Introduction

Certain framings such as mathematical modelling, risk and impact assessment studies, and economic assessments have dominated climate change research (Ford et al. 2010; Bjurström and Polk 2011; Jurgilevich et al. 2017), driving it conceptually and methodologically. Several authors have called for pluralising the methodological approaches and languages we use to talk of, respond to, and prepare for climate change (Hulme 2011), highlighting potential contributions from disciplines such as geography (Hulme 2008), anthropology (Cutter 2003), and history (Adamson et al. 2018) in enriching discourses around climate change.

Climate change vulnerability and adaptation (CCVA) research, in particular, has been criticised for overlooking the dynamic nature of vulnerability, and how risks and responses interact to shape local adaptation (Ribot 2014; Bennett et al. 2016; Fawcett et al. 2017; Adamson et al. 2018). Studies are now demonstrating the value of using different methodological and epistemic bodies of work to address this: from case analogues to study temporal vulnerability (Ford et al. 2010; Fawcett et al. 2017), mixed-methods approaches for mapping shared socio-economic pathways (Birkmann et al. 2015), and narrative approaches for adaptation planning (Paschen and Ison 2014).

In this paper, we add to and extend this body of work by demonstrating the value of applying life history (LH) approaches to CCVA research. As part of a 5-year, multi-country, interdisciplinary project Adaptation at Scale in Semi-Arid Regions (ASSAR www.assar.uct.ac.za), we examined socially differentiated vulnerability and the role of migration in a household’s suite of risk management strategies. To provide a counterpoint to the dominant methodological tools of household surveys and focus group discussions (FGDs) in environmental migration studies (Piguet 2010), we report our experiences of using LH interviews. We find that LHs examine migration and livelihood choices in a temporally sensitive manner that acknowledge climatic risks as one of the many factors affecting household responses (Bagchi et al. 1998; Nielsen and Reenberg 2010) and are a people-centred approach (Adger 2010) examining decision-making in risk-prone environments (Ayeb-Karlsson et al. 2016). This paper charts our experiences and identifies the strengths and limits of using LH interviews for CCVA research.

There are three critical gaps in CCVA research that LH approaches can help fill. First, the insights on dynamic vulnerability and temporality that LHs offer are especially relevant when studying climate vulnerability, livelihood trajectories, and migration because they are dynamic and shaped by past experiences and future aspirations. Second, adaptation studies continue to focus on households as the appropriate unit of analysis without an equal emphasis on intra-household dynamics. Third, how people perceive risks and respond to them remains under-researched despite a growing acknowledgement of the role of risk perception in adaptation decision-making (Singh et al. 2016b; Burnham and Ma 2017). We argue that a LH approach enables us to, in part, address these gaps and expand the range of methodological options available to researchers.

In the subsequent sections, we review current methodological approaches used in CCVA research and their gaps (in ‘Current methodological approaches to climate vulnerability and adaptation’) followed by an overview of LH approaches and the research design we used (section on ‘Life history approaches’). We then report from our experiences of applying a life history approach to climate change research. We end with a discussion on the the values and limitations of LHs to CCVA research.

Current methodological approaches to climate vulnerability and adaptation

How have people studied vulnerability and adaptation?

Vulnerability and adaptation research has a diverse conceptual legacy (Ford et al. 2010; Malone and Engle 2011; Joakim et al. 2015; Jurgilevich et al. 2017; Mortreux and Barnett 2017), moving from stressor-response approaches from hazard and disaster management studies (Füssel and Klein 2006) to livelihood approaches that articulated vulnerability as dependent on five capitals—human, social, natural, financial, and physical (Clay 2018). Poverty and development studies added to this by highlighting the importance of entitlements, capacities, and capabilities to vulnerability reduction (Sen 1981; Roy and Venema 2002; Adger 2006). Resilience studies went on to emphasise how multi-scalar linkages and cycles of repair and renewal shape vulnerability and adaptive capacity (Cannon and Müller-Mahn 2010; Tanner et al. 2014). From the social sciences, studies have focussed on the multiple, incremental drivers of vulnerability (O’Brien et al. 2004; Ribot 2014; Nyantakyi-Frimpong and Bezner-Kerr 2015), and value-based and justice approaches highlight how certain adaptation pathways can create winners and losers (Eriksen et al. 2011). More recently, a growing focus on linking risk perception and adaptation behaviour (Grothmann and Patt 2005; Nielsen and Reenberg 2010; Granderson 2014; Singh et al. 2016b; Burnham and Ma 2017) is beginning to explain how adaptive capacity leads to adaptation outcomes (Mortreux and Barnett 2017).

