Assessing Refugee Preferences for SDG 2 (Zero Hunger) Solutions in Irbid Camp and Sakhra Region: Cultivated Roofs and Refrigerators as Food Banks Interventions

Abstract

Addressing hunger, particularly within impoverished communities in Jordan and globally, demands innovative, practical solutions. The research focused on refugee populations and their preferences for interventions aligned with Sustainable Development Goal (SDG) 2: Zero Hunger remains limited. This study explores the preferences of refugees in the Irbid Camp and Sakhra region, Jordan, for two potential interventions—cultivated roofs ($CRs$) and refrigerators as food banks ($RaFB$). Surveys conducted among 402 households serve to determine refugee preferences in hunger reduction, the influence of demographic attributes on these choices, and the feasibility of each proposed intervention. Chi-square tests were utilized to establish correlations between refugee intervention preferences and demographic variables, such as age, gender, education level, and family size. The results reveal a strong preference (90%) for $RaFB$ over $CRs$ (10%). While no significant demographic influence was identified on the acceptance of $CRs$, a strong correlation was discovered between the education level and the acceptance of the $RaFB$ intervention. $RaFB$ was predominantly favored due to its lower implementation costs, reduced effort, lower risk, cultural compatibility, and demonstrated success in similar contexts. Conversely, highly educated refugees were more likely to reject $RaFB$, indicating potential influences from diverse cultural perspectives or access to alternate solutions. This study provides valuable insight into the potential advantages and challenges of implementing $CRs$ and $RaFB$ projects. It further underscores the need for policymakers to consider demographic factors and cultural nuances in future intervention designs to achieve SDG 2 more effectively.

1. Introduction

Food security remains a pressing global challenge, particularly in refugee camps where access to adequate, safe, and nutritious food is often limited. The United Nations’ Sustainable Development Goal (SDG) 2, established in 2015, aimed to end hunger, achieve food security, improve nutrition, and promote sustainable agriculture by 2030. As we move past the initial 2030 target, the United Nations has advanced a new set of development goals to address the persisting and emerging challenges [1,2]. These renewed goals continue to prioritize food security but with an increased focus on climate resilience, equity in access, and the promotion of nutritious diets. Achieving these goals in vulnerable populations, such as refugees residing in camps and host communities, necessitates innovative and context-specific solutions that address these groups’ unique needs and constraints [3]. In this study, we explore two potential solutions, cultivated roofs ($CRs$) and refrigerators as food banks ($RaFB$), in the context of poverty pockets in Jordan.

Hunger and poverty pockets, like those in the Irbid Camp and Sakhra region, often lack infrastructure and resources, exacerbating their inhabitants’ already precarious food security situation. In addition to the immediate need for food assistance, there is a growing recognition that sustainable and long-term solutions must be sought to address food insecurity and promote self-reliance among refugee populations [4]. One such approach is exploring alternative food production and storage methods appropriate for the camp environment [5].

In Jordan, the economic shock resulting from the coronavirus pandemic in recent years has deepened previous trends of poverty rates, despite governmental efforts to support the social safety net, which have contributed somewhat to slowing the acceleration of poverty numbers. The poverty rate in Jordan is estimated at 24%, according to the latest study conducted by the General Statistics Department in 2021, an increase of 6% due to the repercussions of the coronavirus, while this percentage increases to 33% or more in some poverty pockets and camps [6,7].

Ten officially recognized Palestinian refugee camps in Jordan are serviced by UNRWA and are spread across six provinces [8]. The Sakhra region in Ajloun was selected because it is one of the poorest areas in Jordan, with a poverty rate between 28% and 33% in some regions. The proportion of the population under 15 exceeds 40% in the governorate, and the unemployment rate is high. It is a rural area suitable for agriculture. On the other hand, the Irbid camp was chosen because the poverty and unemployment rates exceed 60% and the camp lacks any developmental projects. The camp is home to more than 29,800 citizens living in a small area of 244,000 m² that cannot be expanded [9,10].

All the member states of the United Nations adopted the 17 Sustainable Development Goals (SDGs) in 2015, serving as a global call to action to eradicate poverty by 2030. The second SDG aims to eliminate hunger and malnutrition and achieve sustainable food production by 2030. This goal is premised on the idea that everyone should have access to sufficient nutritious food, which necessitates promoting sustainable agriculture on a broad scale, which includes supporting small-scale farmers and achieving equality in access to land, technology, and markets. It also requires international cooperation to ensure investment in infrastructure and technology to improve agricultural productivity, increase investment, and properly operate food markets [2].

Global evidence indicates that the number of hungry people worldwide is increasing, reaching 828 million in 2021, an additional 46 million compared to the previous year and an additional 150 million compared to 2019, according to the World Food Security and Nutrition Report for 2022 issued by the Food and Agriculture Organization of the United Nations. Hunger has risen over the past three years, returning to levels not seen in a full decade. This setback sends a clear warning that more efforts are needed urgently to achieve the Sustainable Development Goal of eradicating hunger by 2030 [11,12,13].

This research presents the findings of a study aligning with the targets outlined in the United Nations’ Sustainable Development Goal (SDG) 2: Zero Hunger. We aim to end all forms of hunger and malnutrition by 2030, ensuring all people–especially children and the more vulnerable–have sufficient and nutritious food all year round. To do this, promoting sustainable agricultural practices, including supporting small-scale farmers and allowing equal access to land, technology, and markets, is vital, and international cooperation is necessary to ensure investment in infrastructure and technology to boost agricultural productivity.

In the current study, we investigated refugees’ preferences and the demographic influence on refugees in the Irbid Camp and Sakhra region for two proposed interventions that align with these targets: cultivated roofs ($CRs$) and refrigerators as food banks ($RaFB$). The $CRs$ intervention includes cultivating food crops on the roofs of houses inside the district, a sustainable agricultural practice providing year-round access to nutritious food. Simultaneously, the $RaFB$ intervention aligns with food preservation, storage, and access, positioning community refrigerators to preserve and store food. In addition to assessing the overall preference for each intervention, this research also surveyed the prospective advantages and challenges of employing these solutions in pockets of poverty across Jordan. Age, gender, education level, and family size were analyzed to determine their influence on the refugees’ preferences for each project. By understanding the preferences and needs of refugees concerning food security interventions, this study aims to contribute to the development of contextually appropriate and sustainable solutions for achieving zero hunger in refugee camps and vulnerable areas. The insight gained from this research can further inform policymakers and aid organizations in designing and implementing effective food security strategies in these settings, thus aiding in the advancement of SDG 2.

The study’s problem lies in the poverty pockets in Jordan, especially in the Irbid Camp and Sakhra region, which suffer from severe poverty and hunger, and several Palestinian refugee camps whose members suffer from poverty. Therefore, the importance of the research is to shed light on the issue of hunger in refugee camps and poverty pockets and try to find pioneering projects that contribute to raising the economies of families and solving the problem of hunger and poverty.

The research questions in this study were developed in response to the increasing global and local challenge of food security, particularly within refugee populations and areas of extreme poverty. Previous studies have proposed various strategies for alleviating hunger [14,15,16], yet none have specifically focused on the context of Jordan’s refugee camps and poverty pockets and the unique interventions of cultivated roofs ($CRs$) and refrigerators as food banks ($RaFB$). The current research fills this gap, contributing to the literature on development economics by examining the feasibility and acceptability of these two interventions in these specific settings. Thus, the current research aims to answer the following research questions:

RQ1: 

What is the overall preference among refugees for cultivated roofs, refrigerators as food banks, or their ideas for achieving zero hunger in the camps and vulnerable areas (i.e., Irbid Camp and Sakhra region)?

RQ2: 

How does the demographic profile of the refugees (e.g., age, gender, education level, and family size) influence their preferences for cultivated roofs or refrigerators as food banks for achieving zero hunger?

The remainder of this paper is structured as follows: Section 2 provides an in-depth review of the existing literature on the interventions of $CRs$ and $RaFB$ and food security in refugee and impoverished settings. Section 3 details the methods used in our study, including the data collection and analysis. Section 4 presents our findings, while Section 5 discusses these findings in the context of the existing literature. Finally, Section 6 concludes with a summary of our research and its implications for future policies and practices.

2. Literature Review

The literature review section explores food security’s broad and complex dynamics in refugee settings, focusing on implementing Sustainable Development Goal 2 (Zero Hunger) solutions in refugee communities. A deep exploration of various scholarly sources was undertaken to understand the intersection of refugee experiences, food insecurity, and innovative interventions to combat hunger. Furthermore, the review illuminates the context of the Irbid Camp and the Sakhra Region, shedding light on the unique challenges these regions face and the opportunities for sustainable interventions. In examining previous studies, we pay close attention to the emerging themes, potential gaps, and overall trajectory of research on food security among refugees. This literature review lays the groundwork for our research question: “How can we effectively implement and integrate cultivated roofs and refrigerator food banks within the refugee context, specifically in the Irbid Camp and Sakhra Region, and what are the preferences of refugees for these interventions?” Through the lens of the literature, we aim to frame this question within its broader thematic, regional, and theoretical context.

