Survey-based socio-economic data from slums in Bangalore, India

In 2010, an estimated 860 million people were living in slums worldwide, with around 60 million added to the slum population between 2000 and 2010. In 2011, 200 million people in urban Indian households were considered to live in slums. In order to address and create slum development programmes and poverty alleviation methods, it is necessary to understand the needs of these communities. Therefore, we require data with high granularity in the Indian context. Unfortunately, there is a paucity of highly granular data at the level of individual slums. We collected the data presented in this paper in partnership with the slum dwellers in order to overcome the challenges such as validity and efficacy of self reported data. Our survey of Bangalore covered 36 slums across the city. The slums were chosen based on stratification criteria, which included geographical location of the slum, whether the slum was resettled or rehabilitated, notification status of the slum, the size of the slum and the religious profile. This paper describes the relational model of the slum dataset, the variables in the dataset, the variables constructed for analysis and the issues identified with the dataset. The data collected includes around 267,894 data points spread over 242 questions for 1,107 households. The dataset can facilitate interdisciplinary research on spatial and temporal dynamics of urban poverty and well-being in the context of rapid urbanization of cities in developing countries.


| Introduction
The survey of Bangalore slums carried out in the year 2010 covered 36 slums across Bangalore. The slums in Bangalore were stratified based on the following parameters: Age of the Slum (Old,New), Location in the city (Core, Periphery, North, South, East, West); Size of the slum; Land Type (whether the slums are on Public land or Private land); Declaration Status (Declared or Not Declared); Major Linguistic Group (slums that contain a majority of Kannada, Tamil, or Telugu speakers); Major Religious Group (slums that contain a majority of Hindi, Muslim, or Christian populations) and State of Development (Redeveloped slums, Resettled slums, In situ developed and planned slums). In order to ensure the quality of data, we short-listed 36 slums (based on access to the slums) that satisfied the criteria listed above. In this study, we decided to have a 10% sample size within each slum, i.e every 10th house, working from the street layout of the slum was part of the survey (as a formal list of households was unavailable).
A total of 1107 households was surveyed. Each survey had 242 questions (see attached questionnaire) spread across the following themes: demographics, employment, migration, transport, water, housing, loans, health-care, assets, aspirations, and problems.
This survey was part of the Bangalore Urban Poverty Study conducted in 2010 by multiple partners, and available through a Creative Commons license at: http://ngil .cstep.in/projects/urban_poverty_survey.html (Last accessed on 30 September 2013). Key assistance was provided by Mr. Issac Arul Selva and Mr. Issac Amruthraj. This survey was partly funded by the Jamsetji Tata Trust and the Next Generation Infrastructure Foundation.
The verification, authentication, structuring and analysis of this data, and this report on living wages was carried out by Fields of View in collaboration with the International Institute of Information Technology -Bangalore, and was funded by the University of Amsterdam.
The data collected from this survey underwent extensive cleaning, and was stored in a relational database for further analysis. This document describes the relational model of the database, how to use it, the variables in the dataset, the variables constructed for analysis and the issues identified with the dataset.
Chapter 3 desribes the database structure with instructions on using it, and the Enhanced Entity-Relationships models. Chapter 4 describes the additional variables that were constructed for analysis. Chapter 5 describes the documented issues with the data, and their resolution.
Appendix A provides details about the different ration cards provided by the Government of Karnataka. Appendix B provides a guide to reading the EER diagrams. Appendix D describes all the variables in the dataset.

| Setup of the database
The data is packaged as a MySQL .sql file, which is to be extracted into the MySQL server before use 1 .
The minimum requirements for setting up this database are: • A computer running either Linux, Windows or Mac • MySQL server version 5.5+ • MySQL client version 14+, distribution 5.5+ This chapter describes the instructions to install the database on a typical computer where the MySQL server and client are running on the same computer.
The following steps are to be followed to set-up the database: 1. Login to MySQL server: $ mysql -u <rootusername> -p<rootpassword> 2. Create a database: mysql> create database urbanpoverty2010 3. Create a user and grant access: Consider the username to be urbanpovertydatauser and password as urbanpovertydatapwd mysql> grant all on urbanpoverty2010. * to >'urbanpovertydatauser'@'localhost' >identified by 'urbanpovertydatapwd'; 4. Logout of MySQL by pressing Ctrl+C mysql> quit 5. Create the tables and populate the data with the following command: $ mysql -u urbanpovertydatauser -purbanpovertydatapwd mysql> use urbanpoverty2010 mysql> source urbanpoverty2010.sql This will create the necessary tables and populate it with the data. 1 It is assumed that the reader is well acquainted with the basics of MySQL

