.It was rather the qualitative interviews which revealed that among the resettlers who gave their consent for relocation were those whose former lands were delineated by the government for reforestation.When asked if they volunteered, most of them replied 'yes'; when asked for the reasons, they explained that the government insisted it was a must to move although the relocation would be based on people's volition.The procedures of evicting people from their farmlands and providing them the option of being resettled 'voluntarily' highlighted the complexity of the process.As this complexity required a thorough understanding of the social-environmental and political-economic conditions of the resettlers' living places, a qualitative approach was crucial to explore these conditions.Moreover, the sensitivity of the topic required a relationship of trust between the researchers and the informants, which is central in qualitative approaches.
Our interview material shows that hierarchical power relations and the risks of marginalization had a strong effect on people's position vis-à-vis the state in 'accepting' the resettlement proposal.The government-led resettlement programs in Ethiopia have been characterized by '[i]nadequate planning and resourcing…and rushed timeframe, blocking of NGO and other independent monitors' access' (Hammond 2008, p. 517).Under these conditions, the voluntary-involuntary dichotomy proved too categorical to illuminate the dynamics involved (Rahmato 2003), as it does not allow for considering the nuances and ambivalences between the poles of volition and compulsion (LaRue et al. 2021).The options available to people and the circumstances under which they make decisions are crucial when judging if a resettlement is voluntary or not (Schmidt-Soltau and Brockington 2007; Wayessa and Nygren 2016), which is why the MM approach contributed significantly to enhanced understanding of the resettlement schemes.The quantitative data helped measure the quantifiable implications of the resettlement, instead of relying just on a few people's experience of the events.Simultaneously, the qualitative data revealed the problematic nature of strict categorizations when trying to understand the subtleties of the resettlement as a social process tied to hierarchical power relations and political negotiations, in which formal and informal decision-making procedures are tightly interwoven.
Our analysis of the policy documents in combination with the interview material revealed that exaggerated promises, although were not part of the resettlement policy discourses (OFSPCO 2002), constituted an overwhelming part of the implementation.In the interviews, people expressed how government officials promised to provide them with 'good quality housing' and 'good quality land across the board.'The relocation sites were portrayed as evergreen, and pictures of verdant coffee fields were shown to people who were later relocated to non-coffee-growing areas.Officials promised the resettlers to transport their livestock along, yet on departure day, they prohibited even taking along simple household utensils and farm tools.
The act of evicting people for the purpose of 'environmental reclamation' through state-initiated reforestation was not mentioned in the official documents.This objective became evident to us only when we visited the last resettlement site for this study, Tulama, where according to the survey results, the majority of those who did not volunteer to resettle (23.8%) were living, while all other sites constituted 16.1% combined.The analysis of qualitative interviews, in integration with the questionnaire data, unveiled the specific circumstances of these resettlers' removal.Yusuf, a resident of Tulama, stated in an interview that they would not have moved from their former settlement had they not lost their lands because of the government's reforestation plan: 'We refused to go.Then the district administrator said, those who are willing can go, and those who are not, can stay; it is voluntary and there is no coercion, but cultivating this land is strictly forbidden.' In some cases, the relocation involved outright violence.Suleyman, a resettler in Tulama, recounted a shocking violence: A man refused to recognize the law that forced him to leave his land, house, and properties.He asked if he had the right to sell his properties before leaving.When he refused and picked up his hoe to cultivate the land, a militia fired bullets.When the man picked up his meencaa [a tool used for land clearing but also in fighting], the militiaman shot him.[His wife] came out.She was pregnant.He… killed her too.
Many resettlers were essentially forced off their land, and some of those who refused to be resettled were marginalized during the land distribution, having to live on a smaller parcel of land or becoming landless.In this respect, our results resonated with those of Hammond (2011, p. 427), according to whom the officials threatened people to leave their lands or face imprisonment.An integrated MM approach helped us understand the conditions of such coercion.The quantitative data showed the commonality of feelings of unfairness involved in the relocation, while the interview data revealed the power differences and forms of pressure practiced by the governmental authorities.The interview data also illustrated people's perceptions of unfairness that the quantitative data indicated but was unable to elucidate as an everyday experience of injustice and suffering.