Researchers argue that labor market integration for refugees, such as the Syrian refugees in Jordan, necessitates the alignment of various perspectives. These include legal regulations of the host state, refugee accessibility to the labor market, host community responses, and international aid. Their research emphasizes the need for harmonized efforts to promote refugees’ unrestricted participation in various sectors [17]. Meanwhile, the child marriage trends among Jordan’s specific communities, particularly refugees, are analyzed. Based on interviews with 64 Jordanian and Syrian adolescents, their research offers insight into perceptions surrounding child marriage amid the Syrian crisis. Through the unique methodology of narrative discourse and visual timeline construction, they capture refugee and host populations’ lived experiences and reflections [18].

While these two studies provide critical insight into refugee circumstances, it is evident that more literature must be referenced to survey the field and recent research trends adequately. Hence, future revisions of this section should aim to succinctly articulate the points from existing references while incorporating more diverse studies. This balance will provide a more comprehensive review of the relevant literature, enhancing the research’s grounding in the current scholarly discourse. The precarious employment conditions facing young individuals in Egypt, Jordan, and Tunisia were highlighted in a comprehensive examination of youth labor market vulnerabilities. This study found that unstable initial jobs frequently led to continued low-quality employment, with family wealth, parental education, and the father’s occupation being significant determinants of labor market outcomes [19].

Researchers have also focused on psychosocial support as a tool for refugee family empowerment in Jordan. Their research involved a sample of 32 refugees, split into an experimental group, which received psychosocial support, and a control group. Using the Family Empowerment Scale, they collected pre- and post-test data and a follow-up test for the experimental group, suggesting the potential of such programs to enhance the empowerment of refugee families [20].

These studies contribute valuable insight into distinct aspects of the refugee experience, namely labor market vulnerabilities and the potential benefits of psychosocial support. This further reinforces the need for a multi-faceted refugee integration and empowerment approach.

Researchers also discovered positive attitudes toward this approach in exploring blended learning (BL) among Syrian refugees. The authors surveyed 93 refugees and found that age, rather than gender, affected the participants’ perceptions, with older individuals demonstrating a greater appreciation for BL. The identified challenges offer a roadmap for enhancing BL methodologies tailored to the needs of refugees [21]. Additionally, others studied the impact of the Syrian refugee influx on land use/land cover (LULC) changes in the Irbid district of northwestern Jordan. Leveraging Landsat imagery and the Google Earth Engine platform for analyses, they reported increased urban and agricultural land use during the Syrian crisis, attributed to the refugees’ housing needs and reliance on agriculture for livelihood [22].

These findings underscore the varied impact of refugee influxes on host societies, ranging from educational systems to land use, highlighting the need for comprehensive and multi-faceted approaches to addressing these challenges.

The social integration of elderly refugees presents a unique set of challenges. Elderly refugees, as a vulnerable group within refugee communities, often face amplified difficulties due to the convergence of the struggles associated with aging and migration. This study focuses on their social integration in light of the increasing population of older adults and refugees. The key issues faced by elderly refugees include health and language barriers, social isolation, and bureaucratic complexities. Furthermore, female refugees face additional educational, language proficiency, income, and employment disadvantages. The study identifies language barriers, poverty, and foreign surroundings as major obstacles to social integration for elderly refugees. However, a shared religion, host country characteristics, and established social networks could facilitate their integration. As such, it is essential to establish specific services and social work practices that mitigate these challenges and capitalize on facilitating factors to promote effective social integration for elderly refugees [23].

In addition, a systematic review and meta-analysis approach was employed to examine interventions related to sustainable development goals (SDGs) in refugee camps. The systematic review methodology allows for a comprehensive identification, assessment, and synthesis of all the relevant studies on a particular topic. The researchers developed an extensive search strategy to identify all the peer-reviewed articles that present interventions for SDGs in refugee camps. They screened the titles and, when necessary, the abstracts of 1108 publications, with 72 deemed to have relevant evidence that were subsequently reviewed in detail. The data extracted from these studies were then pooled using a meta-analysis, a statistical procedure for combining data from multiple studies, to provide summary estimates of the effectiveness of the existing procedures [24].

The researchers also comprehensively evaluated interventions in refugee camps to improve the refugees’ quality of life and ameliorate their conditions. Previous studies have discussed these interventions, yet there has been no formal systematic review and meta-analysis assessing the relative effectiveness of these strategies in alignment with sustainability and the 2030 agenda. This study implemented an exhaustive search strategy to identify peer-reviewed articles that discuss interventions related to sustainable development goals (SDGs) within a refugee camp context. Out of 1108 publications screened for relevance, 72 studies containing pertinent evidence were analyzed in detail. The data from these studies were subsequently compiled through a meta-analysis to provide summary estimates of the effectiveness of the current methods. The study determined that the health and education sectors were the most frequently addressed SDGs. The findings and recommendations from the included studies were classified into seven sectors: planning and development, shelters, health and well-being, education, water and sanitation, energy, and work and economic growth [25].

A critical view of global policymaking concerning refugee populations was presented, arguing for a shift in discourse from vulnerability to empowerment. The author uses the case of Canada’s interaction with the global refugee regime to argue for a more active role for refugee women in policymaking, suggesting a feminist geopolitical framework that prioritizes their individual experiences [26]. On the other hand, other researchers delve into the complexities of responding to a large-scale, prolonged urban displacement crisis using Jordan as a case study. They dissect the dynamics of inclusion and exclusion in displacement responses, exploring the factors contributing to exclusion among the displaced. Their findings point to the need for a more inclusive approach to humanitarian interventions during urban displacement crises. These studies call for more nuanced, inclusive refugee policymaking and intervention design strategies [27].

Recent studies explored the unique circumstances of refugee entrepreneurs as differentiated from other categories of immigrants, primarily due to the distinct situations they face compared to non-forced immigrants. Critical differences exist between forcibly displaced individuals and other migrants that can influence their economic choices, including those related to entrepreneurial ventures. Socio-economic heterogeneity is demonstrated significantly through individual refugee characteristics, such as age, gender, and education levels. It was noted that the existing research has not dedicated sufficient attention to gender-based studies or the experiences of refugee women entrepreneurs, particularly in entrepreneurship. Consequently, this investigation examines the adversities and obstacles encountered by refugee women entrepreneurs and identifies the present strengths and opportunities that could bolster their integration into Jordan’s host economy [28].

Several recent studies have explored the influx of refugees and their impact on the environment and energy consumption. A study examining the period between 1996 and 2019 in countries such as Bangladesh, Ethiopia, Jordan, Lebanon, Pakistan, Sudan, and Uganda through the panel quantile autoregressive distributed lag (PQARDL) and causality methods found evidence of a long-term relationship between the sustainable environment, refugee population, governance, economic growth, energy consumption, and other variables, such as HDI, trade deficit, and financial development. The study further indicated unidirectional causality from political and economic governance to greenhouse gas (GHG) emissions and deforestation, as well as from refugees to GHG emissions and deforestation [29]. A similar inquiry using the Fourier bootstrapping ARDL (FBARDL) and Granger causality with the Fourier method, in the context of Ethiopia, Sudan, and Uganda from 1985–2020, found evidence of unidirectional causality running from the refugee population and traditional energy consumption (firewood, charcoal, and coal) to deforestation and environmental pollution. Both studies underline the necessity of considering refugee populations when developing long-term environmental and energy policies, highlighting the urgency of these issues in sustainable development [30].

The literature review has underscored the intersection of food security, refugee experiences, and the potential for innovative interventions, such as cultivated roofs and refrigerators as food banks. This body of work highlights the significance of SDG 2 (Zero Hunger) and its importance to the well-being of refugee populations, particularly in areas such as the Irbid Camp and Sakhra Region. However, it also reveals gaps in the existing research, particularly concerning refugees’ preferences and perspectives on these innovative food security interventions. As such, our study aims to bridge these gaps and contribute to a more nuanced understanding of how such interventions can be implemented in refugee settings, with direct input from the refugees themselves. By considering the refugees’ preferences and ideas, we can develop more sustainable, effective, and context-specific strategies to combat hunger and ensure food security in these vulnerable communities.

3. Methodology

This study employs a mixed-methods approach, facilitating a comprehensive exploration of the social, economic, educational, health, and housing situations of 402 low-income families in the Irbid camp and Sakhra region. Qualitative and quantitative data are crucial in understanding these families’ dietary and consumption patterns and their attitudes toward the proposed entrepreneurial projects to reduce hunger and improve their economic status.