| Database Structure and Usage
The data collected included 242 questions for 1107 households. This data needed to be stored in a manner which was easily accessible, while maintaining the relationships between the data points.
For data integrity, ease of access, and for the ability to "pull" the data into a software/ tool of the users' choice for analyses, a Relational Model of data storage was chosen. The rationale for choosing a relational model is explained in more detail in the appendix. The entities and relationships within the Relational Model were designed such that the model resembled the survey instrument, while maintaining data integrity. The naming convention, described in Section 3.1, was designed so that people familiar with the survey instrument could easily identify the Relation (the RDBMS table) in which the corresponding data could be found.

Naming Convention
The database follows a standard naming convention across all its tables and columns.

Table names
(a) For tables which store data about multiple questions, the naming convention is <questionnumber_from>_<questionnumber_to> _ThemeOfTheTable For example: i. The table for basic demographics of the household is called 1_5_Household as it contains the responses for questions 1 to 5, and is about the Household.
ii. The table for details about people is called 8_15_Person as it contains the responses for questions 8 to 15.
(b) For tables which store data about one question, the naming convention is <questionnumber>_ThemeOfTheTable For example: i. The (b) For columns which store similar data but about multiple questions, the naming convention is <questionnumber_from>_<questionnumber_to> _questionDescription For example, the column 61_66_71_cost stores the travel cost for travel to work (question number 61), travel to education (question number 66) and travel fr household purposes (question number 71). In the table, these are differentiated by the column "type_of_travel", which could be 1 for work, 2 for education or 3 for household purposes 1 .
3. Code tables: The table names for the code tables follow the following naming convention <NameOfTheCode>.
The column names in these code tables follow the following naming convention <name_of_the_code>_id and <name_of_the_code>_name. For example, the table "TravelMode" stores the various travel modes people use. The column names are "travel_mode_id" and "travel_mode_name" which store the unique id of the mode and the name of the travel mode (Bus, Car, Two wheeler, etc.) respectively.
Certain accepted standard RDBMS naming conventions were followed. Table names are in singular (for example 8_15_Person, and not 8_15_People). Column names are not camel case, but underscore separated (18_occupation_id and not 18_occupationId).
However, table names are camel cased to distinguish them from column names, and for readability. All tables are normalized upto the third normal form (Elmasri & Navathe, 2008).

Enhanced Entity-Relationship diagrams
The database was designed by modelling the data using Enhanced Entity-Relationship (EER) models (Elmasri & Navathe, 2008). The first version was designed using Aquafold Data Studio 2 for PostgreSQL 3 , and the consequent versions were designed for MySQL 4 using MySQL Workbench 5 .
This chapter provides the EER diagrams for this database. Since the database has 260 tables, these tables have been arranged theme-wise. Figure 3.1 is the EER model for the tables relating to the household demographics data. Figure 3.2 is the EER model for employment data, Figure 3.3 for the migration data and Figure 3.4 for the travel data. Figures 3.5, 3.6 and 3.7 are the EER models for the data relating to water supply, cost and availbility. Figures 3.8, 3.9 and 3.10 are the EER models for the housing data. Figure 3.11 is the EER model for income, expenditure and assets. Figure 3.12 is the EER model for the qualitative data on issues, agencies and benefits.
Appendix B provides a guide for reading the EER diagrams. For a detailed explanation of relational database design and EER diagrams, please refer Elmasri et. al (Elmasri & Navathe, 2008) and Ramakrishnan et. al (Ramakrishnan & Gehrke, 2000

| Construction of variables
Along with the existing variables in the data set, more variables were constructed to help in the analysis. This chapter describes in detail the processes followed in constructing these variables, the results from the different iterations, and the final definition of these constructed variables.
The following sections describe these variables which were constructed using the collected data. Section 4.1 describes the process of categorisation of the households based on their composition and on their earning potential. Section 4.2 describes the process of categorisation of the different occupations reported by the respondents into different occupation categories. Section 4.3 provides the definition of Kutcha and Pucca housing, and the reasons for their differences with the National Sample Survey Organisation definitions. Section 4.4 describes how the income data (individual and household income) was categorised into different bins and Section 4.6 describes the process for defining the age groups. Section 4.7 describes the process that was followed to ascertain the age of the slums.

