Group Projects‎ > ‎

Group 04: Livelihood and Security of Urban Refugees in Delhi, India

Livelihood, Security, and Access to Services among Urban Refugees in Delhi, India

Project Summary

Refugees and internally displaced persons (IDPs) are increasingly moving to urban areas, including in Delhi, India, where the UNHCR serves more than 24,000 refugees (JIPS Delhi Report). In collaboration with the UNHCR and Tufts University, the Joint International Profiling System (JIPS) surveyed over one-thousand households in Delhi, focusing on refugees from Afghanistan, Myanmar, and Somalia, as well as Indian citizens. Using this dataset, community-level service information, and spatial analysis, we assessed household level livelihood security and perceptions of physical safety and compared them to infrastructure within Delhi. We found that areas with high living standards (defined by our livelihood index scoring system) tended to be clustered, and that there were larger areas with highly clustered livelihood scores than areas with highly dispersed values. Surprisingly, we found that households within walking distance to public services – police stations, hospitals, schools, markets, and religious sites – did not have a significantly higher livelihood scores than other households. However, households close to police stations did report higher perceptions of safety and lower rates of assault than households that were far from a police station. We hope that this report helps inform JIPS regarding refugees in Delhi and that it contributes to a larger dialogue on the potential of spatial analysis for understanding and supporting urban refugee populations.


Refugees are following the global trend of urbanization, and increasing numbers of refugees live in urban areas, integrated with the native population, rather than isolated in camps. A 2009 study by the United Nations High Commissioner for Refugees (UNHCR) estimated that 4.8 million refugees (46%) live in urban areas, compared with 2.5 million (24%) who live in camps (UNHCR Statistical Yearbook 2009). According the UNHCR, this trend has continued, and more than half of the refugees UNHCR serves now live in urban areas.

Integrated with the native population, urban refugees face the myriad challenges common to urban living, including poverty, unemployment, crime and access to health services and education, but refugees also face discrimination and abuse (Chin Refugees in Delhi, 2013; Chin Human Rights Organization, 2009).

In India, the government has no national legislation or implementing framework to protect the rights of refugees, nor has it signed the 1951 Convention Relating to the Status of Refugees or the related 1967 Protocol.  This is particularly concerning because India has 187,482 documented refugees as of August 2013 (UNHCR India Fact Sheet, 2013).  In the National Capital Territory (NCT) of Delhi (which includes the capital city of New Delhi), UNHCR assists over 24,000 urban refugees[1] from its headquarters office in New Delhi (JIPS Delhi Report, 2013).  This represents just 0.001 percent of Delhi’s total population of about 23 million (JIPS Delhi Report, 2013).

Refugees in India come primarily from Afghanistan, Myanmar, and Somalia, and the Delhi refugees represent a subsection of these populations (See Table 1 and Figure 1) (JIPS Delhi Report, 2013; UNHCR India Fact Sheet, 2013).

Table 1. Country of Origin of Refugees and Asylum-Seekers in Delhi, India

Source: JIPS Delhi Report, 2013. Current as of Jan. 2013.


Figure 1. The majority of refugees and asylum-seekers who make their homes in Delhi, India, come from Afghanistan, Myanmar, and Somalia.

In early 2013, UNHCR and Joint Internally Displaced Persons Profiling Service (JIPS) partnered with the Feinstein International Center at Tufts University to conduct a study in Delhi to better understand the living conditions of urban refugees. They surveyed 1,063 households in the NCT of Delhi, including refugees from Afghanistan, Myanmar, and Somalia as well as Indian citizens (JIPS Delhi Report, 2013).  The survey did not conduct a random sampling, but rather targeted areas with known refugee populations, identified using UNHCR refugee registration data.    

Survey respondents were asked about their experiences in Delhi.  Questions particularly focused on four categories, identified by JIPS: employment security, housing security, financial security, and physical safety (JIPS Delhi Report, 2013).  Initial statistical analysis conducted by JIPS suggests that urban refugees in Delhi face greater challenges than their Indian neighbors due to discrimination, but that refugee experiences may also differ according to country of origin (JIPS Delhi Report, 2013).

Our goal was to add spatial analysis capabilities in order to better identify and understand geographic patterns related to refugee security.  We first created a Livelihood Index (LI) score for each household, based on survey responses to questions in each of the four categories.  This enabled spatial analysis of the distribution of high and low LI scores with respect to one another, ethnicity, and proximity to public services. 