Host Consultations
As part of the implementation process and prior to moving the resettlers, the Ethiopian government sought to obtain the consent of the hosts for undertaking resettlement in their living areas.However, according to our interview data, the government's approach was largely one of reinforcing its power and giving directives for the relocation rather than genuinely seeking people's consent.Although there were serious concerns among the hosts about the land distribution, they did not promote open resistance because host communities felt that the government's policies were invincible.A sense of incapacity to challenge the status quo was highly evident in our data.The quantitative results indicated that most of the hosts (63%) strongly disagreed with the statement resettlement undertaken with the consent of the hosts (Fig. 3).The nuances of the forms of consultation and the power relations involved could not have been captured without further exploration through qualitative interviews and participant observation.
Public meetings were the common platform for host community consultations.In the meetings, the government officials emphasized that the resettlers share the same ethnicity as the hosts, which would, according to them, prevent experiences of displacement and long-term hostilities between the different groups.By appealing to hosts' sense of solidarity, the information that the resettlers belonged to the same ethnic group probably enhanced consent, or at least attenuated resistance among the hosts.The following quotes illustrate the state's tactics to get the consent and hosts' feelings of being cheated.Galata,3 a 33-year-old man, who resided in Kenaf and practiced carpentry besides farming, stated: During the Derg regime [1974][1975][1976][1977][1978][1979][1980][1981][1982][1983][1984][1985][1986][1987][1988][1989][1990][1991]…they brought resettlers belonging to another ethnic group.They were inciting conflict among people.This time… there were no problems related to cultural issues or living together, but there was a concern that both people will be exposed to problems because of land shortage.It would have been better if they had been resettled in areas that were not occupied.Now there is overpopulation here, and people have only two hectares or one hectare of land.Some people do not have even one hectare.
Galata continued highlighting how officials addressed the hosts: 'they told us that people, who belong to [our] ethnic group, cannot afford to eat even as much as [we] do because of land shortage and drought.Because these are [our] people, [they said] it is problematic if [we] enjoy and they suffer.'Considering that ethnicity evokes strong emotions in Ethiopia due to the country's turbulent and contested political history, the officials' repeated reference to ethnicity as a common heritage among the resettlers and hosts reflects the authorities' intention to persuade people to approve the decision they had already made.Many interviewees stated that the questions related to how the land and other resources will be distributed were never thoroughly discussed, nor were hosts engaged as stakeholders with a voice in decision making.Concerning the government's tactics during the host consultations, Dabala, a 53-year-old man, who previously worked for the state farm in Kenaf, explained: We were told that the resettlers are Oromo…who were affected by drought and lack of food.You must host them, they told us...so what else could we say?... Now we feel coerced.Having brought the resettlers, it confined us to smaller land and…we have disputes over land…[The government] said that '…you must withdraw from the land that was formerly state farm'.That was coercive, I mean particularly coercive!Dabala's reiteration of the term 'coercive' challenges the government's claim of meaningful consultation and informed consent.According to Dabala, part of his land was given to a resettler, which he considered an oppression ('hacuuccaa') and a lack of recognition.Other hosts reported that they were forced to leave the area after sowing crops, without any compensation.Sutuma, a 35-year-old man living in Lugama was less blunt in his responses, but his expressions reaffirmed a pattern that became evident in the integration of our survey and interview data: 'Yes, the chairman explained in the meetings that people who would resettle here are people like us…and that we should live together in harmony.' When asked if the hosts agreed resettlers' arrival, Sutuma explained, 'Yes, whatever occurs, can we farmers of Lugama stop it from happening?'Thereafter, he raised many questions, implicitly hinting at hierarchical decision making: 'What can we do against that which comes from above?What can we do if the government makes it compulsory?'When we asked citing the official guidelines, which declare that the resettlement should be based on the consent of the resettlers and hosts, Sutuma noted: 'Do you want to know my views?Then, I would say, we gave our consent for them to resettle here'.A careful analysis of this statement, together with other interview material and survey data, revealed the deeper, subtle meanings of Sutuma's statements, indicating that the host population did not willingly accept the relocation but rather gave their 'consent' under pressure in a context of authoritarian governance.Sutuma's questions indicated profound power differentials between the state authorities and the residents, where the state imposed its will on people, and people did not have the opportunity to question the procedures or to mount meaningful resistance.
In Tulama, Gada, a 35-year-old smallholder, expressed a strong sense of regret about giving consent, lured by promises that were never fulfilled.According to him, the officials promised that there will be enough land for everyone, which did not come true: I lost land.When the district administration called us for discussion, I asked what was considered for the landless, and for the future generations, noting that we had people who were…jobless.[The administrator] replied that Tulama land can accommodate twice as many people.When it was almost time to carry out the resettlement, district officials, including village elders, started demarcating the land boundaries…While the hosts were outcrying, officials redemarcated the land inappropriately, including houses [of the hosts] and farmlands with crops allocated [to resettlers].We had no option…He was the district administrator and had the power.We tried our best to refuse but [failed].
Tolasa, a young man in his twenties, added: [Government officials] said the resettlers would be relocated at a distance of a five-hour walk from the host community, and people agreed…they failed to relocate them even at a distance of a five-minute walk…The [hosts] had to leave their homes because the surrounding area was given to resettlers.As a result, some people became daily laborers to feed themselves, and many people starved.
Overall, a careful integration of the survey, interview, and participant-observation data showed that the distinctions, between the voluntary vs. involuntary resettlement, and the informed vs. forced consent, were blurred.This relates especially to contexts of authoritarian governance and hierarchical power relationships, which characterize Ethiopia and many other parts of the global South.The unrealistic promises, and the hidden forms of political pressure during resettler recruitment and host consultation revealed the unequal forms of negotiation and persuasion involved.

