Overcoming resource constraints with online mapping tools
An important part of most Oxford degrees, both undergraduate and postgraduate, is the opportunity to carry out individual, original research. Yet students are often constrained by the (limited) resources available to them. For researchers in disciplines such as Geography and International Development, financial and time restrictions can render survey research almost entirely infeasible. In particular, household registers and administrative data for low- and middle-income countries, especially rural areas, are notoriously difficult to obtain.
In order to save cost and manpower without compromising the quality of his project in rural China (Gansu), Marco Haenssgen developed an innovative approach to sampling villages and households using Google Maps and their geographical coordinates. This methodology has since been shared with other young researchers in the field to support their studies.
How the method was first used
At the start of the project, the only administrative data available to Marco was a complete list of villages in his three selected districts (approximately 2,000 villages in total). Since he did not have the resources to physically go out and list each building in these areas, he carried out the following tasks as part of his unique method:
- verified the location of select villages through Google Maps (using their Chinese names);
- extracted the geographical coordinates of these sites;
- stratified the sample according to the village distance to the nearest township (16 villages plus 32 replacement villages were selected in this way); and
- used satellite imagery from Google Maps and Bing Maps to establish the sampling frame* for each village.
In this case, the chosen sampling frame consisted of maps containing up to 950 numbered and labelled houses for each village, which he produced by extracting and collating highest-resolution aerial images (5 to 40 images per village). He segmented the villages, some of which were highly dispersed, in order to ensure better spatial representation of the households. Once the houses had been stratified by segments, Marco selected individual households using a systematic random sampling method based on the household number per segment, a random starting point and a fixed interval. Through this method, he selected 25 households per village.
Marco also used printed maps, compasses, smartphone map applications, and handheld GPS units to help:
- approach the village via the fastest route;
- brief the field investigators about their selected households;
- locate the selected houses in the village; and
- verify afterwards whether the investigators had indeed interviewed members of the correct household.
A practical, cost-effective method to facilitate research
This effective approach can be replicated in a multitude of contexts where resources for household listing are limited, sampling frames cannot be produced from administrative data, and residential structures are homogenous and distinctive. As a result, student researchers can save substantial expenses that would otherwise be required for transportation, accommodation, subsistence, and insurance expenditures. For Marco, the use of this strategy helped reduce the workload of his research team in China by at least 64 to 80 person-days and consequently, saved the project approximately £4,500.
The importance of this work has been recognised in several ways. For example, it has been taught to DPhil students in Trinity Term as a means to extend their methodological toolbox . It was also presented at the 6th Conference of the European Survey Research Association in July 2015, and will likely result in a journal publication to reach a broader audience.
Top tips for following in Marco’s footsteps
Marco offers three key strategies to help student researchers maximise their gains from collecting data in this way:
- Before using this methodology, you should have completed basic training in survey sampling and be comfortable working with maps.
- Make sure you have the correct equipment, including a laptop with Microsoft Office and a VPN client (e.g. to access Google Maps), a colour printer to produce maps, a smartphone to navigate within the field, and compasses and GPS units to verify specific households.
- It is wise to sample more house than you actually need in case some of the houses selected from satellite maps turn out to be abandoned or non-residential.
- i.e. a list of all those within a population who can be sampled, and may include individuals, households or institutions.
- Read other examples of student innovation.