We focused particularly on physical safety because safety is such a basic component of quality of life and because the literature suggests that urban refugees routinely face discrimination and physical assault (Chin Refugees in Delhi, 2013; Chin Human Rights Organization, 2009). With respect to physical safety, in the survey, JIPS asked heads of households: (1) whether they felt safe on their commute to work, (2) whether they had been assaulted, and (3) whether they felt safe in their neighborhood.  We questioned whether physical safety, as assessed by responses to these questions, would correlate with proximity to police stations.  Intuitively, it seems that crime rates would be lower and therefore safety would be higher near police stations.  However, in 2010 more corruption cases were filed against the Delhi Police by Delhi’s anti-corruption branch than were filed against any other department, which could mean that people feel less safe near police stations (Chauhan 2011).  This could be especially true for refugees, if they felt that the police were a source of discrimination.

[1] Although there are legal distinctions between refugees and asylum-seekers, for the purpose of this report and our analysis, we consider documented and un-documented refugees and asylum-seekers to be “refugees.”

Interactive Map

The map below shows where services are located in our study area of Western Delhi, India, where interviewed households are located, and what locations are optimally close to services. If the Google Gadget map is not loading, please click here.

For more detailed maps and analysis, please see our interactive storyboard map. You can add or omit layers to get a better understanding of our study area, and you can even measure distances from households to services yourself to determine which households have the best access to services. 

Research Objectives & Study Area

Research Goals

Our project aimed to understand the drivers of livelihood and safety for refugees in Delhi, India. We also wished to explore differences in livelihood and safety experienced by refugees from different countries and by Indian citizens. Our main questions were:

i)                   Is there evidence of clustering of livelihood or safety in different neighborhoods?

ii)                 What effect does geographic proximity to public services have on perceived and actual livelihood and safety?

We also sought to develop a framework for a Livelihood Index (LI) score that could be used by JIPS in future analysis to explore further patterns in refugee quality of life.

Study Area

Our study focuses on the National Capital Territory (NCT) of Delhi, India, in order to support the on-going JIPS study of urban refugees in Delhi.  As mentioned in the introduction, Delhi has a large number of refugees (over 24,000), though they comprise a small percent of the total population of 23 million (0.001%) (JIPS Delhi Report, 2013).  Delhi is one of the world’s most populous cities, and it suffers from poverty and crime.  An estimated 8% of the city’s population live on less than US$11 per month (Economic Survey, 2006).  Delhi is also home to the highest rates of kidnapping and violence against women in India (National Crime Record Bureau, 2012).  Urban refugees in Delhi therefore face challenges, including access to public services, that may affect livelihood and safety.

We conducted two levels of analysis at two geographic scales.  First, we explored the spatial distribution of Livelihood Index (LI) scores and physical safety component scores for all 1,063 households distributed throughout the NCT of Delhi (Figure 1).  The NCT of Delhi comprises 9 districts, including the city of New Delhi, and the borders are defined by the Indian Government (Indian Census Administrative Districts, 2011).

Second, we conducted an in-depth assessment of livelihood and safety as they relate to proximity of public services.  We did not have the capacity to gather public service infrastructure data for the entire NCT of Delhi.  We therefore chose to focus our analysis on a cluster of 674 households in western Delhi, a geographic space we defined using GPS coordinates (bottom left corner= 28.59, 77.028; top right corner = 28.659, 77.136) (Figure 2).  This cluster was also identified by JIPS as one of their main survey sampling clusters in its 2013 Report.  It was the largest cluster of survey respondents (63% of total households), and it contained Indian, Myanmarese, and Afghan respondents (Figure 3).

Figure 2. Study area in western Delhi, selected to enable in-depth assessment of public services infrastructure in an area with a representative selection of refugee households.

Figure 3. Ethnicity of Household Respondents in Western Delhi (data from JIPS Delhi Report, 2013).


Our methodology consisted of several steps, corresponding to each of the two research questions.

1. Is there evidence of spatial clustering of livelihood or physical safety indicators?

To answer this research question, the following methodological steps were taken:

(1)  Geocoding of household locations;

Geocoding was done using the latitude and longitude data corresponding to each household as reported in the JIPS survey.