The conceptual diversity in CCVA studies also predisposes research to follow specific methodological approaches (Ford et al. 2010; Singh et al. 2017; Adamson et al. 2018). These include qualitative case study-based methods (Campeanu and Fazey 2014; Joakim et al. 2015), spatially mapping vulnerability hotspots (Varadan and Kumar 2015; Bouroncle et al. 2016), indicator-based assessments (Hahn et al. 2009; Gupta et al. 2010; Gerlitz et al. 2017), participatory methodologies that co-produce adaptation solutions (van Aalst et al. 2008; Fazey et al. 2010; Howe et al. 2013; Ayantunde et al. 2015; Butler et al. 2016), and more recently, historical and temporal approaches to examine trajectories of change (Nielsen and Reenberg 2010; Fawcett et al. 2017; Adamson et al. 2018; Gajjar et al. 2018). This methodological diversity is detailed in Supplementary Material 1.

Within the umbrella of CCVA research, environmental migration (often identified as a key risk management and/or adaptive strategy in developing countries) is also explored through various methodologies that focus on temporality, livelihood choices, and household dynamics (Piguet 2010; Fussell et al. 2014; Neumann and Hilderink 2015; Arnall and Kothari 2015). Piguet (2010) identifies six families of environmental migration methodologies: spatial and multi-scalar analyses, large-N surveys, historical analogues, hotspots approaches using vulnerability indices, and qualitative field case studies using ethnographic methods. Migration research has also drawn on various disciplines with growing acknowledgement of the value of ethnographic (McHugh 2000), life history (Locke and Lloyd-Sherlock 2011; Singh 2018), and biographical (Erel 2007) approaches in understanding the ‘“quantum haze” of human mobility’ (McHugh 2000:72).

The scholarship on environmental migration has conceptualised livelihoods as inherently dynamic, acknowledging multiple, multi-scalar drivers, where agency and structure interact and feedback on one another (Massey et al. 1999; Bardsley and Hugo 2010; Castles 2010; Bakewell 2010; Ransan-Cooper 2016). This has led to a methodological expansion where issues of identity, history, culture, and values are being examined more rigorously through qualitative methods focussing on temporality and migration decision-making (Piguet 2010; Ransan-Cooper 2016; Piguet et al. 2018).

Gaps in the literature

Several gaps remain in the current literature. First, there is increasing focus on understanding drivers of vulnerability and what mediates adaptation but relatively fewer studies exploring how these drivers of vulnerability interact to shape household response behaviour. Despite growing evidence of barriers to adaptation, there is a tendency to identify if there are barriers and describe which types they are (Biesbroek et al. 2013) rather than how they interact and shape responses. Several studies examine factors shaping adaptation, i.e. decision-making around climate change responses (van der Linden 2015), but quantitative approaches tend to mostly use large-n, structured surveys while qualitative approaches typically draw on semi-structured individual or group interviews. While these approaches are useful to explain broad drivers of responses, they often give an inadequate understanding of individual decision-making (notable exceptions include Grothmann and Patt 2005; Nielsen and Reenberg 2010; Granderson 2014; Singh et al. 2016b; Burnham and Ma 2017).

There are growing calls for unpacking CCVA studies to understand the role of intra-household dynamics in mediating risk management choices and their outcomes (Djoudi et al. 2016; Rao et al. 2020). Empirical evidence from gender and feminist political ecology is filling this gap by focusing on intra-household bargaining and gendered vulnerability (Rao et al. 2020; Rao et al. 2017; Singh 2019), but these remain peripheral to mainstream CCVA research (Ravera et al. 2016; Thompson-Hall et al. 2016).

CCVA research in general and environmental migration in particular have also been critiqued for a relative silence on temporality (Obokata et al. 2014; Fawcett et al. 2017; Adamson et al. 2018) and overreliance on static methodologies such as household surveys to examine inherently dynamic processes such as adaptation (Singh et al. 2017). This counters growing evidence that household vulnerability is embedded in ‘pre-existing precarity’ (Ribot 2014: 673) and a result of complex structural conditions (Tschakert et al. 2013), which are often not captured by static, snapshot methodologies (Fawcett et al. 2017; Singh 2018). Different disciplines have used various approaches to study temporality such as case analogues, trend studies, pathways approaches, and oral histories (Supplementary Material 1). However, these methodological approaches tend to occupy a niche within CCVA studies. Cross-disciplinary work has recently broadened the methodological repertoire of CCVA research. For example, Adamson et al. (2018) forefront the role of historical approaches in developing long-term place-specific empirical studies, tracking institutional evolution to understand how historical events and decisions inform and constrain present adaptation, and providing a compass to follow discursive shifts and check historically grown power structures.

Finally, the implications of changing aspirations are seldom examined in adaptation research and practice. With growing recognition that adaptation is often about behavioural shifts (García de Jalón et al. 2015; Davidson 2016; van Valkengoed and Steg 2019), there is a need to understand how shifting aspirations mediate adaptation behaviour (Rao et al. 2020). Household surveys have been largely used to determine aspirations and motivation but usually do not capture intra-household dynamics and the context for individual aspirations within the household.

Life history approaches

What are life history approaches?