A quantitative data analysis was initiated with descriptive statistics, which were used to summarize and comprehend the data. The frequency distributions, percentages, and standard deviations were calculated to understand the data’s variability around the mean. Various indicators were explored, including income level, educational attainment, and dietary patterns.

By enhancing the quantitative analysis, probit and logistic regression analyses were introduced. These regression models enable the examination and modeling of potential causal relationships between various independent and dependent variables. For instance, the impact of variables such as income level and education on the likelihood of families’ adoption of the proposed entrepreneurial projects can be assessed. Binary outcomes are used to model these relationships, with the probit and logistic regression models offering complementary insight. The qualitative data, often collected through field surveys and questionnaires, were converted into a quantitative form using coding techniques. This conversion allows for the quantitative exploration of attitudes toward entrepreneurial projects, adding depth and detail to the statistical analysis.

The statistical package for social sciences (SPSS) was employed for data processing and analysis, and geographic information systems (ARCGIS 10.4) software was utilized for the spatial analyses. This software facilitates the exploration and visualization of geographical patterns and trends in the data. The extended methodological approach adopted in this study ensures a thorough and robust analysis of the data collected, adding more value to the findings and conclusions. This methodological integration provides a comprehensive view of the factors influencing these families’ lives, informing the design of more effective interventions to support them. Figure 1 visually represents these processes, depicting the study’s updated methodological framework and sequential procedures.

4. Primary Data Statistics

This section outlines the demographic features of the study participants, including age, gender, the highest level of education, family size, and origin, in addition to their preferences for cultivated roofs ($CRs$) or refrigerators as food banks ($RaFB$). Such statistics offer a basis for interpreting the constituents of the targeted population in the study, providing perceptions into how these demographic attributes can be connected with their preferences for various interventions to achieve SDG 2 (Zero Hunger). More integrated, practical, and tailored strategies and interferences can be reached by robustly analyzing these factors, thus paving the road toward contemplating the distinctive needs and inclinations of the targeted groups, eventually improving the efficiency and accomplishment of the projected solutions. Figure 2 represents the demographic features of the surveyed sample, while Figure 3 represents the project preference.

The data were collected over one month (December 2022) utilizing a structured questionnaire. Eighteen trained field volunteers, fluent in Arabic, the respondents’ native language, conducted the data collection exercise. To ensure the collection of accurate and reliable data, the respondents were briefed about the purpose of the study, the confidentiality of their responses, and how the data would be used. The questionnaires were administered face-to-face in the respondents’ homes, where the respondents were encouraged to answer honestly and without influence. The relevant authorities ethically approved the data collection process, adhering to all the established guidelines and protocols. This additional information will be included in the manuscript’s updated version to clarify the data collection process and timeline. Survey data were collected through surveys administered to 402 Irbid Camp and Sakhra region households. The data included the demographic characteristics, such as age, gender, education level, family size, and the respondents’ preferences for the $CRs$ and $RaFB$ interventions. This comprehensive dataset spans various demographic profiles and perspectives, providing a rich basis for our analysis.

5. Results and Discussion

The study aimed to determine the overall preference among refugees for cultivated roofs, refrigerators as food banks, or their ideas for achieving zero hunger in the camps and vulnerable areas, specifically in the Irbid Camp and Sakhra region. The results show that 10% (26) of the refugees preferred cultivated roofs, while a significant majority, 90% (233), preferred refrigerators as food banks, with some refugees suggesting alternative ideas, such as opening a grocery store, a sewing lab for teaching and operation, a supermarket project, breeding, and rental of wedding and holiday supplies.

The lower preference for cultivated roofs (10%) can be attributed to various factors, including the unsuitability of the camp buildings’ roofs for cultivation, restrictions on roof usage in rented houses, lack of available land inside the camp, insufficient knowledge of roof cultivation among the refugees, marketing challenges for agricultural products, the expense of materials and equipment required for roof cultivation, and difficulties in storing agricultural products due to rapid deterioration.

On the other hand, most refugees (90%) preferred refrigerators as food banks due to this solution’s convenience, low risk, and cost-effectiveness, and the fact that it does not require specific skills and aligns with the community culture. Additionally, refrigerators have been successfully implemented in other locations worldwide. The key factors contributing to the preference for refrigerators as food banks include convenience and accessibility, food preservation capabilities, security, storage capacity, fostering a sense of independence among the refugees, and the potential for sustainability through renewable energy sources, such as solar power.

5.1. Detailed Descriptive Analysis

To determine the project preference among the refugees, an analysis was conducted to assess the receptiveness of the refugees towards the two proposed interventions; the results are presented in

Table 1. Distribution of participants’ acceptance of proposed interventions.

	Frequency	Percent
Cultivated Roofs ($CRs$)	26	10%
Refrigerators as Food Banks ($RaFB$)	233	90%
Total	259	100%

. Based on these findings, it can be concluded that refrigerators as food banks ($RaFB$) are the predominant preference among refugees for achieving zero hunger in the studied hunger and poverty regions. It is worth mentioning that there were some ideas from the refugees for achieving zero hunger in addition to the proposed ones (e.g., opening a grocery store, a sewing lab for teaching and operation, supermarket project, breeding, and rental of wedding and holiday supplies). Such ideas pave the road toward future research for achieving zero hunger in poverty pockets across Jordan.

The influence of the demographic factors, such as age, gender, education level, and family size, on refugee preferences for cultivated roofs, refrigerators as food banks, or their ideas for achieving zero hunger was investigated using various statistical tests. For the age attribute, no significant relationship was found between age and refugee acceptance of the cultivated roofs project nor the acceptance of refrigerators as food banks project. Such conclusions were extracted after testing the hypothesis in

Table 2. Hypothesis testing results assess the significance of the age attribute concerning the acceptance of the CRs and RaFB solutions.

Cultivated Roofs (CRs)
H0:	No significance exists between age and refugee acceptance of the cultivated roofs project.
H1:	There is a significance between age and refugee acceptance of the cultivated roofs project.
	Number	Mean of age	Std. deviation	95% Confidence Interval for Mean	Minimum	Maximum
Accept	25	48	11.55	(43.4–52.9)	18	72
Reject	357	47	12.04	(45.4–47.9)	19	90
Cultivated Roofs (CRs)
Sum of Squares		Df.	Mean Square	F	Significant	p-value
50.5		1	50.5	0.350	0.554	0.277
Refrigerators as food banks (RaFB)
H0:	No significance exists between age and refugee acceptance of refrigerators as food banks project.
H1:	There is a significance between age and refugee acceptance of refrigerators as food banks project.
	Number	Mean of age	Std. deviation	95% Confidence Interval for Mean	Minimum	Maximum
Accept	226	47.7	11.80	(46.1–49.2)	18	85
Reject	156	45.5	12.21	(43.5–47.4)	19	90
Refrigerators as food banks (RaFB)
Sum of Squares		Df.	Mean Square	F	Significant	p-value
453.182		1	453.182	3.164	0.076	0.061

for the cultivated roofs ($CRs$) and the refrigerators as food banks ($RaFB$). In both cases, the $p-\mathrm{v}\mathrm{a}\mathrm{l}\mathrm{u}\mathrm{e}$ was greater than the level of significance $\alpha =0.05$, indicating the hull hypothesis ($H_0$) failed to be rejected, and there is no significant relationship between age and refugee acceptance of the cultivated roofs ($CRs$) and refrigerators as food banks ($RaFB$) projects. Thus, the lack of preference for the cultivated roofs ($CRs$) and refrigerators as food banks ($RaFB$) solutions was unrelated to the age attribute, suggesting that other variables may influence the two preferences.

For the gender and the family size attributes, no significant relationship was found between the gender and refugee acceptance of the cultivated roofs project, nor the acceptance of refrigerators as food banks project. Such conclusions were extracted after testing the cultivated roofs ($CRs$) and the refrigerators as food banks ($RaFB$) hypothesis, as shown in

Table 3. Hypothesis test results assess the significance of the gender attribute concerning the acceptance of the CRs and RaFB solutions.

CRs
H0:	No significance exists between gender and refugee acceptance of the cultivated roofs project.
H1:	There is a significance between gender and refugee acceptance of the cultivated roofs project.
			Refugee Acceptance		Total
			Accept	Reject
Gender		Male	21	292	313
		Female	5	84	89
Total			26	376	402
Chi-Square
		p-value	Sig.	Df.	value
Pearson Chi-square		0.356	0.712	1	0.136
Number of valid cases		402
RaFB
H0:	No significance exists between gender and refugee acceptance of refrigerators as food banks project.
H1:	There is a significance between gender and refugee acceptance of refrigerators as food banks project.
			Refugee Acceptance		Total
			Accept	Reject
Gender		Male	183	130	313
		Female	50	39	89
Total			233	169	402
Chi-Square
		p-value	Sig.	Df.	value
Pearson Chi-square		0.35	0.700	1	0.149
Number of valid cases		402

and

Table 4. Hypothesis test results assess the significance of the family size attribute concerning the acceptance of the CRs and RaFB solutions.