(2)   Calculation of an overall Livelihood Index and separate sub-indices;

An overall Livelihood Index was calculated based on four dimensions: employment security, housing security, financial security, and physical safety. Each dimension is composed of multiple indicators based on survey data from Joint IDP Profiling Service (JIPS), as shown in Figure 4.  Below are the equations that were used:


All indicators and dimensions are equally weighted, as we think that they all have an equal impact on overall livelihood. Our partner organization, JIPS, also felt that this was a reasonable assumption.

The Safety indicator was calculated for all of the 1,063 households in the JIPS survey dataset (as opposed to just those located in the western area of Delhi). Three of the sub-indicators could be constructed using the full “Households” dataset, containing 1,063 observations. However, the remaining two indicators – related to whether the breadwinner of the household had experienced physical abuse at work and whether he/she perceived the commute to work as being safe – were only available for a subset of households. Namely, while the “Breadwinners” dataset, which contains responses pertaining to these two sub-indicators, has 1,394 observations, not every household has matching breadwinner data while some households have data for two breadwinners. Moreover, 20 of the records in the “Breadwinners” dataset did not have data corresponding to these particular two questions. Ultimately, there were 964 households that had data on all five Safety sub-indicators. Households that had corresponding data for not just one but two breadwinners were assigned the score that is the “lowest denominator” of the two.  For example, if one of the household’s breadwinners reports having experienced physical assault at work while the second breadwinner reports not having ever experienced assault at work, the household is assigned a score of 0 (“breadwinner has experienced physical assault at work”).

(3)   Implementation of a hotspot analysis on the overall Livelihood Index and the sub-indices to determine whether there are within-neighborhood clusters of particularly high or low values and whether there are distinct patterns by household ethnicity.

Hotspot analysis was run for the overall Livelihood Index scores for the entire western Delhi sample, followed by separate analyses for (1) the different sub-indicators and (2) the two ethnicities – Indians and Myanmarese – that live in this area. A spatial extent of 6 kilometers was defined in conducting the analysis, given the spatial distribution and distances between the individual households. Following the Hot Spot analysis, the resulting z-scores were used to create Inverse Distance Weighted raster and Density Kernel raster surfaces separately for each sub-indicator and for each of the two ethnicities’ total livelihood scores.

Figure 4: Livelihood index framework

2..   What impact does proximity to different types of services have on perceived and actual livelihood and safety?

To understand whether the proximity of services – such as schools, hospitals, or public transportation – have on the livelihood and physical security of the interviewed households, the following methods were used:

(1) Geocoding of household locations using the geographic coordinates recorded in the JIPS survey;

See Section 4.1. (1).

(2) Calculation of an overall Livelihood Index and separate sub-indices using JIPS survey data;

See Section 4.1. (2).  Figure 5 below shows the distribution of overall Livelihood Index values across all the households in the western Delhi study area. The distribution appears to be approximately normal, although the scores are somewhat more concentrated around the median values.

Figure 5: Histogram of distribution of Livelihood Index scores


(3)   Digitization of various types of services hypothesized to be potentially important for livelihood and physical security;

As noted earlier, the geographic focus on analysis was an area in the western part of Delhi, defined by GPS coordinates: bottom left corner= 28.59, 77.028; top right corner = 28.659, 77.136.  The following types of services were digitized in this area using Google Maps: hospitals, bus routes and stops, metro routes and stops, churches, Hindu temples, mosques, and schools. However, in order to be able to analyze the impact of distance to police stations on physical safety scores, police station locations were digitized for the entire Delhi metropolitan region (marked in purple in the locator map). The decision to expand the safety-specific analysis to the entire city was made due to the team’s particular interest in the safety question as well as to enhance the statistical power and external validity of this particular analytical component.

Locations for each of the services were identified with the search tool, and locations were marked with vector lines (metro and bus routes) or points. The locations were saved as GPS points onto personal Google maps, and the data were exported as KML documents. The KML documents were converted to Shapefiles on ArcGIS and projected into the Azimuthal Equidistant coordinate system.