Life history (LH) approaches refer to a suite of approaches that typically involve open-ended interviewing to draw on life experiences and provide rich descriptions of a ‘life as a whole’ (Atkinson 2002, p. 123). Thus, LHs narrate not only life stories but also perceptions, memories, aspirations, and experiences that shape them (Elder 1998, 2012; Atkinson 2002; Chimhowu and Hulme 2006; Locke and Lloyd-Sherlock 2011; Wimark et al. 2017; Singh 2018). In doing so, they go beyond ‘methods and approaches…defining and addressing problems’ to examining ‘the intersection of nature, humanity, and technology in multiple ways’ (Moezzi et al. 2017).

LHs go beyond typical semi-structured interviews by focussing on narratives, i.e. exploring ‘a sequence of events into a whole so that the significance of each event can be understood through its relation to that whole’, thereby conveying ‘the meaning of events’ (Elliott 2005:3 in Lewis 2008). While semi-structured interviews can also capture temporality, they are relatively limited in emphasising how life events and key decisions are embedded in wider dynamics often beyond the individual’s control (Singh 2019; Chimhowu and Hulme 2006; Singh 2018). LHs thus ‘link personal experience with broader patterns of institutional change’ (Lewis 2008: 561) connecting individual and the social (and at a conceptual level, structure, and agency) within the social sciences. Most critically, LHs reconfigure interviewer–interviewee power dynamics by placing value on peoples’ perceptions of their life story (Locke and Lloyd-Sherlock 2011), thus ‘humanising research’ and giving ‘history back to people in their own words, potentially rescuing it from dominant elitist discourses’ (Thompson 1988: 265).

LHs have specific characteristics that are valuable for CCVA research. First, they delve into ‘micro-historical (individual) experiences within a macro-historical (history of the time) framework’ (Hagemaster 1992, p. 1122) thereby capturing cross-scalar linkages critical for household risk management (Adger et al. 2005; Osbahr et al. 2008; Eakin et al. 2009; Shucksmith et al. 2016). Second, LHs allow for a nuanced understanding of individual decisions that explain motivations and perceptions, challenging assumptions that decisions are discrete, rational choices (Hagemaster 1992; Atkinson 2002). This is in line with recent CCVA research that demonstrates the iterative, and multi-causal nature of adaptation decision-making (Grothmann and Patt 2005; Singh et al. 2016b; Burnham and Ma 2017). Third, LHs forefront the temporal nature of lives and livelihoods (Murray 2002; Locke and Lloyd-Sherlock 2011), giving insights into how household decisions are shaped by dynamic risks, changing aspirations, and consequent response pathways (e.g. Singh and Basu 2019; Singh 2019). This allows LH narratives to explore the dynamic contexts within which adaptation decisions were taken and can potentially help ‘see’ future decisions and identify entry points to shape adaptation behaviour. Critically, the focus on temporality goes beyond typical methodologies in livelihoods research that tends to be ahistorical (de Haan and Zoomers 2005) and assume risks and responses are static.

In development studies particularly, LHs have been used to examine poverty (Kothari and Hulme 2004; Bird and Shinyekwa 2005), well-being and economic empowerment (Locke and Lloyd-Sherlock 2011), and policy development and its implications (Chimhowu and Hulme 2006; Lewis 2008). Although LHs have ‘now developed into a significant, theoretically dense, and diverse sub-set of historical and social-scientific enquiry’ (Godfrey and Richardson 2004:144), they are not used in mainstream CCVA research which tends to be dominated by surveys, semi-structured interviews, and risk and impact modelling and assessments (see Supplementary Material 1).

There is a relatively small but growing body of CCVA research using LHs (or some form of them) to examine household trajectories of resilience and vulnerability (Bagchi et al. 1998; Murray 2002; Sallu et al. 2010; Shucksmith et al. 2016; Singh 2019; Singh & Basu 2019), livelihood dynamics and their implications for household resilience (Rogaly and Coppard 2003; Ayeb-Karlsson et al. 2016; Singh and Basu 2019), and as a way to creatively understand and communicate climate concerns (Moezzi et al. 2017). These studies demonstrate that LHs go beyond case studies by ‘providing respondents the chance to reflect on their own stories and report what they feel were crucial turning points rather than just submit a chronology of life events to be examined and interpreted by researchers’ (Singh 2018 p., 3).

It is necessary to highlight that LHs are limited in certain ways—they are not amenable to generalisation (Lewis 2008) and cannot describe broad patterns in large populations. In ‘the section Recognising the limits of life histories’, we reflect on these limitations based on our experiences.

Operationalising life histories: research design and data collection

We draw on data collected in communities living in semi-arid regions of four countries; Ghana, India, Kenya, and Namibia. The research was part of a larger 5-year study, and while we report experiences of conducting research on migration, livelihood dynamics, differential vulnerability, and adaptation outcomes in this paper, other methodologies were used to capture dynamics in the wider social–ecological system. These other methods included structured household surveys (to capture demographic details, climate change perceptions, livelihood dynamics and risk management strategies, migration behaviour and well-being outcomes), and qualitative and participatory approaches such as group discussions,Footnote 1 Vulnerability and Risk Assessment (Morchain and Kelsey 2016) and mental modelling (Rivers III et al. 2018). We do not describe each method in detail since it is beyond the scope of this paper but suggest references for those wanting to understand these approaches better.