Family Categories
Familes are categorised according to their composition and by their earning potential.

First version of the categorisation
The following was the first version of the family categorisation based on composition: The following was the first version of the family categorisation by their earning potential: 1. Female headed nuclear family with one earning member 2. Female headed nuclear family with multiple earning members 3. Female headed joint family with one earning member 4. Female headed joint family with multiple earning members 5. Other-headed nuclear family with one earning member 6. Other-headed nuclear family with multiple earning members 7. Other-headed joint family with one earning member 8. Other-headed joint family with multiple earning members 9. Single households

Second version of the categorisation
The following concerns arose from the first version of the categorisation: • The above categorisation does not provide a definition for Joint and Nuclear families • The categorisation by composition and categorisation by earning potential do not follow the same categorisation criteria. The categorisation by composition was done with the primary criteria being whether the families were nuclear or joint, whereas the categorisation by earning potential was done with the primary criteria being whether the household was male or female headed.
Based on these, a second version of the categorisation was done. The following are the definitions being used: Nuclear Family A nuclear family consists of a married couple and their dependent children. A nuclear family can have multiple earning members only if both the married members in the family earn. The household is a joint family if the children earn as well.
Joint Family A joint family is one where the household consists of more than one adult (not married to each other) and/ or multiple earning members.
Female headed household Households which have reported a female member as the head of the household, or households which have a female member as the primary earning member, are classfied as Female Headed households.
Male headed household Households which have reported a male member as the head of the household, or households which have a male member as the primary earning member, are classfied as Male Headed households.
The final categorisation is based on female-headness as the first criteria, nuclear/ joint family as the second criteria and the number of earning members as the third criteria.
The final categorisation based on family composition is: 1. Female headed nuclear family 2. Female headed joint family 3. Female, Single 4. Male headed nuclear family 5. Male headed joint family 6. Male, Single The final categorisation based on family composition is: 1. Female headed nuclear family, single earning member 2. Female headed nuclear family, multiple earning members 3. Female headed joint family, single earning member 4. Female headed joint family, multiple earning members 5. Female, Single 6. Male headed nuclear family, single earning member 7. Male headed nuclear family, multiple earning members 8. Male headed joint family, single earning member 9. Male headed joint family, multiple earning members 10. Male, Single Households in which the members had not reported any income would be categorised only be their composition (female-headedness) and not by their earning potential. Also, households which have not reported a head of household would not fall under any of these categories.

Occupation Categories
During the survey, the respondents were asked their occupations, and this was recorded without coding. These occupations were then categorised during data clean-up. The categorisation underwent 7 iterations.
First version The first version of the categorisation is given in Some occupations, however, could not be categorised. For instance, snake charmers did not fall under any of the categories.
Second version It was found that even after incorporating occupations from this study into the existing categorisation, it could not account for a large range of occupations in domestic services and home based manufacturing. Thus, these two new categories were created.
Third version Since Trade and Self-employment were in the same category, the occupations which were reported as self employed (during the survey) were identified for further analysis in the next iterations.
Fourth version Sweepers, paura karmikas, corporation work, and toilet cleaners were moved to manual labour category, from domestic services. Coolies, painters and plumbers now fell under manual labour so as to differentiate between manual labour and construction work.
Fifth version Mesthri was moved to semi-professional category from construction work as this was a supervisory role in construction work. Pigmy was moved from selfemployed category to semi-profesional. "Mysore sandal factory supervisor" was moved to professions. Agarbathi packing was moved to home based manufacturing. Tailor, mobile repair and motor repair were moved to self-employed category. Plumber and painter were moved to construction work from manual labour.
Sixth version A separate category called Sales and Small Enterprise was created, with occupations such as tea powder making, book binding and bakery work. Domestic Services was renamed to Domestic and Cleaning Services to include housemaids (who work at homes) and house keeping (who work as staff at various establishments such as colleges, malls etc.). Manufacturing and repair work were split into two categories as well, as repair work falls under services.
Final version The first category, "Professional and Semi professional" was split into Professional White Collar and Professional Blue Collar as the income differences between the two were significant. The final categories are given in Table 4.2.