CRs
H0:	No significance exists between family size and refugee acceptance of the cultivated roofs project.
H1:	There is a significance between family size and refugee acceptance of the cultivated roofs project.
	Number	Mean of family size	Std. deviation	95% Confidence Interval for Mean	Minimum	Maximum
Accept	26	5.15	2.167	(4.28–6.03)	1	8
Reject	376	4.56	1.942	(4.37–4.76)	1	8
CRs
Sum of Squares		Df.	Mean Square	F	Significant	p-value
8.466		1	8.466	2.211	0.138	0.069
RaFB
H0:	No significance exists between family size and refugee acceptance of refrigerators as food banks project.
H1:	There is a significance between family size and refugee acceptance of refrigerators as food banks project.
	Number	Mean of family size	Std. deviation	95% Confidence Interval for Mean	Minimum	Maximum
Accept	233	4.61	1.945	(4.36–4.86)	1	8
Reject	169	4.59	1.986	(4.29–4.89)	1	8
RaFB
Sum of Squares		Df.	Mean Square	F	Significant	p-value
0.031		1	0.031	0.008	0.929	0.4645

. In both cases, the p-value was greater than the level of significance $\alpha =0.05$, indicating the hull hypothesis ($H_0$) failed to be rejected, and no significant relationship between gender or family size and refugee acceptance of the cultivated roofs ($CRs$) and refrigerators as food banks ($RaFB$) projects exists. Thus, the lack of preference for the cultivated roofs ($CRs$) and refrigerators as food banks ($RaFB$) solutions were unrelated to the gender or family size attribute, suggesting that other variables may influence the two preferences.

For the level of education attribute, no significant relationship was found between the level of education and refugee acceptance of the cultivated roofs project. Such conclusions were extracted after testing the hypothesis for the CRs, where the p-value was greater than the level of significance $\alpha =0.05$, indicating the hull hypothesis ($H_0$) failed to be rejected, and no significant relationship exists between the level of education and refugee acceptance of the CRs project. However, a significant relationship was found between the level of education and refugee acceptance of refrigerators as food bank projects. Such conclusions were extracted after testing the hypothesis for the RaFB, where the $p-\mathrm{v}\mathrm{a}\mathrm{l}\mathrm{u}\mathrm{e}=0.0395$ was lower than the level of significance $\alpha =0.05$, indicating the hull hypothesis was rejected, and there is a significant relationship between the level of education and refugee acceptance of the RaFB project. It was observed that refugees with higher educational levels were more likely to reject the solution to zero hunger by installing refrigerators as food banks. This might explain why the refugees with higher educational levels were more likely to reject the solution to achieving zero hunger by installing refrigerators as food banks. Refugees with higher educational levels may have different cultural beliefs, values, or expectations than those with lower educational levels. They may also have different perceptions of the effectiveness or practicality of the proposed solution. Alternatively, it could be due to socio-economic status, prior experiences with similar interventions, or access to alternative solutions. Further research would be necessary to better understand the factors influencing the acceptance or rejection of the proposed solution by refugees with varying educational levels.

Table 5. Hypothesis test results assess the significance of the level of education attribute concerning the acceptance of the $CRs$ and $RaFB$ solutions.

			Refugee Acceptance for the CRs Project			Refugee Acceptance for the RaFB Project
			Accept	Reject	Total	Accept	Reject		Total
Level of Education		Illiterate	0	33	33	16	17		33
		Can read and write	1	29	30	16	14		30
		Elementary	0	23	23	17	6		23
		Preparatory	5	56	61	38	23		61
		Basic	4	115	119	69	50		119
		Vocational education	0	1	1	0	1		1
		Secondary	13	91	104	56	48		104
		Intermediate diploma	2	16	18	15	3		18
		B.A.	1	9	10	6	4		10
		High school diploma	0	2	2	0	2		2
		M.A.	0	1	1	0	1		1
Total			26	376	402	233	169		402
CRs
H0:	No significance exists between the level of education and refugee acceptance of the cultivated roofs project.
H1:	There is a significance between the level of education and refugee acceptance of the cultivated roofs project.
Chi-Square
			Value	Df.		Sig.		p-value
Pearson Chi-square			13.94	10		0.176		0.088
Number of Valid Cases			402
RaFB
H0:	No significance exists between the level of education and refugee acceptance of refrigerators as food banks project.
H1:	There is a significance between the level of education and refugee acceptance of refrigerators as food banks project.
Chi-Square
			Value	Df.		Sig.		p-value
Pearson Chi-square			15.362	10		0.119		0.0395
Number of valid cases			402

represents the hypothesis testing results for the level of education attribute with the two proposed projects.

The results obtained from the survey shed light on the potential benefits of implementing an $RaFB$ project, including (1) reduced food waste because refrigerators can help to preserve perishable food items, minimizing waste and ensuring that available food resources are used effectively; (2) improved food security by providing a secure location for food storage, as refrigerators can reduce the risk of theft or spoilage, enhancing overall food security in the poverty pocket regions; (3) enhanced nutrition, as access to refrigerated storage can encourage refugees to consume a more diverse range of fresh and nutritious food items, contributing to better overall health and well-being; (4) promoting self-reliance because refrigerators can empower refugees to manage their own food resources, fostering a sense of autonomy and independence; (5) scalability, as refrigerators can be easily scaled up or down depending on the needs of the refugee population, making them a flexible solution to food storage challenges; and (6) achieving environmental sustainability via the utilization of renewable energy sources, such solar power, refrigerators can provide a sustainable and environmentally friendly option for food storage in refugee camps.

In addition, the obtained results can be considered a solid foundation for determining the potential challenges of implementing the $RaFB$ project, including (1) limited access to electricity, as many refugee camps face challenges in accessing reliable electricity sources, which could hinder the effective use of refrigerators as food banks; (2) cost and maintenance, as the initial cost of purchasing refrigerators and ongoing maintenance expenses can be significant, potentially limiting the feasibility of implementing this solution in some contexts; (3) cultural acceptance, where the acceptability of using refrigerators as food banks may vary across different cultural contexts, potentially impacting the success of the intervention; (4) logistics and transportation, as the transportation and installation of refrigerators in refugee camps may pose logistical challenges, particularly in remote or difficult-to-reach locations; (5) coordination and management, as the effective operation of refrigerators as food banks may require robust coordination and management systems to ensure equitable access to food resources and prevent misuse or abuse of the intervention; and (6) security concerns, as ensuring the security of the refrigerators and their contents may require additional resources and planning, particularly in contexts where theft or vandalism is a concern.

5.2. Probit and Logistic Regression

In this sub-section, we investigate deeper into the determinants of intervention preferences by employing two widely used techniques in the realm of econometrics: probit and logistic regression. Despite their similar purposes, these two methods offer unique insights and advantages, allowing us to analyze the variables influencing refugee households’ intervention preferences comprehensively. Probit and logistic regression are particularly useful for examining binary outcomes, as with our interventions, where the response of interest is the household’s preference for an intervention, expressed as a binary choice. Through these methods, we aim to quantify the relationship between the predictor variables, including household demographics, characteristics, and amenities, and the probability of a household opting for a given intervention. This approach will deepen our understanding of the factors influencing the refugees’ preferences, thereby guiding more effective intervention strategies.

The subsequent tables (i.e.,

Table 6. Probit regression results for the refrigerators as food banks ($RaFB$) intervention.

Variables	Coefficients	s.e.	Wald	$\mathit{p}\text{-}\mathbf{V}\mathbf{a}\mathbf{l}\mathbf{u}\mathbf{e}$	Lower	Upper
Location (Irbid Camp or Sakhra)	−0.012	0.224	0.003	0.957	−0.451	0.427
Nationality	−0.099	0.831	0.014	0.905	−1.729	1.531
$Log(House Area)$	−1.590	1.267	1.577	0.209	−4.073	0.892
House material (Concrete or Blocks)	−0.0717	0.1872	0.1467	0.7017	−0.4387	0.2953
Owned_House	4.486	11.226	0.160	0.689	−17.516	26.489
$Log(Monthly Rental)$	−2.214	7.326	0.091	0.762	−16.573	12.144
Refrigerator_Available	−0.101	0.212	0.229	0.633	−0.517	0.315
Cooker_Available	0.098	0.290	0.114	0.736	−0.470	0.666
Microwave_Available	−0.381	0.263	2.106	0.147	−0.896	0.134
Gender	0.209	0.306	0.466	0.495	−0.391	0.808
Age (Years)	−0.008	0.072	0.011	0.916	−0.149	0.134
Age2	0.000	0.001	0.000	1.000	−0.002	0.002
Family Size	−0.033	0.064	0.263	0.608	−0.159	0.093
Education	−0.931	0.121	58.991	2.548 × 10−7	−1.169	−0.694
Roof Suitability for Cultivation	0.673	0.436	2.387	0.122	−0.181	1.527
$Log(Roof Area)$	0.142	0.472	0.090	0.764	−0.783	1.066

,

Table 7. Logistic regression results for the refrigerators as food banks ($RaFB$) intervention.