In order to capture the diversity of households within the research locations, we used a purposive sampling strategy developed based on scoping visits and initial household surveys. The researchers iteratively co-developed a set of reasons to include/exclude households based on socio-economic characteristics, livelihood portfolios, and household composition and headship (Table 1). Based on these criteria, fifteen households were studied across the four sites.

Table 1 Criteria used to sample households in the research sites
Full size table

Following the research objective (using narratives to understand how individuals within a household make response decisions, negotiate their risk landscape, and choose to migrate or not), the household was treated as a case and the main unit of analysis. Within each case, linkages were made to larger scales (institutional changes, biophysical dynamics) and finer dynamics (intra-households relations, well-being, and individual aspirations).

For this study, LHs were employed as a method to help examine changing relations of individuals within the household, set against the broader landscape of the community/socio-cultural context and larger institutional trajectories. Typically, two members within each household were interviewed to examine dynamics in their decision-making processes in relation to each other. In some cases, this was not possible due to local conditions. For example, in a Muslim household in India, the research team (female interviewer, male translator) was only able to speak to the male household head. Despite a good rapport built over several months, the team was unable to speak to the wife, in keeping with socio-cultural norms and the need to respect local sensitivities and customs.

We followed an open-ended interview process to explore personal accounts of the temporal nature of vulnerability, risks that their household (and they as an individual) are exposed to and how they respond, how and if mobility featured as a household response strategy, and perceptions of how well the household is doing. The LHs typically recalled changes over a 10-year period (some going further back to 30–40 years) and were split into three parts (preceded by the informed consent process). The first part generated quantitative data on the household. The second part was unstructured and focused on (1) significant life events and decisions undertaken, (2) changing household structure and decision-making, (3) shocks and stresses faced, and (4) reflections on drivers and outcomes of migration/immobility. The third part sought opinions on subjective and relational well-being, and individual and household hopes and aspirations for the future.

The interview transcripts and researcher notes were translated (where necessary) and transcribed. The transcriptions were coded by hand using a set of descriptive codes based on emerging themes from the interviews. The codes were co-developed between the researcher team through two workshops to facilitate comparative analysis. The four large code families were (1) description of livelihood dynamics (risks, responses, type of livelihood shifts), (2) intra-household dynamics and decision-making, (3) outcomes of responses (e.g. trade-offs, well-being outcomes), and (4) pathways (cycles of poverty, future aspirations). The analysis was an iterative and collaborative process, facilitated by the small dataset and intimate understanding all the researchers had with the context and households studied.

Applying a life history approach to climate change research

To examine the implications of the LH approach on CCVA research, we draw on multi-country data which examined household and intra-household risk perception and management, livelihood choices, and migration decision-making. In the following sections, we juxtapose findings from structured household surveys, FDGs, semi-structured group interviews (SGIIs), mental modelling, and participatory vulnerability assessmentsFootnote 2 with data from LHs based on four themes: drivers of responses (why people act the way they do), intra-household dynamics and power relations, temporality (changes in livelihoods, risk portfolios and response strategies over time), and aspirations.

Drivers of responses

The household surveys were useful in drawing patterns of livelihood shifts and the role of migration within them. For example, in South India (n = 825), we found 39% households migrated (81% of these being men) in search of better jobs (56%), to escape unprofitable agriculture (26%) and to overcome debt (8%). These patterns were one-time snapshots of livelihood decisions and did not examine how household responses to dynamic risks change over time. When discussing risks within the survey, respondents were asked yes or no questions on risks they faced and their relative importance, which helped understand what risks were faced but not why and how.

In Namibia, the survey (n = 647) recorded shocks people experienced (from a list of 20, determined through previous scoping visits and participatory rural appraisal methods) and responses undertaken using a list of ‘coping strategies’.Footnote 3 The data showed that early maturing crops (66%), changing timing of planting (68%), and spatial planning (15%) were the most common coping strategies for crop farming, while supplemental feeding (51%), moving livestock (26%), destock (11%), and changing herd mix (11%) were the most common coping strategies for livestock farming. In this way, the survey captured the range of response strategies households employed but did not explain the drivers behind these choices. In the LH interviews, all strategies were seldom discussed but those discussed were better able to capture detailed motivations behind response decisions, the decision-making process, and outcomes of responses. Some of these ‘value additions’ that LHs provided compared with structured household surveys are explained in Table 2.