Housing
The National Sample Survey Organisation (NSSO, 2001)defines the following housing structures: 1. Pucca structure: A pucca structure is one whose walls and roofs are made of pucca materials such as cement, concrete, oven burnt bricks, hollow cement/ ash bricks, stone, stone blocks, jack boards (cement plastered reeds), iron, zinc or other metal sheets, timber, tiles, slate, corrugated iron, asbestos cement sheet, veneer, plywood, artificial wood of synthetic material and poly vinyl chloride (PVC) material.
2. Kutcha structure: A structure which has walls and roof made of non-pucca materials is regarded as a katcha structure. Non-pucca materials include unburnt bricks, bamboo, mud, grass, leaves, reeds, thatch, etc. Katcha structures can be of the following two types: (a) Unserviceable katcha structure includes all structures with thatch walls and thatch roof i.e. walls made of grass, leaves, reeds, etc. and roof of a similar material and (b) Serviceable katcha structure includes all katcha structures other than unserviceable katcha structures.
3. Semi-pucca structure: A structure which cannot be classified as a pucca or a katcha structure as per definition is a semi-pucca structure. Such a structure will have either the walls or the roof but not both, made of pucca materials.âȂİ However, some households in this survey have both, the walls and the roofs as asbestos sheets. As asbestos walls are not stable and are likey to fall without adequate support, a house ought to be categorised as pucca only if the roof is made of asbestos, but not the walls. Such type of housing cannot be classifed as pucca housing, according to the NSSO classfication.
To address such issues, the definitions of kutcha, pucca and semi-pucca housing was modified as: 1. Pucca structure: A pucca structure is one whose roofs are tiles and walls are wood, G.I/ metal/ asbestos sheets or whose roofs are stone, concrete, and wall is wood, G.I/ metal/ asbestos sheets, burnt brick, stone, concrete, cement bricks.
2. Semi-pucca structure: A semi pucca structure is one whose roofs are tiles, stone, concrete and wall is grass/ thatch/ bamboo, plastic/ polythene, mud/ unburnt brick, wood, G.I/ metal/ asbestos sheets or whose roofs are grass/ thatch/ bamboo, plastic/ polythene and walls are burnt brick, stone, concrete, cement bricks.

Income Categorisation -Individual Income
Out of the data of the 5634 people in the survey, 2049 people reported their income. As seen in Fig 4.1, the income distribution resembled a long-tail distribution. The income categorisation had to be chosen which best represented this distribution.

First version of the categorisation
To identify the intervals and categories of the income, the following process was followed: 1. Clustering to identify possible interval sets: k-means clustering was used on the income vectors to identify clusters and intervals. This clustering algorithm was run three times to identify 3, 4, 5, 6, 7, 8, 9 and 10 clusters, and their respective intervals. This was performed to identify the number of intervals we would require to represent the income distribution.
2. Identifying interval sets which best represent a long-tail distribution: From the results of the clustering, the following interval sets were identified as best representing a long-tail distribution: 3. Narrowing down from the selection to one interval set: The interval set number 3 (Fig 4.4) was chosen as it best represented a long-tail distribution.

Income Categorisation -Household Income
Household income is calculated by calculating the sum of the individual incomes in the household. Similar to individual incomes, the household income distribution resembled a long-tail distribution (Fig 4.6). The income categorisation had to be chosen which best represented this distribution.
1. Clustering to identify possible interval sets: k-means clustering was used on the income vectors to identify clusters and intervals. This clustering algorithm was run three times to identify 3, 4, 5, 6, 7, 8, 9 and 10 clusters, and their respective intervals. This was performed to identify the number of intervals we would require to represent the income distribution.

Age Groups
Age categorisation was done based on the requirement that literacy and occupation data need to be analysed according to age. The first categorisation of age data was: The age group [0,6) was added to identify the outliers in occupation data. This category is excluded from analysis of occupation data.
For the second version of the categorisation, the age group [18,25) and [25,30) were merged to form [18,30). The age group following this are in increments of 10.
Furthermore, it was identified that splitting the age groups from 6 to 14 into two groups [6-10) and [10-14) would not be useful when analysing occupation data, but very important to analyse literacy data. Therefore, people were categorised into two age groups, one for literacy data analysis and one for occupation data analysis.

Age of the slums
Ages of the slums were required to understand its relationship with the facilities and infrastructure available within these slums, and to understand how slum demographics vary based on whether the slums are rehabilitated, non-rehabilitated, declared, old or new.
The first set of ages of the slums was acquired through a personal interview with Mr. Issac Arul Selva of Slum Jagatthu. The slum ages from the first version of the documentation are given in Table 4.3. This data contains information on when the slum came into existence and when the slums got rehabilitated. The case studies carried out during the survey were referred to look for details of the age of the slum. Table 4.4 shows the data obtained through the case studies.   The age of the slums was corroborated based on these multiple versions, and a final version is presented in Table 4.6.

