Push and Pull Factors Influencing Household Resettlement on the 2021 Nyiragongo Lava Flows in Goma (DRC): A Micro-Level Survey Dataset

 

1.     Introduction

The city of Goma, in the eastern Democratic Republic of the Congo, is characterised by rapid urban growth under conditions of chronic risk, driven in part by its proximity to Mount Nyiragongo—one of the most active and hazardous volcanoes in Africa. The eruption of May 2021 generated extensive lava flows that destroyed residential areas and infrastructure, triggering large-scale displacement. However, in the months following the eruption, a notable pattern emerged: affected populations began to return to, or newly settle on, the recently solidified lava flows.

This apparent paradox—resettlement in highly hazardous and recently affected areas—raises critical questions about the drivers of settlement decisions in post-disaster contexts. Rather than being solely determined by risk awareness, such decisions are shaped by a complex interplay of push factors (constraints and pressures that drive people away from alternative locations) and pull factors (perceived advantages that attract people to specific sites, including hazardous ones).

This dataset was designed to systematically capture these dynamics at the household level with 958 participants. It provides detailed empirical evidence on the factors influencing decisions to settle or re-settle on the 2021 lava flows of Mount Nyiragongo. The data encompass socio-demographic characteristics, prior exposure to volcanic eruptions, evacuation experiences, and multidimensional impacts of displacement, alongside a rich set of variables explicitly measuring perceived push and pull factors.

Push factors documented in the dataset include constraints such as high housing costs, limited access to land, insecurity in alternative neighbourhoods, inadequate access to services, and social dislocation. These pressures can reduce the viability of safer locations and effectively compel households to seek alternatives, even in risk-prone zones.

Conversely, the dataset captures a range of pull factors associated with settlement on lava flows. These include the relative affordability and availability of land, proximity to economic opportunities and urban centres, perceived improvements in accessibility, and the possibility of rebuilding livelihoods. In addition, social and relational dimensions—such as proximity to family networks, community presence, and social integration—play a key role in shaping these decisions.

A distinctive contribution of this dataset is its integration of subjective perceptions and objective conditions. It includes Likert-scale measures of perceived ease, affordability, security, and access to services, alongside indicators of place attachment (emotional, cultural, and memory-based). This allows for a nuanced understanding of how individuals weigh risks against opportunities when making settlement decisions.

Importantly, the dataset also captures how prior disaster experience—particularly exposure to the 2002 and 2021 eruptions—influences current behaviour. It provides insights into how evacuation experiences, perceived impacts (physical, health, psychological), and risk awareness interact with structural constraints to shape post-eruption mobility and settlement patterns.

By focusing on the case of lava flow resettlement in Goma, this dataset contributes to broader debates in disaster risk reduction, urban resilience, and human geography. It highlights that resettlement in hazardous areas is not merely a failure of risk communication or governance, but often a rational response to constrained choices within complex socio-economic environments.

The dataset is particularly relevant for researchers and practitioners seeking to design more effective resettlement strategies, land-use policies, and risk reduction interventions in rapidly growing, hazard-exposed cities. It provides a robust empirical basis for understanding how trade-offs between risk, affordability, accessibility, and social belonging shape settlement decisions in post-disaster contexts.

2.      Structure of the Dataset

Each row represents a single respondent, while columns correspond to variables grouped into the following thematic categories:

a.      Socio-demographic characteristics

• Gender: Respondent’s gender (Male/Female)
• Age: Age category (coded numerically)
• Income: Household income level (ordinal scale)
• Household: Household size or composition
• Room: Number of rooms in the dwelling

b.     Experience with volcanic eruptions

• Experiencen / Experiencef: General experience of volcanic events (binary or categorical)
• Eruption: Reported eruption(s) experienced (1977, 2002, 2021)
• Specific variables for each eruption:
• Erupt1977n / Erupt1977f
• Erupt2002n / Erupt2002f
• Erupt2021n / Erupt2021f

These variables capture whether respondents experienced each eruption and, where applicable, qualitative aspects of that experience.

c.      Residence and evacuation

• Residencen / Residencef: Residential status during eruption
• Evacuationn / Evacuationf: Whether evacuation occurred
• EvacuationChoice: Type of evacuation (e.g., mandatory, self-evacuation, both)
• Perception variables:
• EvacEasiern / EvacEasierf: Perceived ease of evacuation
• EvacNecessaryn / EvacNecessaryf: Perceived necessity of evacuation

d.     Social disruption and separation

• Separationn / Separationf: Whether households were separated
• NberSeparation: Number of separations experienced

 

e.      Impacts of eruptions

• Impactedn / Impactedf: Whether respondent was impacted
• Impact dimensions:
• Physicn / Physicf: Physical impacts
• Healthn / Healthf: Health impacts
• Phsychn / Phsychf: Psychological impacts

Severity is typically measured using ordinal scales (e.g., “Not severe at all” to “Extremely severe”).

 

f.       Attachment to place

Variables prefixed with AT capture different dimensions of place attachment:

• ATemotion: Emotional attachment
• ATbeauty: Perception of landscape beauty
• ATmemory: Memory-related attachment
• ATculture: Cultural attachment
• ATstay: Desire to remain

Responses are measured on Likert scales ranging from low to high attachment.

g.      Location attractiveness and living conditions

Variables prefixed with LA assess factors influencing settlement decisions:

• LAobtain: Ease of obtaining land
• LAproximity: Proximity to services or city
• LAprice: Affordability
• LAstability: Stability of location
• LAsecurity: Perceived safety

 

h.     Community attributes

Variables prefixed with CA describe social and community dynamics:

• CAlink: Social ties
• CApresence: Presence of family/community
• CAsocial: Social life
• CAintegration: Integration into community
• CAsupport: Social support networks

i.       Access to services

Variables prefixed with AC evaluate access to essential services:

• ACservices: General services
• ACschool: Schools
• ACtransport: Transport
• ACshops: Markets/shops
• ACsecurity: Security services

j.       Living conditions and constraints

Variables prefixed with AL:

• ALconstrained: Constraints in current living conditions
• ALenough: Adequacy of housing
• ALcost: Cost-related constraints
• ALotherhome / ALhome: Availability or preference for alternative housing

k.     Preparedness and risk perception

Variables prefixed with PR:

• PRassessment: Risk assessment capacity
• PRpreparation: Preparedness level
• PRknowledge: Knowledge of volcanic risks
• PRmeasure: Awareness of mitigation measures
• PRinformation: Access to information

l.       Data Characteristics

• Type of data: Cross-sectional survey
• Measurement scales: मिश्र of categorical, ordinal (Likert scale), and binary variables
• Coding:
• Numerical codes are used for ordinal responses (e.g., 1–5 scales)

Text labels provide corresponding qualitative meaning (e.g., “Strongly disagree” to “Strongly agree”)