Table 2 Value added by life histories across the four research sites
Full size table

The LHs also served to highlight how multiple risks are experienced and intersect to shape daily realities. For example, a 48-year-old man who commuted between his village and a large city in India said,

Growing up, I have seen my father struggle to feed the family. The only way he could do so was by incrementally selling off his land… I dropped out of school very early and became a wage labourer to supplement my family’s income… People can make as much money in the village as they do in the city. However, there are social constraints. Caste is the main one; it governs … what circles you are part of, what livelihoods you are allowed to practice, what support and opportunities you can access, and … what things you can acquire for yourself and your family… I commute to Bangalore daily and work there as a gardener…At least my children have finished secondary school and can support themselves and the family, if needed. [IND1]

The quote captures the trajectory and sequence of risks and responses experienced and how they shape his and his children’s lives. While his father responded to food insecurity by selling the family land, he diversified first by becoming a wage labourer and then commuting to Bangalore city for work. Looking forward, he notes that these choices shape the opportunities his children have, through education. In the interview, he also alluded to how broader socio-cultural norms (caste) and assets (social networks) play a role in managing risk.

The second form of data from group discussions tended to allow for more detailed discussions on intersecting drivers of responses.Footnote 4 The gender-segregated discussions also provided insights into what men and women do and why. These narratives were typically at community scale (thus often capturing inter-household dynamics but masking intra-household dynamics). For example, in Kenya, the group discussions with pastoral communities were segregated by age and gender. Older men reported loss of different types of fodder due to erratic and low rainfall exacerbated by overgrazing. Issues of reduced fodder were compounded by ongoing, low-level ethnic conflict (between the Meru and Borana) that has limited access to local markets, forcing people to travel further. Cumulatively, these pressures have affected businesses, like miraaFootnote 5 trading, since producers, middlemen, and customers now travel longer distances to access markets. In contrast, younger men highlighted that long dry spells affected miraa prices, with grave consequences for youth reliant on it as a livelihood, while women noted that the severe drought in 2011 led to acute pasture shortage, lack of water, and widespread livestock deaths. Thus, the group interviews constructed a picture of biophysical risks constraining local livelihood systems.

Let us now examine the findings from the LHs. In Kenya, for example, a widowed woman who runs a small shop and trades fruit, vegetables, miraa, and other household products reported that ethnic conflicts meant that she had to rely on miraa stock she had. The ethnic conflict had disrupted water availability from the mountains (provided through tankers), making her rely on bottled water ($0.50 for 1 l) or buying water from the police ($0.30 for 20 l). In another LH, a 27-year-old father of four children lives and works in Isiolo, Kenya, primarily as a construction labourer. Having previously owned a small herd of sheep, he reported how the severe drought of 2011 left very little pasture, forcing him to move his sheep farther. By the time rains came, his two remaining sheep were so weak that they also died. During this period, he often resorted to buying fodder he could ill afford and his wife would collect vegetable scraps from nearby houses and the local market for the sheep. At the height of the drought, his wife’s small business of buying and selling vegetables provided their only income. Thus, the LHs alluded to biophysical risks and ethnic conflicts but added experiences of interactions between multiple risks, responses, and how these were sequenced.

A third form of data was collected through participatory exercises such as community-scale VRAs. VRA exercises were conducted as workshops, focusing on the entire social–ecological landscape, to build and strengthen relationships between stakeholders and enable marginalised voices to be heard (Morchain and Kelsey 2016). In Namibia, participatory, multi-stakeholder VRA workshops were conducted to analyse multi-scalar causes of vulnerability (Hegga et al. 2016).

The VRA workshop provided critical insights into beyond-community drivers of vulnerability and responses. For example, as expected, drought and erratic rainfall were the highest ranked risks. Other drivers of vulnerability were ‘cultural beliefs stopping new practices’ (ranked third) and institutional and informational issues (e.g. limited availability of climate forecasts and limited agricultural extension services ranked fourth and fifth, respectively) (Hegga et al. 2016). By bringing together multiple stakeholders and using a system approach, the VRA results explored structural drivers of vulnerability such as historic inequalities and long-standing lacunae in governance systems. This was more holistic than in LHs where some of these aspects were referred to, but their role in mediating responses was not always clear. The VRA also involves a step called ‘Aligning Findings with Opportunities’ which is a solution-oriented forward-looking exercise. Through this, the VRA workshops developed response strategies based on identified vulnerabilities or impacts, giving critical insights into local needs and possible priority areas.

One drawback of the VRA was that the exercises did not focus on temporality; risks and responses were listed as static options rather than dynamic entities.Footnote 6 Morchain and Kelsey (2016, p. 7) note that ‘considering the increasing complexity of landscapes and the people inhabiting them, a VRA should be repeated roughly annually or biannually, or when important changes take place. This will ensure the findings reflect the latest dynamics’ (emphasis in original). Such a reflection highlights that dynamics are important but given multiple strands of inquiry participatory exercises already have, dynamic risks and responses often get overlooked or reduced to checklist type assessments. Further, participatory risk assessment exercises require significant time and skill (van Aalst et al. 2008). However, with its focus on equity and agency, the VRA proved to forefront marginalised voices and concerns more than standard group discussions or LHs.