Farmland Size
The government promised to provide every resettler household with two hectares of farmland, regardless of household size. 4However, most of the households received less than two hectares.Table 1 shows the descriptive statistics of resettlers' and hosts' access to key livelihood assets before and after resettlement, including farmland.However, when integrating the survey and the interview data, it became clear that there were political and extralegal reasons for differentiated access to land despite the same promise.Some interviewees reported bribery and connections to village cabinet members as mechanisms for securing land.In some instances, resettlers who received land first were allocated two hectares, whereas those who received later were allocated less.In several cases, the population destined for a resettlement site was beyond the capacity of the site to provide each household with two hectares of land.
According to our survey data, for all the resettlement sites, except Tulama, the mean size of the resettlers' farmlands was higher after the resettlement (Fig. 4).The differences in the average farmland size between new and former settlements were pronounced for resettlers in Jirma (1.38 ha), Baqqo (1.31 ha), Machara (1.18 ha), Fig. 4 The average farmland size of resettlers and hosts before and after resettlement according to resettlement site and Dhidhessa (1.08 ha).Among the resettlement sites, the lowest means were recorded for Lugama (1.16 ha), Tulama (1.32 ha) and Chawaqa (1.42 ha).Contrary to that of the resettlers, the average farmland size for the hosts was smaller after the relocation in all the resettlement sites.These were Kenaf, Dhidhessa, and Jirma, with losses of 1.25 ha, 1.16 ha, and 1.02 ha, respectively.Thus, our quantitative analysis shows a general increase in farmland size for resettlers and a decline for hosts.Examined through the concept of displacement (Cernea and Schmidt-Soltau 2006), this suggests that the hosts were displaced due to the resettlement.The results of our qualitative analysis further reinforced our survey results, illustrating people's experiences of land losses and the socio-political conditions under which they took place.According to the survey results, land was unevenly distributed among the resettlers.Most of the respondents reported obtaining two hectares of land, while some received more.Our further examination of the land distributions through qualitative interviews revealed that some resettlers obtained six or more hectares of land.The quantitative results showed the pattern of land distribution, while the qualitative data revealed its how and why aspects.Only after careful integration and cross-fertilization between the survey results and interview results, we were able to comprehend the process that led to the inequitable land distribution.
Considering land loss by the hosts, quantitative measures illustrated the extent, and showed how common and varied the loss was among the sample and, by extrapolation, among the target population.The magnitude of the loss-which we were able to reveal through quantitative data-must have had an influence on the meanings people attached to it.Considering the small size of their farmlands, losing one hectare is different from losing two hectares of land.A loss suffered through a coercive process is felt differently from a loss to which one has given an informed consent, or a loss for which a fair compensation is received.Whereas quantitative methods revealed the extent and pattern of the distribution of such losses, qualitative methods uncovered what they meant to people.This is crucial as most of the hosts had cultivated the land for generations and had a strong cultural attachment to living from the land.