(4)   Implementation of a buffer analysis to determine if households optimally located within walking distance of all services are better off than households who are not located near all services;

The proximity analysis, focused on the Western Delhi study area, examined the effect of being proximate to services that were hypothesized to have a positive impact on livelihood and security: police stations, bus routes, churches, temples, mosques, hospitals, markets, metro stations, and schools. The buffers for churches, temples, and mosques were combined to create a “religious buildings” layer and the buffers for bus routes and metro stations were combined to create a “transportation layer.” Our analysis was based on a buffer distance of 0.5 miles, based on the findings from transportation literature where 0.5 miles is deemed to be “walking distance” (e.g. Moudon et al., 1997).  However, we used a larger buffer of 1 mile for police stations and hospitals, where the effectiveness of the serve is related less to being within a “walkable” distance and more on being within quick emergency response time.  

Figure 6: Example service buffer of 0.5 miles around markets in western Delhi

Next, all of the buffers—religious buildings, transportation, hospitals, markets, schools, and police stations—were intersected to create a layer of optimal housing locations based on proximity to services. Figure 7 illustrates the overall buffer zone we identified.

Our final outputs are a set of households that are close to all services and a set of households that are not close to all services. The Livelihood Index values for these two groups of households were then compared using a t-test in the R statistical analysis software to determine if the groups were statistically different. Since we found that the two groups were not different in a statistically significant way, we decided to explore the exact relationship between the distance to each type of service (here, treating distance as a continuous variable as opposed to a binary “in/out” type factors) and the calculated Livelihood Indices.

Figure 7: Combined service buffers to identify optimal areas within walkable distance of services within western Delhi.

(5)   Statistical validation of the buffer analysis results to determine the exact impact of distance to each type of service on the households’ overall Livelihood Index;

Euclidian distance from each household to the nearest (i) market, (ii) school, (iii) hospital, (iv) bus station, (v) metro station, (vi) church, (vii) mosque, and (viii) Hindu temple was calculated in ArcMap, using the “Near” function. Ordinary Least Squares (OLS) regression analysis was then used to statistically estimate the relationship between the distance to each type of service (independent variable) and the overall Livelihood Index (dependent variable). As shown in the table below, on average, the interviewed household distance to services varies significantly by type of service, with schools and bus stations typically being located the closest and metro stations and markets – the furthest.

Table 3: Household distance to nearest service (m), by service type









Bus station


Metro station






Hindu temple



(6)   Statistical analysis of the relationship between the households’ physical Safety indicators and their proximity to police stations.

As noted previously, the Safety-specific statistical analysis focused on the entire Delhi area. Similarly to other services, police proximity was calculated using the “Near” function in ArcMap. For each of the households, the function identified the nearest police station and calculated the exact Euclidian distance to that station.  The shortest distance turned out to be 128.2 meters, while the longest was 2,296.6 meters.  Figure 8 shows the overall distance-to-police distribution across all households – it appears to roughly follow the normal curve.

Figure 8. Histogram of distribution of household distance to police stations

In addition, we calculated distances based on gender and ethnicity so we could later explore whether these factors also affected overall perceptions and safety (e.g. women may be more afraid for their families or may experience more violence against women, or Somali refugees may experience more discrimination than Myanmarese refugees). We found that, while the average distance to police stations was slightly larger for female respondents (1,297 meters compared to 1,271 meters for males), the median distances were exactly equal – 1,281 meters. By ethnicity, Somali households, on average, live closest to police stations, followed by Indian households. Both in terms of mean and median distance, Myanmarese and Afghan households live the furthest away from police – approximately 1,330 meters. As also shown in the boxplots below, however, the range of distance to police stations is the widest for this Somali ethnic group.

Figure 9: Average Distance to Police, by ethnicity of respondent’s household

Following the distance calculation and its diagnostic statistical analysis, Ordinary Least Squares (OLS) regressions were run to statistically confirm whether or not distance to police stations is a significant predictor of perceived or actual safety.

Data Sources


1 Is there evidence of clustering of livelihood or physical safety in different neighborhoods?

The hot spot analysis for the household sample as a whole, across the National Capital Territory (NCT) of Delhi showed that:

  •  Employment security scores tended to have high clustering of low z-score values (i.e. cold spots) compared to other livelihood indicators, indicating a dispersed distribution of employment attainment. This result suggests that there is a relatively minimal negative economic dependence among households living in the study area, although this pattern could be different (more clustered) for individual socioeconomic groups or ethnicities that are linked through social networks.
  • Financial security indicators showed both distinct clustering of particularly high and particularly low z-score values, indicating that individual neighborhoods have higher-than-normal clustering of financial security scores while others – more dispersed distribution of financial security as compared to random. This result is also fairly intuitive and can possibly be explained by neighborhood-specific social network effects that are manifested spatially – i.e. those involved in more community groups and receiving more financial help from these and other types of entities relay this experience to their neighbors who then are able to enhance their financial security.  
  • Physical safety also showed intense clustering on both particularly high and particularly low z-score values; however, the clustering of particularly high values was more distinct, indicating higher-than-normal spatial clustering of physical safety scores. Interestingly, when the police stations were added to the hot spot raster surface, they tended to be located in the areas with the high clustering, intuitively suggesting a distinct (either positive or negative) relationship between reported and perceived safety and proximity to police. This finding suggests that the local presence of law enforcement, even when suffering from institutional corruption, can be a positive influence on refugee safety. However, without estimating the exact statistical impact, it is difficult to tell whether particularly high or particularly low physical safety scores cluster in the police station areas.
  • Housing security hot spot analysis predominantly showed large areas of high z-value clusters, indicating higher-than-normal clustering of housing security values. This outcome is intuitive from a spatial point of view, given that housing security is probably the one livelihoods indicator that has many physical attributes and therefore can be expected to be correlated across space. Also, survey respondents were sometimes located within the same apartment complex, so if one apartment is of good quality, it is expected that others at the same location will also be of good quality.
  • Livelihood Index scores reflected all these different patterns; on balance, there appeared to be more distinct clusters of high, albeit not very high, z-score values compared to low ones.

There was significant difference in the hot spot derived surface rasters for the LI between Indian and Myanmarese households in Western Delhi. While “hot” and “warm” spots covered nearly the entire surface in the Indian-specific analysis, the Myanmarese LI score raster was dominated by high clustering of particularly low values.  This result seems counter-intuitive at first, given anecdotal evidence from the JIPS report that refugees were highly cooperative and co-dependent.  Our results suggest that such cooperation is not translating into increased LI scores, at least as measured through spatial patterns.  Another possible alternative is that the refugees with highest or lowest LI scores are not as tied into the refugee mutual support networks.

2 What effect does geographic proximity to public services have on perceived and actual livelihood and safety?

Overall Livelihoods

Individual OLS regression analyses focusing on the impact of service proximity on overall Livelihood security in Western Delhi showed that:

  • Distance from markets, hospitals, schools, churches, and bus stations has a statistically significant positive effect on overall livelihood index scores, meaning that livelihood scores were in fact higher for households living further away from these services. This result is somewhat surprising and contradicts our initial hypothesis. However, it could be explained by selection bias – i.e. that services tend to be located in areas (such as the inner city) that are less residential and therefore less well-off, implying a negative correlation between overall well-being and presence of services. An alternative explanation to this finding is the presence of informal services – markets, schools, or health facilities – that are perhaps more important for refugee day-to-day well-being but are not easily identifiable in data sources such as Google Maps. 
  • Distance from metro stations has a statistically significant negative effect on overall livelihood scores.  This result suggests that physical mobility plays an important role in enhancing the livelihood security of the interviewed households.  In fact, the effect of physical mobility on livelihood security might be through its role on access to other services – such as formal or informal markets, schools, and other facilities. Still, while highly statistically significant, distance to metro stops explains only about 4% of the overall variation in the household Livelihood scores. 

The difference between metro stations and bus stations is interesting, as we conducted our analyses based on the assumption that all forms of public transportation were equal.  However, this is unlikely to be true, and it may be that metro service is a better predictor of high livelihood security.

Regression 1: Effect of distance from metro stops on overall Livelihood index

                                  Estimate               Std. Error                t value                 Pr(>|t|)   

(Intercept)                 5.632e-01              2.032e-02                27.719                 < 2e-16 ***

Metro distance       - 6.169e-05              1.133e-05                 -5.447                  7.15e-08 ***

Residual standard error: 0.1188 on 688 degrees of freedom

Multiple R-squared:  0.04134,         Adjusted R-squared:  0.03994


The main findings from the series of OLS regressions, focused on the city-wide (entire NCT of Delhi) reported Safety scores as related to distance to police stations, are as follows:

  •  Distance to police stations has a statistically significant negative impact on overall safety scores, perceived neighborhood safety, actual physical assault, and the household breadwinners’ perceived safety while commuting to work. This indicates that the households living closer to police stations not only perceive their surroundings as more safe but are indeed better protected from physical harm. Regression 2 below shows the exact estimated coefficients, standard errors, and significance scores. It also shows that, despite exerting a statistically significant influence on safety, distance to police only explains about 1% in the variation in reported safety, as indicated by the R2 value.