NOTE:
The slum names have been anonymised ("ANON") to protect the identity of slum dwellers.

| Issues and Resolution
While the dataset has undergone multiple stages of clean-up, both before and after entry into the relational database, due to the highly disaggregated nature of the data that was collected, a portion of the data still contains a few issues.  Ration card data from the surveys was inaccurate as the respondents reported "Pink" and "Red" ration cards.
Resolved in the database.
After a detailed review of the types of ration cards, the final list of types of ration cards was created and is currently in use in the database.
3 Slum with slum_id = 7 in the database is Thubarahalli. While data was collected from this slum during the survey, during the data clean-up process, it was ascertained to be unreliable and inaccurate.
Not resolved in the database. This is to be excluded from all analyses.

4
The age field contains 99 as the age of some people.
Resolved in the database.
The age was made NULL.

Issues with the current dataset
Sl.no Issue Resolution 5 Occupations reported by the respondents were categorised into twelve categories. However, some occupations could not be categorised as they did not fall under any of these categories.
Resolved in the database.
These occupations are to be excluded from analyses of occupation data.

6
The travel cost for the people whose only mode of travel was walking, was found to be nonzero. This was made 0 as walking does not have an associated travel cost.

Resolved in the database.
Travel cost for people whose mode of travel is only walking is made NULL.

7
People who have reported their occupation as "Driver" (tempo, car, autorickshaw) have reported their travel cost, time and distance for the duration of their work-day. Since drivers have constant access to a mode of travel, considering this data for analysis might skew the results.
Not resolved in the database. This is to be excluded from all analyses.
All rows with type_of_travel=3 are data relating to travel to household purposes.
9 Some people had reported daily costs of travel, with the mode of transport being a bus.
Resolved in the database. These costs have been multiplied by 26 to represent monthly costs.

Issues with the current dataset
Sl.no Issue Resolution 10 Migration data needs further verification before use. Additionally, the responses to the question of where the individuals migrated from is unreliable and inaccurate.
Not resolved in the database. Certain fields such as 37_location_name require further verification. However, state and district data, and the year of migration is available for use, although it is advised not to use the migration data until full clean-up is complete.

11
The expenditure on water per pot is reported for only four slums. Additionally, the costs reported from Gangodanahalli are too high, and are most likely outliers.
Not resolved in the database.
These costs are to be excluded from analyses. For expendiure related to water, the table 194_207_Expenditure is to be used.

12
Water expenditure was collected on a per-pot, per-month and per-litre basis. This was ascertained to be very inaccurate and unreliable.

Not resolved in the database.
For expenditure related to water, the Partially resolved in the database.
These households are to be excluded from analyses of income, expenditure and family categorisation. They have not been assigned a family category by earning potential. Household identifiers: 11,54,75,110,160,187,199,203,206,277,279,291,344,369,477,534,576,622,634,657,687,717,811,837,879,887,893,927,965,982 16 Some people had reported weekly wages instead of monthly Resolved in the database. This was multipled by 4 to represent monthly costs.

One individual has reported an income of Rs 30000
Not resolved in the database. This has a high probability of being an outlier as it is the only response which is greater than 18000. This is to be excluded from analyses.

18
Households have reported multiple reasons for sending/ receiving remittances. For instance, one household has reported that they received a certain amount for marriage and construction purposes.
Not resolved in the database. These are to be excluded from analyses as it is not possible to identify the split of the amount according to the reason mentioned.

19
A five-year old was reported to earn Rs. 5000.

Resolved in the database.
This was removed from the dataset 20 The data relating to assets is not cleaned. The fields for year of purchase and cost are interchanged in the database. Also, the codes for assets need to be verified before use.
Not resolved in the database. However, this has undergone multiple rounds of cleaning before entry into the database.