Variables	Coefficients	s.e.	Wald	$\mathit{p}\text{-}\mathbf{V}\mathbf{a}\mathbf{l}\mathbf{u}\mathbf{e}$	$\mathit{E}\mathit{x}\mathit{p}\left(\mathit{b}\right)$	Lower	Upper
Location (Irbid Camp or Sakhra)	−0.080	0.395	0.041	0.839	0.9227	0.923	0.425
Nationality	0.008	1.378	0.000	0.996	1.0076	1.008	0.068
$Log(House Area)$	−0.9064	1.0264	0.7797	0.3772	0.4040	0.0540	3.0205
House material (Concrete or Blocks)	−0.1645	0.3349	0.2413	0.6233	0.8483	0.4401	1.6353
Owned_House	9.426	20.900	0.203	0.652	1.24 × 104	1.24 × 104	7.64 × 1021
$Log(Monthly Rental)$	−4.881	13.566	0.129	0.719	0.0076	0.008	2.68 × 109
Refrigerator_Available	−0.182	0.368	0.245	0.621	0.834	0.405	1.714
Cooker_Available	0.133	0.512	0.067	0.795	1.1422	−1.142	−1.419
Microwave_Available	−0.588	0.444	1.754	0.185	0.5557	0.556	0.233
Gender	0.297	0.534	0.308	0.579	1.3454	1.345	0.472
Age (Years)	−0.023	0.129	0.033	0.855	0.9768	0.977	0.759
Age2	0.000	0.001	0.008	0.930	1.0001	1.000	0.997
Family Size	−0.053	0.111	0.226	0.634	0.9484	0.948	0.763
Education	−1.730	0.246	49.582	1.902 × 10−12	0.1773	0.177	0.110
Roof Suitability for Cultivation	1.173	0.753	2.424	0.119	3.2315	3.232	0.738
$Log(Roof Area)$	0.246	0.793	0.096	0.757	1.2788	1.279	0.270

,

Table 8. Probit regression results for the cultivated roofs ($CRs$) intervention.

Variables	Coefficients	s.e.	Wald	$\mathit{p}\text{-}\mathbf{V}\mathbf{a}\mathbf{l}\mathbf{u}\mathbf{e}$	Lower	Upper
Location (Irbid Camp or Sakhra)	0.2993	0.1590	3.5443	0.0598	−0.0123	0.6108
Nationality	5.8093	4264.1691	1.859 × 10−6	0.9989	−8351.8	8363.4271
$Log(House Area)$	−0.5362	0.5861	0.8368	0.3603	−1.6849	0.6126
House material (Concrete or Blocks)	−0.649	0.263	6.093	0.014	−1.165	−0.134
Owned_House	0.4153	6.0939	0.0046	0.9457	−11.5286	12.3592
$Log(Monthly Rental)$	−0.0958	3.9784	0.0006	0.9808	−7.8933	7.7017
Refrigerator_Available	0.2250	0.1548	2.1109	0.1463	−0.0785	0.5284
Cooker_Available	0.1895	0.2396	0.6258	0.4289	−0.2801	0.6591
Microwave_Available	0.2819	0.1780	2.5092	0.1132	−0.0669	0.6308
Gender	−0.3224	0.1909	2.8512	0.0913	−0.6967	0.0518
Age (Years)	−0.0766	0.0360	4.5405	0.0331	−0.1471	−0.0061
Age2	0.0007	0.0004	3.1896	0.0741	−0.0001	0.0014
Family Size	−0.0242	0.0466	0.2686	0.6043	−0.1156	0.0672
Education	−0.0178	0.0424	0.1775	0.6736	−0.1009	0.0652
Roof Suitability for Cultivation	1.8724	0.3794	24.3606	7.99 × 10−7	1.1288	2.6159
$Log(Roof Area)$	0.0658	0.4112	0.0256	0.8729	−0.7402	0.8717

and

Table 9. Logistic regression results for the cultivated roofs ($CRs$) intervention.

Variables	Coefficients	s.e.	Wald	$\mathit{p}\text{-}\mathbf{V}\mathbf{a}\mathbf{l}\mathbf{u}\mathbf{e}$	$\mathit{E}\mathit{x}\mathit{p}\left(\mathit{b}\right)$	Lower	Upper
Location (Irbid Camp or Sakhra)	0.5436	0.2775	3.8359	0.0502	1.7222	0.9996	2.9670
Nationality	19.8449	9705.4	4.18 × 10−6	0.9984	4.15 × 108	0.000	-
$Log(House Area)$	−2.783	2.097	1.761	0.185	0.0618	0.062	0.001
House material (Concrete or Blocks)	−1.046	0.448	5.435	0.020	0.3515	0.351	0.146
Owned_House	0.9793	9.9623	0.0097	0.9217	2.6625	0.0000	8.04 × 108
$Log(Monthly Rental)$	−0.3566	6.4961	0.0030	0.9562	0.7001	0.0000	2.37 × 105
Refrigerator_Available	0.3629	0.2708	1.7959	0.1802	1.4374	0.8455	2.4438
Cooker_Available	0.3703	0.4301	0.7414	0.3892	1.4482	0.6233	3.3648
Microwave_Available	0.4883	0.3075	2.5209	0.1123	1.6295	0.8918	2.9774
Gender	−0.6115	0.3499	3.0541	0.0805	0.5425	0.2733	1.0771
Age (Years)	−0.1300	0.0599	4.7079	0.0300	0.8781	0.7809	0.9875
Age2	0.0011	0.0006	3.3033	0.0691	1.0011	0.9999	1.0023
Family Size	−0.0350	0.0805	0.1891	0.6636	0.9656	0.8246	1.1307
Education	−0.0377	0.0739	0.2601	0.6101	0.9630	0.8331	1.1132
Roof Suitability for Cultivation	3.2308	0.7211	20.0746	7.44 × 10−6	25.2996	6.1564	103.9683
$Log(Roof Area)$	0.0956	0.7076	0.0183	0.8925	1.1003	0.2749	4.4036

) encapsulate the results of our probit and logistic regression models for each intervention. These tables include the essential statistical parameters, including the coefficients, standard errors (s.e.), Wald’s test, $p\text{-}\mathrm{v}\mathrm{a}\mathrm{l}\mathrm{u}\mathrm{e}\mathrm{s}$, odds ratios ($exp\left(b\right)$), and confidence intervals (lower and upper). The coefficients (or beta values) indicate the magnitude and direction of the relationship between each independent variable and the dependent variable, i.e., the likelihood of a household opting for a specific intervention. The standard errors and Wald’s test provide us with information about the reliability of these coefficients. The $p\text{-}\mathrm{v}\mathrm{a}\mathrm{l}\mathrm{u}\mathrm{e}\mathrm{s}$ then allow us to test these relationships statistically. The odds ratios provide a more intuitive understanding of the results by expressing the change in odds for a one-unit change in the predictor variables. The confidence intervals offer a range within which we can be reasonably confident that the true parameter lies. Together, these measures provide a robust foundation for understanding the determinants of intervention preferences.

As shown in Table 6, the probit regression results for the refrigerators as food banks (RaFB) intervention suggest a complex relationship between the variables. The location, coded as Irbid Camp or Sakhra, has a coefficient of −0.012. This value and a high p-value of 0.957 suggest that the location has no statistically significant effect on the outcome. The nationality variable also seems to have no significant effect, given its high p-value (0.905). The house area-related variable has a negative coefficient (−1.590), indicating a negative relationship with the intervention’s outcome. However, the high p-value (0.209) again implies that this effect is not statistically significant. It is worth mentioning that taking logarithms for certain variables was vital to remove the unit effects.

The variable house material-related variable is coded as either concrete or blocks. Its coefficient of −0.649 suggests that a shift from concrete to blocks is associated with a decrease in the dependent variable, suggesting that those living in houses made of blocks are less likely to adopt the RaFB intervention. The p-value of 0.7017 confirms the statistical insignificance of this relationship. Interestingly, the house ownership-related variable has a large positive coefficient of 4.486, yet it is statistically insignificant due to a very high p-value of 0.689, implying that the relationship might be due to chance.

