Data on Demand

What is Data on Demand?

A product of IDinsight, Data on Demand uses quick-turnaround surveys powered by local enumerators, Data on Demand produces trustworthy, data-driven insights for decision-makers.

Our Data on Demand team helps practitioners understand how people think, feel, behave, and live so they can provide services that change lives.

Data on Demand is dramatically faster and cheaper than traditional survey methods, enabling leaders to urgently respond to people’s evolving needs.

How DoD works

Household surveys provide essential information to understand people’s needs and their use of essential services. Traditionally, household survey data has been time-consuming and expensive to collect.

IDinsight’s Data on Demand is transforming household data collection with extraordinarily fast turnaround time. It takes days or weeks, compared to months or years, to obtain high-quality results.

We deploy a network of trained local surveyors and apply cutting-edge technology to process and clean large volumes of data in near real-time.

The start-up costs – to recruit, train, and transport surveyors – are eliminated.

We optimize software applications like SurveyCTO, and use data science and machine learning methods to turn household surveys into invaluable feedback that is timely and affordable feedback for decision-makers.

Clients served:

• We are serving NITI Aayog by tracking socio-economic outcomes for 27 of India’s poorest districts, which has enabled a pioneering competitive federalism governance approach. We are now expanding to 112 districts (out of 741) to facilitate unit-level verification of government administrative data, enable DoD implementation by third-party survey firms, and develop a sustainable model for the government to commission future high-quality sample surveys.
• Since May 2020, we have been working with the Delhi Government as Delhi’s official COVID-19 technical partner. We established an end-to-end patient and facility data system for COVID-19 and supported a range of complementary survey and data analysis efforts. We worked extensively with all levels of government to ensure data accurately reflected the ground reality while generating actionable insights for Delhi’s most senior leaders.
• In 2020, DoD conducted seven rapid response phone surveys to inform the Government of India on COVID-19 knowledge, behavior change, and economic effects. The typical request-to-report turnaround was three weeks, with data collected from 37 high-need districts across eight states in India, which represent 80 million of India’s poorest residents. These results have informed government decisions on economic relief and national behavior change communication strategies.

Expansion plans

IDinsight aims to expand the DoD model within Asia and Africa for both in-person and telephone surveys. We are looking to support the initial investment in recruiting and training local surveyors and tailoring the approach we use in India to other contexts. Once established, we expect the DoD system in a country to yield immediate benefits across both NGO and government data users.

We have a deep commitment to partnering with peer organizations and government statistical agencies. As we expand, we will engage with partners to strengthen existing household survey data collection efforts.

Innovation

DoD relies on technical and operational innovations to reduce time lags and assure accurate data. We have developed, tested, and introduced a range of novel approaches to date, and have extensive plans for future innovations.

These include:

• New ways to obtain and use sampling frames. Sampling frames determine whether survey data will accurately represent a population of interest. In India, we tested the accuracy of voter rolls, which provided more up-to-date and comprehensive population listings than census data. We are now experimenting to see how satellite data can be used for sampling, and whether random digit dialing can produce a representative sample in areas with high rates of cellphone use.
• New ways to recruit, train, and manage surveyors. Household survey data collection is labor-intensive. Our current system uses software we are developing that can be used on- or off-line to recruit, train, manage, and assess the quality and productivity of surveyors. We have experimented with natural language processing on a 100 percent sample of calls to monitor surveyor performance in different languages. This has the potential to dramatically reduce costs even more while enhancing inclusive employment practices, consistency, and quality.
• New ways to ensure unbiased results. Bias can occur in the sampling strategy in a number of ways: using only the dominant language, having only male surveyors, identifying males as the household head by default, asking questions that carry assumptions about gender roles, and/or in data collection methods, like phone surveys, that may produce unrepresentative or incomplete responses. IDinsight is committed to representation through data collection and we invest in testing and refining our methods to eliminate sources of bias. For example, we have done in-depth testing to ensure we are capturing women’s voices in phone surveys specifically.
• New ways to assure data quality. The value of household survey data depends on scrupulous attention to data quality. Our data science team uses the power of machine learning to identify error patterns as data are transmitted, for immediate feedback to survey managers and surveyors. Our tailored software instantly cleans data using transparent and rigorous methods, generating data tables, and visualizations based on the most accurate information.

In all our research and development activities, we are seeking to create tools and knowledge that can be used by other organizations as a means of contributing to the larger household survey field.

If you are interested in learning more about how DoD works, our plans for growth, and how we might work together please contact us.