Access to Irrigation
To compare access to irrigation in current and former settlements, the proportion of irrigated land was calculated as a ratio of the size of irrigated land to the total size of farmland.The comparison showed a general decline in access to irrigation for resettlers.There was a wide range, from 4% increase for Dhidhessa resettlers to 30% decline for Kenaf resettlers.On average, the highest loss in access to irrigation was endured by resettlers in Kenaf, followed by those in Baqqo.Through qualitative interviews, it became clear that losing access to irrigation meant losing a survival strategy.Irrigation helped farmers to minimize the risks of rainfed agriculture and associated vulnerability by providing the possibility of producing more than one harvest per year.In their former settlements, the resettlers survived several years of rainfall shortage with the help of irrigation.Before the relocation, the resettlers were informed that rivers are abundant in destination areas and that the topography is so plain that water for irrigation can be diverted without pumps.While recalling this rhetoric, resettlers in Baqqo noted that they had contributed money in groups to buy the pumps since the area was not as convenient as promised.
Our interview material also revealed the crucial role and meaning of land in farmers' livelihoods and lifeworld.In an interview, Abdi, who lived in Kenaf, commented: Our lives depend on land...We are very concerned about land.We do not know what will happen to us….Moreover, there is no irrigation….What we produce once a year is little because of poor land conditions.Despite toiling heavily, people fell under destitution and are now struggling like a baby that has recently started crawling.There is nobody who is optimistic about our livelihood conditions.
During the interview, Abdi, as many others, expressed strong concerns of land and other material and non-material aspects of their livelihoods, including struggles over access to irrigation.The metaphor of 'crawling' gave a 'physical' manifestation to people's struggle to make a living with few resources and scant opportunities.This scenario was not an isolated incident affecting few individuals, as Abdi stated: 'there is nobody who is optimistic about their livelihood conditions.' Besides land size, land quality and soil management practices affect the productivity of land, which can be measured in terms of crop yield per unit of land.However, having realized errors in the data gathered on crop yield, we dropped this aspect from our analysis because our interview data and field observations contradicted the quantitative data on crop yield.We consider this experience as a reminder of the ability of MM approach to provide possibilities for verifying the reliability of results by cross-checking them with different types of data.

Livestock Husbandry
Changes in livestock husbandry were analyzed through livestock ownership measured in tropical livestock unit (TLU), a standardized and aggregate measure computed using conversion factors.Each livestock type was valued in terms of TLU, and these values were summed up to produce total TLU as presented in Table 2.The results show that livestock ownership generally decreased for both resettlers and hosts.Figure 5 disaggregates the results according to resettlement sites.The graphs and tables indicate the general pattern of changes in livelihood resources.However, they do not show whether the differences between the two "moments"-before and after resettlement-are statistically significant or not.To address this gap, we performed mean comparisons of changes in access to farmland, irrigation, and livestock ownership using the Wilcoxon Signed Rank Test.The results are presented separately for resettlers and hosts.

Resettlers
There was a statistically significant increment of farmland size as resettlers moved to the new settlements, z = − 9.35, p < 0.001, with medium effect size (r = − 0.34).The median score on the size of landholding increased from pre-resettlement (Md = 0.63) to post-resettlement (Md = 1.5) (Table 3).Following similar interpretation principle, the situation after resettlement was statistically significantly different from the situation before the resettlement for all the variables in question -farmland size, proportion of irrigated land, and livestock ownership.
There were significant increments in farmland size, whereas there were significant declines in the proportion of irrigated land, and livestock ownership.Regarding change in farmland size, the situation in Tulama exhibited a different scenario, which was evidenced also in qualitative interviews.According to the interviews, part of the lands allocated for resettlers were from the vicinity of the Birbir River, about three-hour walk from Tulama.The other part was provided close to their homesteads.Notwithstanding the problem associated with the distant location, the land cultivation in Birbir was worsened by the presence of hippopotamuses, which destroyed crops.The qualitative information about plot distance was confirmed through quantitative analysis which revealed a significant  difference among resettlement sites in the distance of plots from homesteads (H(8) = 66.97, p < 0.001).5 Tulama exhibited very high median (Md = 150.77).
The analysis thus far shows the general picture of the gains and losses the resettlers experienced after relocation.However, we do not yet know if they varied according to resettlement sites.To tackle the question of statistical significance of variation in farmland size, proportion of irrigated land, and livestock ownership, we computed changes in these variables as the difference between their values in former and current settlements (Change = Value after relocation-Value before relocation).Then, we tested if they showed statistically significant difference across resettlement sites, using the Kruskal-Wallis Test.The analysis shows a statistically significant difference in farmland gain across the resettlement sites, x 2 (7, n = 385) = 81.26,p < 0.001 (Table 4).The changes in proportion of irrigated land are also significant at p < 0.001.This indicates that the losses/gains were influenced by differences in resettlement sites.
The statistical non-significance of the differences in livestock ownership by resettlement site could be misleading if not critically assessed considering the prevailing and emerging problems in most sites.While the prevailing problems may have shown in current quantitative results, which have not exhibited statistical significance, the emerging threats could only be explored qualitatively.One of the major problems was the death of livestock due to trypanosomiasis, a livestock disease common in resettlement sites.Another major constraint was the lack of grazing land.These problems limited the survival and reproduction of livestock and affected people's decisions in investing in livestock husbandry, as expressed by Ibrahim in Baqqo: When we came, there was a swampy area [which could be used for grazing].That area is now divided among the host community, and they are cultivat- ing it…In our village, thanks to Rabbii [God], everybody has at least seven or eight livestock.Some have more.
[However] because of the shortage of grazing land, everybody is getting…discouraged.
In Kenaf, the hosts' cultivation of the riverbanks formerly used for grazing resulted in alarming shortage of grazing land.The process was closely linked to the eviction of the hosts from the lands previously under their cultivation due to increasing national and foreign land investments, or 'land grabs', in which the residents did not have any voice (Fig. 6).The land appropriation or leasing by investors was reported in most resettlement sites, and although its impact was felt disproportionately, land grabbing was a serious source of threat in all the resettlements sites.This trend is evident in many parts of the global South (Borras et al. 2011;Hall et al. 2015;Wayessa 2020).