The educational level-related variable has a notable negative coefficient of −0.931 and an extremely low p-value (2.548 × 10⁻⁷), indicating a strong statistically significant negative relationship. This could imply that as the level of education increases, the likelihood of adopting the RaFB intervention decreases. However, given the complexity of this relationship, this result should be further investigated for potential confounding factors. Such results match the descriptive statistical findings illustrated in the previous section.

The variable related to the educational level showed a statistically significant negative relationship with adopting the intervention in the regression analysis. This contrasts with the descriptive analysis, which highlighted the potential importance of education but did not indicate a clear negative relationship. Notably, the higher the education level, the lower the likelihood of adopting the ‘refrigerators as food banks’ intervention. This counter-intuitive result could be further investigated and interpreted in light of other socio-economic factors.

Many variables in the probit regression, such as location, nationality, house-related characteristics, and demographic characteristics (e.g., age and gender), did not have statistically significant effects. This partly aligns with the descriptive analysis, which did not highlight these factors as critical in adopting the intervention. However, the regression analysis allows for more precise estimates of these effects’ size and significance, reinforcing the descriptive analysis’s understanding. These comparisons underscore the complementary roles of descriptive and regression analysis in understanding the complex factors influencing the adoption of the ‘refrigerators as food banks’ intervention. They both shed light on different aspects of the relationships and provide a comprehensive overview of the situation.

Table 7 illustrates that the logistic regression analysis results provide insightful details on how various factors influence the adoption of refrigerators as food bank interventions. The most significant predictor is education, as demonstrated by its strong negative coefficient and extremely small p-value, which suggests that with each additional level of education, the odds of adopting the refrigerator as a food bank intervention decrease. This is a compelling finding and might suggest that individuals with higher education levels may have other means of securing food, lessening their reliance on refrigerators as food banks. However, this is merely an interpretation and further investigation would be needed to understand the causal mechanisms involved.

Conversely, the variable related to the roof’s suitability for cultivation was found to be a positive predictor, although not statistically significant, suggesting a potential trend that families with roofs suitable for cultivation might be more inclined to adopt the intervention. The notion here could be that families who can cultivate their roofs might be more concerned with food security, hence also more likely to use refrigerators as food banks.

Interestingly, many of the other factors (including location, nationality, house material, house area, gender, and age) were not found to be significant predictors in the logistic regression model, which might suggest that adopting the refrigerator as a food bank intervention transcends these individual and demographic factors, being more influenced by socio-economic factors (i.e., education). However, it is important to remember that these findings are based on the available data and model specification and may not capture all the complexities and nuances of the intervention adoption process.

Table 8 represents the probit regression analysis of the cultivated roofs intervention, provides valuable insight, and reveals some significant findings. The analysis results show that the location significantly impacts choosing the CRs intervention, as the Sakhra Region relies more on cultivation when compared to the Irbid Camp. In addition, the house’s structural characteristics are more suitable for the CRs intervention. Moreover, the average roof area is larger. Additionally, the location, specifically the Irbid Camp or Sakhra, has influenced the adoption of the intervention due to different environmental conditions or community attitudes towards such initiatives.

The study further elucidates patterns that validate and diverge from the preceding qualitative analysis. The variable ‘House Material’ emerged as a significant determinant in the probit regression for the acceptance of (CRs) initiatives. This outcome aligns with the preliminary analysis, which inferred that the housing material could significantly influence the decision to implement rooftop cultivation. Block-built houses are typically not engineered to tolerate the weight and permeability associated with rooftop soil; hence, occupants of such dwellings may be disinclined to embrace CRs methods due to the increased likelihood of water seepage and roof degradation when the rooftop is used for cultivation. While the regression paints a more complex picture, it suggests that individuals residing in block-made houses are less likely to opt for the intervention.

The age variable and its square also showed lower statistical significance. Interestingly, the sign of the coefficient is negative, suggesting that the probability of adopting the cultivated roofs intervention decreases as age increases. This finding may be understood in that older individuals may have less propensity or ability to manage a cultivated roof due to potential physical limitations. However, it is essential to note that the squared term of age, which could capture non-linear effects, was not found to be statistically significant.

The variable ‘Roof Suitability for Cultivation’ stands out with a significant positive coefficient, suggesting that as the roof’s suitability for cultivation increases, the likelihood of adopting the cultivated roof intervention also significantly increases. This result is expected, considering that having a roof that is more suitable for cultivation naturally facilitates the implementation of such an intervention.

Many of the variables in the regression (e.g., nationality and house area) were statistically insignificant. This suggests that these factors do not significantly influence the adoption of the cultivated roofs intervention. Such regression analysis helps understand the variables influencing the implementation of the cultivated roofs intervention, highlighting the critical role of roof suitability for cultivation and, to a lesser extent, age and location. The findings provide an empirical basis for targeted strategies to promote the adoption of this intervention.

Drawing on the results of the descriptive analysis conducted earlier, it is interesting to note how they compare and contrast with the outcomes from the probit regression model for the cultivated roofs intervention. The descriptive analysis pointed to the importance of the house area and the availability of certain amenities, such as a refrigerator, cooker, and microwave, in determining the living conditions of the individuals. However, the regression analysis did not find these variables significant predictors for adopting the cultivated roofs intervention. It suggests that while these factors may be crucial for understanding the general living conditions, they might not directly relate to the decision to adopt this specific intervention.

Interestingly, the regression analysis underscored the roofs’ suitability for cultivation variable as a highly significant factor influencing the adoption of the intervention, which aligns well with the descriptive findings, wherein a visible correlation was noted between the suitability of roofs for cultivation and the likelihood of their use for such purposes. On the subject of age, while the descriptive analysis might not have directly implied its importance, the regression analysis highlighted its role, although it is not as crucial as roof suitability. The inverse relationship between age and the adoption of the cultivated roofs intervention is an interesting finding that was not apparent in the initial descriptive analysis.

The location factor, which seemed to contribute to the living conditions in the descriptive analysis, was also significant in the regression analysis. This shows that the living environment, determined partly by the location, might play a role in adopting the intervention. The comparison and connection of these two analyses provide a nuanced understanding of the factors influencing the adoption of the cultivated roofs intervention. The blend of descriptive and regression analyses gives a more rounded view, emphasizing the complexity of the decision-making processes underlying such interventions.

The logistic regression analysis for the cultivated roofs (CRs) intervention, as shown in Table 9, provides clear insight into the factors affecting the adoption of this intervention. A critical finding from the analysis is the significance of ‘Roof Suitability for Cultivation’ as a positive predictor of CRs intervention adoption. Its large coefficient and small p-value (almost zero) suggest that as the roof’s suitability for cultivation increases, so does the likelihood of CRs intervention adoption. This result aligns with intuitive expectations, as families with roofs more suitable for cultivation would naturally be more inclined to implement the CRs intervention.

Location, specifically, whether the household is located in the Irbid Camp or Sakhra, positively impacts the adoption of the CRs intervention, albeit at the border of statistical significance (p-value of 0.0502). This suggests a slight trend that families in one location might be more inclined to adopt the CRs intervention than those in the other. However, this finding should be interpreted cautiously, given the marginal significance level. In addition, an interesting observation is the age variable’s negative but significant effect. It suggests that as the age of the head of household increases, the likelihood of adopting the CRs intervention decreases. It might be that younger households are more open to adopting new interventions or are more aware of and concerned with environmental issues.

Contrarily, many factors, such as housing materials, house ownership, monthly house rental, and education, do not appear to influence the adoption of the CRs intervention significantly. This indicates that adopting the CRs intervention transcends these individual and socio-economic factors and is more influenced by practical considerations (such as house material and roof suitability) and demographic factors (e.g., age and location). As always, while the results provide strong evidence, they should be interpreted in the context of the data and methodology used and with the understanding that the reality might be more complex.

For the RaFB intervention, the probit and logistic regression models show some similarities and notable differences. Both models suggest that education is a significant predictor for adopting this intervention, with education being both negatively and positively related. This indicates that households with lower education levels are more likely to adopt the intervention, according to both models. However, the statistical significance of the refrigerator’s availability is a key discrepancy between the two models. The probit model shows a positive relationship with the adoption of the intervention, whereas the logistic regression model shows a negative but not statistically significant relationship.

Turning to the CRs intervention, the probit and logistic regression analyses highlight the roof’s suitability for cultivation and the type of housing material as a strong and positive predictor of intervention adoption. The location variable also emerged as significant in both models, but with opposing effects: while the probit model shows a positive effect, the logistic regression model has a negative coefficient. Age is another variable with inconsistent results across the models. The probit model indicates a negative relationship between age and the likelihood of adopting the intervention, but the variable is not significant in the logistic regression model. On the other hand, variables such as house ownership and monthly house rental do not appear to significantly influence the adoption of the CRs intervention in either model.