A | Ration Card types
Ration cards are provided by the government to the families who are in need of state support. The Government of Karnataka provides the following types of ration cards: 1. Akshaya Scheme (Yellow card)

A.1 Akshaya Scheme (Yellow Card)
This card is provided for the weaker sections, backward class families and for senior citizens. For holders of this card, food grains(rice, wheat, sugar) and kerosine are provided free/at a subsidised rate through government ration depoys, with the help of the central government.
The qualifications for this card are: 1. The annual income in urban areas should be less than Rs. 17,000 2. The applicant must be a resident of Karnataka 3. The applicant must be a resident of slums, non wagers, agricultural workers, who has benifits of ashraya scheme, the families who have been living in slums from a long time.
The following are grounds for disqualification 1. Families with permanent landline telephone connections 2. Any government/private labourer whose income is more than Rs. 1000 3. Families which own a own diesel/ petrol vehicles (except mopeds such as TVS 50/ Luna).
4. Families which have outstanding loans of 1 lakh or more

A.2 Anthyodaya Anna Yojane
This card is for people who are below the poverty line, and for the families who can't afford sufficient meals even for a day. Widows, handicaft workers, weaker sections, mentally retarded, schedule caste, schedule tribes, agricultural labourers and tribal groups are eligible for this card. Table A.2 provides the facilities available through this card.

A.3 APL card
The Above Poverty Line ration card is for families who do not qualify for the fixed qualification criteria of the Akshaya, Annapurna, Anthyodaya ration cards. Table A.3 describes the facilities available through this card.

Fields in the tables
Primary key The key symbol in tables indicates its primary key. Most tables have one column as a primary key, while some tables have a combination of two columns as their primary key (a compound key). In the above example, household_id is the primary key of the table 1_5_Household.

Foreign key
Red diamond symbols indicate that the field is a foreign key. In the above example, "slum_id" is a foreign key in 1_5_Household.

Null fields
Diamond symbols with only an outline indicate that the field is nullable. In the above example, "surveyors_name", "data_entry_by" and "date_of_survey" are some of the nullable fields in 1_5_Household.

Not-null fields
Blue diamond symbols indicate that the field is not nullable. In the above example, "1_type_of_household" is a compulsory fields, and cannot be NULL.

Relationships
Relationships in the schema for this data are one-to-many relationships. These relationships are indicated by the following symbology:

Dotted lines
Dotted lines indicate a non-identiyfing relationship: when the foreign key from the referenced table is not a primary key in the referencing table. In the above example, household_id in 8_15_Person is a foreign key from the table 1_5_Household and is not a primary key in 8_15_Person

Non dotted lines
Non-dotted lines indicate identiyfing relationships: when the foreign key from the referenced table is a key in the referencing table. In the above example, house-hold_id in the table 6_MotherTongue is a foreign key from the table 1_5_Household and is a primary key in 6_MotherTongue.

C | Sample SQL Queries
This chapter presents a few example queries, whose outputs can be used to create contingency tables in any statistical software. The listing "Income category and Educational levels of individuals" will output the count of people (N) first grouped by their income category, then grouped by their highest educational training. The output from this query can be used to create a crosstabulation/ contingency table between income categories of people and their highest educational training.
The listing "Occupation and Occupation benefits" will output the count of people (N) first grouped by their occupation category, then grouped by the benefits they get from their occupation. The output from this query can be used to create a crosstabulation/ contingency table between occupation categories and the benefits these occupations provide.
The listing "Travel mode and Slums" will output the count of people (N) first grouped by the slum they live in, then grouped by the travel modes they use. The output from this query can be used to create a crosstabulation/ contingency table between slums and the travel modes their residents use.

D | Variables
The survey had 242 questions spread across the following themes: Demographics Data on the household details, individual details, and other demographic information.
Employment Data about people's occupation, past occupations, seasonal occupation, employment benefits and self employment.

Migration Data on Migration details of individuals and families in the slums.
Mobility Data on individuals' travel time, cost and distance to work and education, their opinions of the Metro, issues relating to cycling and walking.
Water Data on sources of water used in the summer and other seasons, water availability, distance to water sources, waiting time and expenditure on water.
Housing Data on material of houshold roof, wall and floor, kitchen details, bathroom and toilet details, housing finance details and household size details.
Loans and Remittances Data on loans received and loans provided to other households.
Assets Data on what asset the households own, which year it was purchased, and its price at the time of purchase.
Aspirations, Issues, Agencies and Benefits Qualitative data on the aspirations of the people, their current and most pressing issues and how they resolve them or want to resolve them.
Tables D.1 to D.11 provide a complete list of all the questions asked in the survey, and their mapping on to the database. Chapter 3 describes the database design in detail.