Intra-household dynamics and power relationships

Intra-household gender relations are critical to shaping individual vulnerability and response strategies to manage risk (Rao et al 2020; Singh 2019; Ravera et al. 2016; Rao et al. 2017). Based on relative capacities, work burdens, and agency, individuals within a household negotiate different risks differentially, either towards maintaining or enhancing overall household well-being or to consolidate their relative position (Singh and Basu 2019; Rao 2014; Ngigi et al. 2017; Rao et al. 2020).

The LHs helped unpack intra-household dynamics relatively well. By interviewing multiple people within the household, LHs captured (1) trade-offs between household members, (2) how risk perceptions were differentiated by position within the household, and (3) how response strategies were coproduced by household members as well as shaped by individual agency and voice (e.g. see Case Study 1). Insights on iterative negotiations underpinning household decisions were absent from the survey data, which typically collected data from one household member (usually the household head) and did not adequately cover power differentials within the household unit. The survey in Kenya was an exception where two members within each household, the household head and a member of the opposite gender, were interviewed, which led to a more nuanced dataset on intra-household dynamics.

Case Study 1 Intra-household trade-offs

figure a

In Ghana, mental modelling was also used to understand how households deal with food insecurity. Semi-structured interviews were done with young and old, men and women. Mental modelling interviews involved inviting respondents to talk about their experiences of food access and management without interruption or probing. The data were analysed to establish recurring themes and relationships and construct a model influence diagram to understand experiences of food security differentiated. The mental modelling data provided rich insights into why people respond the way they do (drivers of responses) with implications for understanding inter- and intra-household response behaviour. However, it did not show how experiences with food security change over time, although some dynamics were established during analysis.

The LHs drew out how food security (especially access) is shaped by multiple factors such as drought incidence, household structure, dependence on relatives, and changes over time. For example, members in one household traced how the death of the head affected food access and availability:

‘The family of my late husband took all the properties I had including the farmland. This was when I was already suffering. They took the livestock and the harvest we had including 30 pigs, 20 guinea fowls, 7 bags of beans, 6 bags of corn and 7 bags of rice. This made living conditions difficult for me and my children. We couldn’t even have two-square meals in a day’. (GHA3)

We found that mental modelling interviews collected rich data on individual roles in maintaining household food security and captured differentiated food access at inter- and intra-household levels. However, given the uninterrupted format of interviewing, key risks and responses were often left out and how access changed over time was inadequately captured. The LHs filled this gap by collecting information on a household’s specific context, relationships between household members, and their roles within or around a particular activity (e.g. securing food), over time.

Temporality

How risks, vulnerability, and peoples’ responses change over time is a key question in CCVA research and one that has been repeatedly identified but have relatively low empirical evidence (Fawcett et al. 2017; Adamson et al. 2018). In our research, the survey typically captured temporality (of risks and response outcomes) through recall questions. For example, dynamic risks were explored through perceptions of change in rainfall variability, access to fodder, or water availability. Responses were captured through before/after questions. In Namibia, Ghana, and India, the survey included questions on levels of poverty and changes in subjective well-being presently and 10 years ago. The LHs dealt with temporality—past decisions, present situations, and future aspirations—in detail. The pathways households took were easy to follow since most of the interviews captured risks as they had occurred, and actions taken in response to those risks (Case Study 2). In doing so, the LHs often highlighted how several risks can be sequenced together leading to a compounding of effects, while at other times different responses (even from different household members or external actors) can come together to strengthen household responses (these findings are reported in a forthcoming paper  by Tebboth et al.).

Case Study 2 Temporality and risk and response pathways

figure b

While the group discussions captured changes in risks and responses over time, they tended to be reported as discrete events and seldom captured pathways of responses the way the LHs do. In India, the FGDs used participatory timeline exercises to follow trajectories of biophysical, social, institutional, and political changes (methodological details in Singh et al. 2016a). The 45-year timelines (from 1970 onwards) identified specific risks and consequent responses people had undertaken. The data captured broad changes with specific events signposted, such as significant droughts or large development projects. In contrast, the LHs also alluded to these broader changes but demonstrated how these events had led to specific household/individual responses and gave insights into how reactive and proactive approaches were undertaken. Further, they allowed us to construct household response pathways (as shown in Case Study 2) where households followed certain trajectories as a result of the risks faced and responses taken.

Motivations and aspirations

Aspirations are examined in poverty and social mobility studies (Ray 2006) but seldom discussed in CCVA research (Rao et al. 2020). Understood as ‘conceptions of how human beings engage their own futures’ (Appadurai 2004: 33), aspirations are integral to livelihood decisions such as whether to migrate or not (Suckall et al. 2016; Scheibelhofer 2017), changing labour practices and social norms (Punch and Sugden 2013), and when dealing with socio-ecological transitions in rural areas (Rigg 2006).