While the probit and logistic regression models for each intervention share some common findings, they also present some discrepancies. These discrepancies might be due to the underlying differences in the assumptions and functional forms of the two models; therefore, it’s crucial to consider the context, research question, and data characteristics when choosing and interpreting regression models.

A comparison with the previous logistic regression analysis for the refrigerators as food banks intervention shows similar and contrasting patterns. For instance, house material was a significant factor for the refrigerators as food banks intervention but appeared to have no significant impact on the CRs intervention. Conversely, the age factor was not significant in the previous analysis, but is significant here, highlighting the importance of considering different factors when planning and implementing different types of interventions. It also underscores the value of performing multiple analyses to better understand the unique factors influencing the success of each intervention.

The receiver operating characteristic (ROC) curve is a graphical plot that represents the diagnostic ability of a binary classifier system as its discrimination threshold is varied. The curve is created by plotting the true positive rate (TPR), also known as the sensitivity, recall, or hit rate, against the false positive rate (FPR), the fall-out or probability of false alarm, at various threshold settings. The TPR is plotted on the y-axis and represents the proportion of actual positives (e.g., households adopting a specific intervention) that are correctly identified as such. On the other hand, the FPR is plotted on the x-axis and denotes the proportion of actual negatives (e.g., households not adopting the intervention) that are incorrectly identified as positives. The area under the ROC curve, or AUC, measures how well a parameter can distinguish between two diagnostic groups (diseased/normal or adopter/non-adopter). A model whose predictions are 100% correct has an AUC of 1, while a model with no better than random chance has an AUC of 0.5; hence, the closer the curve follows the left-hand border and then the top border of the ROC space, the more accurate the test. It is worth noting that the ROC curve does not depend on the class distribution, making it a popular choice for evaluating the performance of machine learning models, especially in scenarios where there are imbalanced classes. The receiver operating characteristic (ROC) curves for the probit and logistic regression for each intervention are shown in Figure 4.

Drawing comparisons between the results of the probit regression analysis and the descriptive analysis can offer valuable insight into the intricacies of the study. The descriptive statistics provided initial insight into the preferences of refugees for the refrigerators as food banks (RaFB) intervention over cultivated roofs (CRs). It reported that a substantial majority (90%) preferred RaFB, indicating a clear leaning toward this intervention.

The probit regression analysis further deepened our understanding of these preferences by quantifying the influence of various factors. It confirmed some of the findings from the descriptive analysis but also provided new insight. For instance, the descriptive analysis showed no significant influence of the demographic factors, such as age, gender, and household size, on the acceptance of the interventions. This was also echoed in the probit regression, where these variables proved statistically insignificant, signifying that they did not significantly impact the preference for RaFB.

In contrast, the education variable, seen as a significant factor influencing the acceptance of RaFB in the descriptive analysis, was also found to be significant in the probit model but with a negative coefficient, which implies that the likelihood of preferring RaFB decreases as the education level increases. This complements the finding from the descriptive analysis that those with higher education levels were more likely to reject RaFB. Furthermore, variables such as house materials that were not explicitly highlighted in the descriptive analysis were found to significantly influence the preference for RaFB in the probit model. The negative coefficient suggests that a shift in house material from traditional to modern decreases the preference for RaFB, providing a nuanced understanding that was not clear from the descriptive analysis alone.

While the descriptive analysis provides a broad overview of the refugees’ preferences, the probit regression adds depth by quantifying the impacts of various factors and highlighting potential areas for intervention to increase the acceptance of RaFB. These findings provide a more robust basis for policy recommendations, offering key insight into which factors might need to be addressed to increase the adoption of the preferred RaFB intervention.

Looking back at the previously stated results from the descriptive analysis, the regression results largely align with these while also providing nuance and depth. In the descriptive analysis, we established that a significant portion of refugees indicated a preference for the cultivated roofs (CR) intervention. The regression analysis affirmed this and went a step further to identify the factors that drive this preference. Particularly, the suitability of a roof for cultivation emerged as a strong determinant in the regression analysis, which resonates with the descriptive analysis’ finding of the high acceptability of the CR intervention.

The age factor showed an interesting dynamic. While the descriptive analysis observed a wide range of ages among the refugees preferring the CRs intervention, the regression analysis revealed that the likelihood of preferring CRs decreases as age increases. This suggests that while a broad age group may favor the CRs intervention, it is more popular among younger individuals. Household characteristics, such as the house type, material, area, and appliances (refrigerator, cooker, microwave), were insignificant in the regression analysis, which mirrors the descriptive analysis, which did not highlight any particular patterns concerning these variables. It seems that the preference for CRs is largely independent of these factors. Contrastingly, while the descriptive analysis revealed different preferences for the CRs intervention among various nationalities, the regression analysis suggested that nationality does not have a significant effect. This discrepancy might be due to the high standard error and p-value associated with the nationality variable in the regression model. More investigation may be needed to resolve this inconsistency. In addition, the regression and descriptive analyses offer complementary insight into refugee preferences for the CRs intervention. While the descriptive analysis provides an overall picture, the regression analysis digs deeper into the specific factors influencing these preferences.

Several key trends emerged in synthesizing the findings across the descriptive analysis, probit regression, and logistic regression for both interventions. For the refrigerators as food banks intervention, the descriptive analysis indicated that the number of household members, education, and house materials were important factors, which was generally supported by the probit and logistic regression analyses, with the education and house material variables emerging as significant. Despite this, differences in the results across the analyses—for instance, the marital status factor was significant in the logistic regression but not in probit—illustrate the complexity of determining intervention preferences and the importance of considering different analytical perspectives.

5.3. Most Accessible Food Bank Location Selection

The implementation of refrigerators as food banks in refugee camps has the potential to contribute to achieving zero hunger through the preservation and secure storage of food resources. However, careful consideration of the context-specific challenges, including access to electricity, cost, cultural acceptance, and logistical factors, is crucial to ensuring the successful implementation of this intervention. Further research and investigation into the factors influencing refugee preferences and acceptance of different hunger alleviation strategies can provide valuable insight to inform the design and implementation of contextually appropriate and effective interventions.

5.3.1. Integrating Geographic Information Systems (GIS)

Integrating geographic information systems (GIS) in strategic decision-making processes can immensely augment the effectiveness and efficiency of resource allocation. To combat hunger and improve food accessibility, a deeper understanding of spatial distribution and population densities can be crucial. The following section of our research employs GIS, specifically, kernel density estimation (KDE), as an innovative approach to identify optimal locations for placing refrigerators to serve as food banks in both areas under study. KDE is a non-parametric method used to estimate the probability density function of a random variable, which, in our context, would translate to an accurate representation of the population density and food demand. This method aims to establish an optimal distribution of food banks to maximize their reach and minimize food wastage. The subsequent analysis could potentially contribute to more strategic and evidence-based planning, and most importantly, it could bring us a step closer to ensuring food security for all.

5.3.2. Irbid Camp and Sakhra Region Overview

The Irbid refugee camp and Sakhra sub-district are two locations in northern Jordan considered for our research. The Irbid refugee camp traces its roots back to 1951, when it was established to accommodate Palestinian refugees uprooted by the 1948 war. Situated near Irbid, the camp originally covered an area of 0.24 square kilometers and sheltered about 4000 refugees. According to the Department of Palestinian Affairs’ 2021 data, the camp now spans 244 dunums and provides shelter to 29,894 refugees [31]. On the other hand, the Sakhra region Is part of the Al-Junaid municipality within the Ajloun Governorate. The sub-district encompasses several villages, including Sakhra, Ibbin, Ibilin, Samta, Munif, Deir al-Barak, and Khirbet Fara. Covering a geographical expanse of 57.9 square kilometers, Sakhra is home to an estimated 39,480 people [32,33].

5.3.3. Applying Kernel Density

With the GPS coordinates of the families mapped on ArcGIS, we utilized a kernel density analysis to determine the feature density near these coordinates. The kernel density analysis, applicable to point and line features, allowed us to identify regions with a high family concentration and select optimal locations for establishing food banks, considering factors such as distance and accessibility. After applying the kernel density analysis, we segmented the results into three density-based categories: high, medium, and low. It was observed that mosques make for optimal food bank locations, considering their accessibility and proximity to families. Since mosques remain open throughout the day and the distance between families and these mosques is less than 200 m in the Irbid camp and less than 1 km in the Sakhra region, they serve as ideal locations for the food banks. Figure 5 visually represents the geographical locations of the Irbid Camp and Sakhra region in Jordan, which are the primary study sites for this research. The areas are distinctively marked to highlight their relevance to the study. The map exhibits the precise positioning of Irbid Camp and the Sakhra sub-district within the broader landscape of the Ajloun Governorate in northern Jordan. It provides viewers with a geographical context and perspective, aiding the comprehension of the research scope. Furthermore, it provides an understanding of the regions’ relative positions and distances, which is crucial for analyzing the logistical aspects of the proposed interventions for achieving zero hunger among the refugee populations in these areas.