Hosts
The hosts also experienced changes in farmland size, proportion of irrigated land, and livestock ownership.The Wilcoxon Signed Rank Test indicated a statistically significant reduction of farmland size for hosts, z = − 7.94, p < 0.001, with large effect size (r = − 0.51).The median score on the size of farmland decreased from pre-resettlement (Md = 2) to post-resettlement (Md = 1) (Table 5).Farmland size and proportion of irrigated land showed medium and large effect sizes.
To see the statistical significance of variations across resettlement sites, we computed changes in these variables as the difference between the respective values before and after the resettlement.The Kruskal-Wallis Test showed statistically significant difference among resettlement sites in the loss of farmland, x 2 (6, n = 243) = 22.25, p = 0.001 (Table 6).All the variables were significant at p < 0.01.Kenaf recorded the highest loss (Md = − 1.25), followed by Jirma (Md = − 1) and Dhidhessa (Md = − 1).Interestingly, these sites were among those that showed the highest gains for resettlers in terms of farmland size.The juxtaposition of the survey results of resettlers and hosts showed that the hosts were displaced by the resettlement.This was confirmed by qualitative data, which indicated that the hosts in Kenaf were told to retreat from cultivating their lands and confine their cultivation to the riverbanks.
Our integrated MM analysis showed that even when current figures do not indicate that a variable is statistically significant, qualitative results may hint at phenomena with significant implications before quantitatively verifiable change has happened.For instance, the decline in resettlers' livestock ownership was not statistically-significantly influenced by differences in resettlement sites.Although a 'strict' quantitative orientation would suggest dropping this variable from further analysis, we were able to ascertain its social significance by qualitatively examining the trends in the resettlement sites.Employing MM approach enables addressing the statistical and substantive significance of the variables and the diversity of meanings associated with development interventions.It also enhances the methodological-analytical power of research and the policy relevance of the results, potentially informing opportunities for more just and sustainable pathways in the global South.