Regression 2: Effect of distance to police on overall Safety score

                                        Estimate           Std. Error                t value               Pr(>|t|)   

(Intercept)                      7.816e-01           4.171e-02                  18.74              < 2e-16 ***

Distance                       -9.794e-05           3.139e-05                  -3.12                0.00186 **

Residual standard error: 0.3342 on 962 degrees of freedom

Multiple R-squared:  0.01002,         Adjusted R-squared:  0.008987

We initially questioned whether refugee perception of safety could be accurately correlated with actual physical safety, or whether perception biases (e.g. feeling displaced and identifiable) would enhance the perception of unsafety.  These results are therefore useful in supporting the survey questions’ utility in future research.

  • Distance to police stations does not appear to significantly affect actual theft or the household breadwinners’ actual physical abuse at work. These results are intuitive, since police proximity to home location does not necessarily affect safety at work, unless work locations and home locations are near each other. The results also show that police effectiveness might be higher in protecting refugees from physical than from economic crimes such as theft. Theft is also likely to be more under-reported as compared to physical assault.
  • Being male is associated with having slightly higher overall safety scores; however, this relationship is not statistically significant.  This result is extremely surprising given the rates of violence against women in India generally and in Delhi in particular.  However, there may be alternate explanations; for example, female refugees may not travel as widely and therefore face less discrimination, or they may be unlikely to travel alone or at dangerous hours.  A better understand about gender roles and experiences in Delhi would assist in explaining this result.
  •  The relationship between distance to police stations and physical safety varies across the four ethnic groups. Specifically, the analysis shows that, in fact, Indian households tend to feel slightly safer the further away they are located from police stations. In contrast, the reported safety of Afghan and Somali refugees is notably higher the closer they are located to police stations. Because the overall Safety indicator includes both perceived and actual safety sub-indicators (e.g. perceived neighborhood security as opposed to experiences of actual attacks), the higher reported safety for households living closer to police stations might be due to either subjective perceptions or due to actual police service effectiveness. 

 Figure 10: Relationship between distance to police and safety, by ethnicity

3 Conclusion
Proximity to services did not significantly correlate with higher livelihood scores, which suggests that physical distance to services may not be the most important barrier for urban refugees.  Finances, lack of mobility, or discrimination may play more important roles in determining livelihood index. Isolating elements of the livelihood index—employment security, housing security, financial security, and physical safety—may be a better way of analyzing the importance of physical access to services. For example, isolating the safety element of the livelihood index and analyzing it in relation to proximity to police stations did yield significant results. Refugees who live closer to police stations reported higher perceptions of safety and lower rates of assault. It appears that the positive services provided by the police agency outweigh the negative effects of corruption in the Delhi police agency. Overall, we found that our analysis of spatial pattern was inhibited by the non-random sampling of households.


Future Research

Our ideas for future research can be divided into two branches: 1) further analysis that can be done with the current dataset and 2) further work the UNHCR could pursue based on these results or to otherwise enhance their understanding of urban refugees in Delhi.

Further work that can be done with the current data includes revising the livelihood index and developing a predictive capability for our model. To revise the livelihood indicator, we can change the weights of the indicators and dimensions in the livelihood index or by adding or omitting indicators from the index. To take this a step further, we can try multiple iterations of the livelihood scores, rerun our analysis on each set of scores, and compare the significance of our results to determine the ideal components and weighting for a livelihood index in Delhi, India. We can also focus on each livelihood dimension (physical, financial, housing, employment) to more specifically describe the living conditions of refugees.

With our current dataset, we can also build-in a predictive capability into our model. To do this, we would need to change the basis analysis from yes-or-no questions to an optimality problem. Specifically, we could rank optimal locations for new services or make recommendations about how best to use funds for urban refugees in Delhi. The UNHCR may also have a stake in developing predictive capability as it seeks to support urban refugees in Delhi, as well as refugees world-wide.