The most accessible location for the food bank within these areas is determined by the central feature selection method. This method involves calculating and aggregating the distances from the centroid of each feature to the centroid of all the other features within the dataset. After considering weights, if specified, the feature with the shortest cumulative distance to all the other features is highlighted and transferred into a newly constructed output feature class. As shown in Figure 6, given their several logistical benefits, the suggestion has been advanced to establish food banks at local mosques. The presence of electricity, a caretaker, and the regular opening five times a day contribute to the mosques’ suitability as accessible and operationally viable locations for the food banks.

The main advantages and barriers facing the deployment of the projected interventions were highlighted. The CRs project benefits involve enhanced food production, reinforced nutrition systems, and augmented refugee self-reliance. Nevertheless, challenges with such an intervention include the relatively high initial investment essential for substructure progress, enduring preservation expenses, roof stability, lack of freedom to use rented house roofs, insufficient knowledge in roof cultivation, the high cost of materials and equipment, and the necessity of guidance and capacity building amongst refugees to guarantee efficacious execution. The RaFB intervention offers the advantages of granting an immediate solution for food storage and preservation, decreasing food surplus, and staging food sharing amongst community members. In spite of these advantages, some possible obstacles include acquiring a consistent energy source, confirming identifiable repairs and hygiene, and addressing prospective concerns associated with food safety and equitable access.

The current study’s findings suggest that both CRs and RaFB have the potential to contextually appropriate and provide sustainable food security interventions in refugee camps. However, the demographic influences on project preferences highlight the importance of considering the target populations’ unique needs and characteristics when designing practical solutions for zero hunger within the pockets of poverty and implementing contextually appropriate and effective interventions. It was also revealed that the RaFB is the most applicable and efficient solution vital for achieving zero hunger in the targeted communities.

6. Theoretical and Practical Implications

The current research has various theoretical and practical implications that can positively influence the robustness of the strategic planning for achieving zero hunger in the poverty pockets across Jordan. The theoretical implications include recognizing preferences, as the current study findings contribute to the current body of knowledge by judging the preferences of refugees in reaching zero hunger in camps and vulnerable areas. Thus, the conclusions can be employed to develop guided strategies that accommodate the refugee population’s particular needs and preferences. In addition, the research results are vital for highlighting the association between demographic factors, such as age, gender, education level, and family size, and the preferences of refugees for different approaches to achieving zero hunger. Such information can help policymakers and practitioners better understand the factors influencing refugees’ preferences and develop tailored interventions. Moreover, the research results highlight the significance of involving refugees in decision-making and considering their preferences when designing interventions. Such an approach can empower refugees and foster a sense of ownership and responsibility in achieving zero hunger in their communities.

Theoretically, this study enriches the existing literature by delving into the preferences of refugees in approaching zero hunger in camps and vulnerable regions. The findings align with those of the authors of [19], who identified key factors influencing the youth’s labor market outcomes and vulnerability, including family wealth and parental education. The current study further illuminates how demographic factors, such as age, gender, education level, and family size, interplay with refugees’ preferences for different strategies to attain zero hunger. These findings provide a bridge to [20] work on empowering refugee families through psychosocial support programs. This research underscores the significance of integrating refugees into decision-making processes and respecting their preferences in designing interventions. This aligns with the growing global consensus on refugee participation, as the author of [26] highlighted, and enhances refugees’ sense of ownership and responsibility toward eradicating hunger within their communities.

This study offers policymakers, aid organizations, and other stakeholders a roadmap to allocate resources effectively. By prioritizing interventions like $RaFB$, which most refugees prefer, we increase the likelihood of these strategies being well-received and successful [21]. This echoes the study wherein Syrian refugees demonstrated a positive attitude towards blended learning, a transformative approach to achieving academic objectives. Additionally, the research underlines the importance of crafting interventions specific to refugees’ needs and demographic profiles, an approach also advocated by the authors of [27] in addressing urban displacement crises. This study identifies challenges that could hinder specific interventions and proposes potential solutions that can lead to a more efficient and effective implementation, ultimately maximizing their impact in achieving zero hunger. The current research promotes capacity-building among refugees and encourages innovation in addressing zero hunger. It aligns with the findings of the authors of [22] on the refugee influx’s impact on land use/land cover changes in northwestern Jordan. By exploring refugees’ preferences and ideas, we can develop novel approaches to hunger reduction that cater to the unique needs of this vulnerable population. Overall, this study offers both a theoretical and practical framework for integrating and addressing the needs of refugees in our quest for zero hunger.

The current research has various implications that can be utilized to practically deploy the necessary intervention within vulnerable regions, including guiding policymakers, aid organizations, and other stakeholders in allocating resources more effectively. By prioritizing interventions that most refugees prefer (i.e., $RaFB$), they can ensure that the interventions are more likely to be successful and well-received by the target population. In addition, the study highlights the importance of tailoring interventions to the specific needs and preferences of refugees, considering their demographic profile. Customized interventions are more likely to be successful and sustainable, as they cater to the unique needs of the target population. Additionally, by identifying the challenges associated with implementing specific interventions (such as cultivated roofs), this study can help stakeholders better anticipate and address potential obstacles. This can lead to more effective and efficient implementation of interventions, maximizing their impact in achieving zero hunger.

The study’s findings can be used to inform capacity-building efforts among refugees. For example, by identifying the lack of knowledge in roof cultivation as one of the reasons for the low preference for cultivated roofs, organizations can develop targeted training programs to address this knowledge gap and build the skills necessary to implement this intervention. Furthermore, the study encourages innovation in addressing the issue of zero hunger in refugee camps and vulnerable areas. By exploring the preferences of refugees and their ideas for achieving zero hunger, the study can inspire the development of novel approaches and interventions that consider the unique needs and preferences of the refugee population.

7. Conclusions

The current study is designed to investigate the preferences of refugees in deploying viable interventions to achieve zero hunger in poverty pockets and vulnerable areas across Jordan. The research adopted an assorted-approaches method, blending quantitative surveys and GIS technologies, to generate in-depth statistics regarding the preferences of refugees and the demographic factors that affect these preferences. The study’s outcomes stress that refugees prefer RaFB to the CRs intervention. The current research also sheds light on the function of demographic factors (e.g., age, gender, education level, and family size) in forming these preferences. The results also highlight the potential advantages and challenges of applying each proposed intervention.

The current study’s theoretical implications subsidize an enhanced comprehension of refugees’ preferences in achieving zero hunger within their communities. Moreover, the role of demographic factors in shaping these preferences was illustrated in addition to the significance of involving refugees in the decision-making procedure. The current research’s practical implications are critical to managing food allocation, informing the design of adapted involvements, highlighting the potential advantages and implementation challenges, supporting capacity-building strategies, and inspiring novelty in achieving zero hunger in the pockets of poverty and vulnerable areas in Jordan.

The current research underlines the necessity of prioritizing the voices and preferences of the targeted groups when planning, originating, and executing sustainable interventions to achieve zero hunger. By promoting a sagacity of proprietorship and accountability amongst refugees and tackling this populace’s unique challenges, legislators, support groups, and other stakeholders can work together to advance more operative, practical, and sustainable interventions to alleviate hunger and guarantee food security for refugees and their communities.

The current study’s findings have led to several policy recommendations. Notably, strategies such as refrigerators and food banks ($RaFB$), which were preferred by many refugees, should be prioritized in policy planning. This calls for more efficient resource allocation by policymakers, aid organizations, and various stakeholders, considering the refugees’ specific needs and preferences in the design of interventions. Furthermore, identifying potential challenges linked to implementing certain interventions, such as cultivated roofs (𝐶𝑅𝑠), emphasizes the need for strategic planning to address these issues proactively. The policy framework should also incorporate capacity-building measures, manifesting as targeted training programs to bridge the knowledge and skill gaps about specific interventions, such as roof cultivation, amongst the refugee populace.

The current research has limitations, as it focuses on a particular refugee community within a specific geographical context; caution is suggested when extrapolating these results to other contexts or refugee populations without additional research or adaptations. In addition, the current study’s emphasis on demographic factors may not encapsulate all the influential socio-economic or psychological variables. For more holistic explorations, future research should consider these elements and others, such as cultural nuances. Moreover, future research could assess the efficacy of the preferred interventions, such as $RaFB$, in achieving zero hunger, thereby informing better policy planning. Extending the scope to various geographical areas and refugee demographics could provide a more comprehensive understanding of refugee preferences. Employing sophisticated machine learning models (e.g., $XGBoosting$ and $Lightgbm$) in future studies may also offer a more nuanced analysis of the variables’ correlations.