In India and Namibia, the surveys ended with a question on aspirations, captured through what people aim to invest in and what their hopes were for the future. As a standalone block at the end of a lengthy survey, this did not work well and the data were not robust enough to be analysed. The LHs were able to weave questions on aspirations into the broader interview, as a story spanning the past, present, and future rather than a standalone question which can be unrepresentative of lived realities.

We found that aspirations typically involved wanting to educate children (education being seen as a vehicle of upward social mobility), improve existing livelihoods (e.g. invest in improving farming), and in some cases return to one’s home (e.g. several permanent migrants in India spoke of returning ‘home’, to their villages after earning in the city). These rich narratives of aspirations not only explained what people looked forward to and hence valued but also allow glimpses into what motivates people and hence can incentivise behavioural change towards adaptation. Moreover, we can see how aspirations differ for different elements of people’s lives (for children, livelihoods, safety etc.), holding critical insights into what motivates individuals to undertake different responses.

Recognising the limitations of life histories

There, of course, remain caveats to the LH approach. The narrative data LHs generate cannot be read in a vacuum; the life stories are deeply embedded in wider socio-political environments and dynamics, and analysing this requires intimate understanding of local and wider contexts. Linked to this is the need for advanced qualitative analysis skills to make sense of narrative data. In our research, we overcame this constraint by having experienced and novice qualitative researchers work collaboratively from research conceptualisation to data analysis with regular discussions on how to conduct open-ended interviews, record data, and analyse and write up the narratives (summarised in Davies et al. 2018).

There are certain drawbacks inherent to the process of LH interviews themselves. These include the costs and barriers of accessing respondents, time required to collect and analyse data, and risk of gatekeepers shaping data quality.Footnote 7 Often, LH interviews suggest a number of livelihood decisions, and interviewers must choose one or a few as examples to follow up in detail in the interview in the interest of time and respondent attention. While these can be mitigated by following good practices from ethnographic traditions (for example, immersion in the research site, triangulating data, and rapport building with respondents), they remain time-consuming processes that require significant skills.

While LHs reconfigure researcher–respondent hierarchies, they still rely on the researcher to frame the narrative and decide what parts of the story are reported, highlighting the ‘power of authorial control’ (Lewis 2008, p. 562). In our study, we overcame this to some extent by staying as close to the narrative as possible and minimising reinterpretation on the researcher’s part.

LHs are often criticised for the difficulty of generalising from individual, highly granular data. Here, it is important to revisit what LHs set out to do—they create ‘multiple and diverse ‘little narratives’ that contest ‘the stability of received knowledge’ (Gardner 2002: 28). Thus, generalisation is not a core aim of LH data. However, given that generalisation is an important concern, especially when the research is meant for policy audiences that require the big picture, some studies have shown how LH data can be used to provide an understanding of broader ‘types’ and patterns. For example, examining changing well-being of older people in Buenos Aires, Locke and Lloyd-Sherlock (2011) analysed commonalities and differences in how informants referred to particular relationships, turning points, or changes in circumstances. In their analysis, they focused on contrasting pairs of narratives (around a particular theme) in detail and followed with a brief, generalised account of how that theme appeared in the wider group. This allowed preserving ‘the integrity of individual life histories whilst enabling us to explore the variety of experiences of linked lives’ (p. 1137).

Finally, LHs can suffer from a ‘biographical inevitability paradox’Footnote 8 as they are always told from the position of the interviewee today. Thus, they can be criticised for being self-justifying narratives that attempt to explain current situations (good or bad) with the benefits of hindsight. Respondents can hide and reveal certain stories, thus showing what they want you to see (Baumeister and Newman 1994). In this sense, what people say about their lives may not be considered by positivist researchers as ‘true’ in the objective sense. We did encounter examples of interviewees manipulating stories, masking certain aspects, or giving conflictive narratives. However, we argue that these silences and manipulations can allow CCVA researchers understand what people value and how that drives their decision-making. More importantly, conflicting narratives can often allow insights into the process of sense-making of precarious lives (Granderson 2014). Such a focus uncovers some of the myths of climate change, such as the view that it is a primary driver of livelihood decisions, and highlights the plurality of views about a single event, a specific stressor, or a particular intervention.

Discussion and conclusion

As the scope of CCVA research has expanded from climate modelling and impact assessment, to more qualitative explorations of how and why some people adapt to climate change, the methodological approaches used have also expanded. In this paper, we began by reviewing the methodological landscape of CCVA and found there remain gaps in linking risk perception to decisions people make about their livelihoods and managing risk (‘Current methodological approaches to climate vulnerability and adaptation’ section). While each methodological approach has its own strengths and weaknesses (and it is not our aim to elevate one over the other), in this paper, we argue that life histories (LHs) add value to CCVA research in several ways.