Conclusion
This study has examined the processes and outcomes of state-sponsored resettlement schemes in Ethiopia as a case for MM approach in global development studies.The study employed a quantitative analysis of changes in access to livelihood resources and the impacts of the resettlement on the livelihoods of resettlers and hosts, combined with a qualitative analysis of the resettlers' and hosts' experiences and perceptions of the resettlement schemes.While quantitative analysis demonstrated the patterns and the magnitude of the impacts of the resettlement, qualitative analysis helped understand the meanings of the resettlement policies and practices as a sociopolitical process, and the associated lived experiences of the resettlers and hosts.The results of the study show resettlement as a process in which distinctions between voluntary vs. involuntary relocation and informed vs. forced consent are blurred and tricky.This is relevant especially in the contexts of authoritarian governance and hierarchical power relationships, characteristics of many parts of the global South.The careful integration of quantitative survey data and qualitative interview and participant-observation data showed that the 'voluntary' resettlement involved substantial coercion in the case of the resettlers.Correspondingly, the hosts gave their 'consent' for the resettlement under political pressure.The survey data showed the general tendency in people's exercise of voluntarism, while the qualitative data revealed how the unfulfilled promises, and the hidden forms of persuasion and coercion practiced during resettler recruitment and host consultation were interlinked with high power differentials between the state authorities and the residents, where both the resettlers and hosts had scant opportunities to challenge the procedures.
The results of our MM analysis revealed that the hosts suffered land losses and associated displacement due to the resettlement.Examination of the changes in access to land and other livelihood assets through a quantitative lens elucidated the magnitude of the resettlement's impacts and provided an angle to cross-check the informants' memorized experiences, while qualitative interviews provided nuanced understanding of the resettlers' and hosts' differentiated access to land and other resources, as well as of the complexity of the power relations involved.
Besides thematic-empirical contributions, this study enhances conceptual-methodological understanding of the advantages and challenges involved when analyzing resettlement-displacement projects in the global South.The quantification and qualitative explanation of changes in access to livelihood resources is important for advanced theorization and development practice, both relevant for global development studies.The same holds true when studying multidimensional aspects of land grabbing, resource extractivism, and infrastructural development projects, and the associated environmental-social impacts.These are critical areas of engagement in global development studies.The operationalization of concepts, design of data collection instruments, contextually-sensitive data analysis, and the generalizability and particularity of the results all require considerable knowledge of the study areas.This is critical especially in many contexts of the global South, where the political regimes of governance, systems of land tenure, plurality of resource rights, entanglements between formal and informal ways of decision-making, and diversified meanings attached to land and livelihoods tend to be highly complex (Brockington 2019;Lund 2016;Nygren 2021;Sud 2017).
There are advantages and challenges in employing an integrated MM approach in global development studies.The challenges include the long-term empirical research required for data gathering and the difficulties of getting valid survey data and reliable interview data on socially sensitive topics, especially in authoritarian and politically fragile circumstances.Nevertheless, often the only option to get evidencebased and contextually-grounded data under such conditions is through empirical MM research, which offers possibilities to triangulate data and cross-check the results.This is essential in contexts where quantitative data from official statistics are often out-of-date and fragmentary.Thus, qualitative data gathered through field research improves the probability that the results are evidence-based and contextually grounded.
The justification for combining qualitative and quantitative methods entails evaluating the advantages in comparison to employing only one type of method.This study has shown that quantitative methods can offer insights into the magnitude of changes and statistical significance of the variables involved, thus enhancing understanding of the patterns of change, the extrapolation of results to the population in question, and the application of aggregate measures in estimating changes.Correspondingly, qualitative methods improve understanding of contextualized meanings of quantifiable changes, while enhancing knowledge of unquantifiable dimensions of transformations.Capturing the implications of development interventions as experienced by people involved, qualitative methods provide significant information of the nuances, diversities, and exceptions within general patterns.They are also crucial in understanding embedded power relations and attributed values of interventions, as well as their implications according to intersecting dimensions of social differentiation, including ethnicity, gender, and age.Through MM approach, this study was able to analyze the complex interlinkages between the processes and outcomes, and the general patterns and variegated meanings, of the resettlement schemes.It can inspire researchers in global development studies to examine the processes and outcomes of a wide variety of development interventions in the global South through innovative applications of MM approach.

Fig. 1
Fig. 1 Study sites covered during fieldwork for data collection

Fig. 3
Fig. 3 Hosts' responses to the statement resettlement undertaken with the consent of the hosts (N = 242)

Fig. 5
Fig. 5 Average livestock ownership of resettlers and hosts in TLU by resettlement site before and after resettlement

Fig. 6
Fig. 6 Indian investors' mango plantations in Kenaf (left) and Pakistani investors' sugar cane fields and sugar factory in Lugama (right)

Table 1
Descriptive statistics of farmland size (ha) and proportion of irrigated land before and after resettlement

Table 2
Livestock ownership (TLU) before and after resettlement, for resettlers and hosts

Table 3
Comparisons of resettlers' farmland size, proportion of irrigated land, and livestock ownership before and after resettlement (z represents a data point expressed in standard deviation unit)

Table 5
Comparisons of hosts' farmland size, proportion of irrigated land, and livestock ownership before and after resettlement