Further work that may be done beyond the current dataset include resurveying the population using random sampling and different survey questions and, more broadly, rethinking what livelihood means. To follow up on the JIPS profiling project of urban refugees in Delhi, the UNHCR may wish to do another survey using random sampling techniques so that they can better analyze the data using standard statistical and spatial techniques. They may also wish to perform spatial analysis on another non-clustered dataset and compare the results in order to get a more representative view of refugees in Delhi, India. New questions may also be added to a follow-up survey to better understand some of our key findings. For example, to better understand why refugees feel safer closer to police stations but do not have a higher livelihood index closer to public services, follow up interviews focused on refugee perceptions of local law enforcement and access to public services could illuminate these trends.

        Finally, future work may also involve rethinking our definition of livelihood. Many of the questions in the survey focused on the basic needs of refugees, and basic needs are undeniably important for livelihood, but livelihood may encompass much more. Perhaps livelihood should account for leisure time, access to open spaces, environmental quality and contamination, mental health, or even happiness. There is much more to life than just money, a job, a house, or safety. UNHCR needs to better understand other factors that contribute to livelihood to ensure that refugees have the opportunity to live a happy and healthy life.


Neeraj Chauhan, “Delhi Police, MCD Report Maximum Corruption CasesThe Times Of India – Delhi, 2013.

Chin Human Rights Organization, Waiting on the Margins: An Assessment of the Situation of the Chin Community in Delhi, India (April 2009).

Rob Fiedler, Nadine Schuurman, Jennifer Hyndman, Improving Census-based Socioeconomic GIS for Public Policy: Recent Immigrants Spatially Concentrated Poverty and Housing Need in Vancouver (ACME: 2006).

Indian Census, Administrative Divisions 2011: National Capital Territory of Delhi (2011).

Jeremy W. Crampton, GIS and Geographic Governance; Reconstructing the Chloropleth Map, Cartographica: The International Journal for Geographic Information and Geovisualization, 39: 41-53 (2004).            

Jesuit Refugee Service South Asia, Chin Refugees in Delhi: Realities and Challenges (2013).

JIPS, Urban Profiling of Refugee Situations in Delhi Report (2013).

Anne Moudon, Paul Hess, Catherine Snyder, Kiril Stanilov. Effects of Site Design on Pedestrian Travel in Mixed Medium-Density Environments. Washington State Transportation Center (1997).

National Crime Records Bureau. Crime in India-2012. (Ministry of Home Affairs: 2012). 

National GIS Organization to be Formed, The Hindu Business Line, 4 Dec, 2013.

Planning Department, Government of National Capital Territory of Delhi (2011), "Chapter 21: Poverty Line in Delhi" (PDF). Economic Survey of Delhi, 2005–2006, pp. 227–231.

Donna Schiess Renaud, (2011) "An Analysis of Burmese and Iraqi Resettlement Location and Assimilation in a Midsized City: Implications for Educational and Other Community Leaders." Dissertations. Paper 17.

UNHCR, Statistical Yearbook 2009. 2010.

UNHCR India, Fact Sheet September 2013.


This project was a team effort and would not have been possible without the support of our advisor at JIPS and faculty support at Stanford University.
Stanford University Student Team:
Evie Pless -- MS Student, Biology
Anne Siders --  PhD Student, Emmett Interdisciplinary Program for Environment and Resources
Lauren Steinbaum --  MS Student, Civil and Environmental Engineering
Aiga Stokenberga --  PhD Student, Emmett Interdisciplinary Program for Environment and Resources

United Nations Adviser: 
Assanke Koedam
Information Management Officer 
Joint Internally Displaced Persons Profiling Service (JIPS)

Stanford Professor:
Patricia Carbajales
GIS Lecturer and Geospatial Manager
Stanford University

Sophia Paliza-Carre

Additional Materials

If you are interested in the specifics of our project, please see the documents below. These are the reports we submitted throughout our project and illustrate how the project evolved throughout the semester.  If there are any discrepancies in the approach, please refer to our final report.

SelectionFile type iconFile nameDescriptionSizeRevisionTimeUser
View Download
  1381k v. 1 Dec 7, 2013, 5:30 PM Anne Ronelle Siders
View Download
  695k v. 1 Nov 4, 2013, 9:22 PM Lauren Steinbaum
View Download
  1463k v. 1 Nov 22, 2013, 10:26 PM Lauren Steinbaum
View Download
  667k v. 1 Oct 20, 2013, 1:29 PM Anne Ronelle Siders

View Download
  644k v. 1 Oct 20, 2013, 1:30 PM Anne Ronelle Siders