First, a life history approach examines temporality and demonstrates that household livelihoods and risk management strategies are not static. They are dynamic and continually adjusted in response to external changes (of which climate change is one) and internal dynamics (changing household configurations, aspirations, asset bases etc.). LHs acknowledge and capture how people use ‘social navigation’ (Langevang 2008) to respond to and prepare for an environment and socio-institutional context that is constantly changing. In doing so, we respond to calls for ‘“time-scape” investigations into the ways individual citizens construct ideas of climate’ (Hulme et al. 2009) and offer LHs as an approach to strengthen evidence on temporality in CCVA research.

Second, by demonstrating the processes households undertake to respond, the fixation of articulating causal links between migration and climate change falls apart. Research on environmental migration has evolved from alarmist discourses that envisioned climate change driving environmental refugees to a focus on ascertaining the role of climate change among other, more structural drivers of mobility such as lack of opportunity, conflict, or poverty (Bettini and Gioli 2016; Gemenne and Blocher 2017). Despite growing acknowledgement of this non-linear, complex landscape of migration decision-making (Farbotko et al. 2018), methods to understand how environmental drivers interact with other socioeconomic factors are ‘nascent’ (Bettini and Gioli 2015:7) and rarely pluralistic (Baldwin and Fornalé 2017; Piguet et al. 2018). Against this gap, LHs appear to be a valuable approach to refocus environmental migration research away from causation alone to a more bottom-up, inclusive understanding of how people operate within webs of risks and responses.

Third, a life history approach captures the complexity and messiness of household decision-making and highlights how different people within the same household have different motivations, thus unpacking intra-household dynamics. It forefronts how certain combinations of characteristics, e.g. age, gender, and educational qualifications, can come together to mediate risk management. It can also demonstrate how focussing on households as homogenous units tends to mask trade-offs that individuals make (as shown in Case Study 1). We argue that the intra-household data that LHs collect can add to the growing field of gender and adaptation, thereby increasing the voices and views that are represented in CCVA research. At a larger, policy-facing scale, this has implications on what is seen as effective adaptation and who is included in shaping discourses on effectiveness.

Fourth, the granularity that LHs provide add depth to vulnerability and adaptation studies moving away from a ‘pressure and release model’ (Blaikie et al. 2004) to one of iterative decision-making marked by non-linearity and trial and error. In writing about human vulnerability and environmental change, McLaughlin and Dietz (2008:99–100) argue that further progress will only be achieved if researchers ‘...develop a more robust, integrated perspective on vulnerability, one capable of addressing the interrelated dynamics of social structure, human agency, and the environment’. We believe LHs can add to this agenda of building empirical evidence around the dynamic and interlinked nature of household risks and responses.

Fifth, LHs forefront the importance of perceptions (of risks and well-being) and add to the literature on how risks are perceived and go on to shape responses. By capturing ‘lived experience or embodiment of risk’ (Granderson 2014:57), LHs potentially democratise the sensemaking of risks, showcasing that there is no one ‘truth’. Capturing the subjective, contextual, and dynamic nature of risk perceptions, this differentiated understanding of risk perception can help explain where interventions can potentially be most effective and how uptake can be incentivised.

LHs can also offer insights to adaptation practice and policy which challenge conventional ‘givens’ (Table 3). While these implications can be drawn from other methodological approaches as well, LHs’ focus on temporality, intra-household dynamics, household response pathways, and embedding individual actions in wider shifts are particularly useful.

Table 3 Practice and policy implications of life history data
Full size table

At the level of policy and in common practice, certain assumptions and dominant narratives serve to shape how and what adaptation action is prioritised. Echoing Godfrey and Richardson (2004: 145), we argue that ‘oral testimony tells us as much about the symbolic categories through which reality is constructed as it does about the ‘facts’ of people’s lives’. Thus, the multiple narratives of intersecting climatic and non-climatic risks, dynamic vulnerability, and complex risk management decision-making that LHs capture serve to contest linear mono-causal understandings of vulnerability and unpack the seemingly black box of adaptation differentiation and decision-making. It also helps redress the tendency within policy discourse to focus exclusively on big n data as a way to justify and assess adaptation interventions. Such approaches marginalise issues central to vulnerability and adaptation such as intra-household well-being, gender relations, and changing aspirations (Rao et al. 2020) by reducing complexity.

Different methodologies privilege different facts and portrayals of the truth. To be truly representative and include marginal voices, CCVA research needs to expand current, dominant methodologies to capture these different understandings of risk and how they inform response behaviour. While we do not suggest privileging one methodological approach over the other, we call for recognising and capitalising on the strengths that different approaches bring to CCVA research and practice. Life histories, as we demonstrate in this paper, are one way to expand the range of existing methodological approaches employed and add immense value to who is represented and what is valued in CCVA research and practice. They allow us to challenge conventional assumptions about who is vulnerable and how strategies to deal with multiple risks follow dynamic, heterogeneous pathways. This expanded understanding of risk perception, and risk management behaviour (of which adaptation is one kind of response), we argue, has direct implications on the kinds of adaptation that gets supported, when and where interventions are made, and how they are incentivised.