Using the Global Positioning System (gps) in Household Surveys
2. Using GPS can help understand policy externalities and spillovers
The spatial proximity one of household to another may be directly of interest, particularly for understanding the interactions between actions taken by different households, the role of social networks, and the potential spillovers from policies which treat some households and not others.
One example of interactions between households is the possibility that they learn from one another’s actions. Conley and Udry (2005) study learning in the context of the decision to adopt pineapple in Ghana, and of how much fertilizer to apply to it. They note that the classic identification problem here is that the fact that a farmer is more likely to adopt a new technology soon after his сусідів have done so might just be a consequence of some unobserved variable that is spatially correlated — such as soil types, pests or topographic features – rather than the result of genuine learning. They therefore use GPS to define the geographic сусідів of a given plot to be those within 1 kilometer of the center of the plot, and also collect data on who farmers talk to (informational сусідів). For Controlling the deviation of a farmer’s input from his geographic сусідів, they can then identify learning through the impact of informational сусідів’ choices. 9 Furthermore, they do find evidence of positive spatial correlation in unobserved shocks to the productivity of fertilizer, highlighting the importance of controlling for geographic effects when examining learning.
Another example of using GPS to study learning from сусідів is provided by McKenzie, Gibson and Stillman (2007) who study how negative employment experiences for emigrants affect the expectations of would-be emigrants. These would-be emigrants were all unsuccessful in a random ballot in Tonga that offers an opportunity for ballot winners who obtain employment to move to New Zealand. When interviewed subsequently about their employment (and income) expectations they had moved to New Zealand, the would-be emigrants greatly understated employment rates and incomes compared with the actual outcomes for the emigrants. One factor explaining this understatement is that many ballot winners who moved found that their initial job opening in New Zealand was no longer available, and news of this negative outcome appears to flow back to the would-be emigrants in Tonga. Specifically, if all ballot winning emigrants within a six kilometer circle (based on the GPS measurements) did not take up their initial job in New Zealand, the employment expectations of the ballot losers were lower by 19.6 percentage points.
The standard approach to evaluating the impact of a policy is to compare outcomes for those subject to the policy to outcomes for a comparable group not subject to that policy. However, as Miguel and Kremer (2004) point out, this can give misleading estimates of the effect of a policy when there are externalities. They investigate the impact of a deworming treatment in schools in Kenya which was randomized across schools. Using GPS distances at the level of the school, they control for the number of primary school pupils within a certain distance of the school, and then use the number of treated pupils within this distance to measure health spillovers. They find that naive estimates which fail to take externalities into account would underestimate the program treatment effects, leading to the mistaken conclusion that deworming is not cost-effective.
Miguel and Kremer (2004) use GPS distances at the level of a school, and note that these are subject to some measurement error due to U. S. government using Selective Availability to downgrade GPS accuracy at the time of their survey. Now that more accurate measurement is available, a refinement to their research would be to use GPS locations of the residences of each individual child, which could then be used to construct a child-specific measure of exposure to treated and non-treated children. This would provide more variation in the extent of spillover, which could be used to examine the heterogeneity in treatment effects.
3. Information on the spatial distribution of population and services is essential to understanding access to services
One of the most common uses of GPS information to date in developing countries has been to measure access to infrastructure and social services, particularly health care. For example, Perry and Gessler (2000) use GPS to measure access from communities to primary health care facilities in Andean Bolivia and use this to propose an alternative model of health distribution in the area study.
In addition to providing purely descriptive measures of access, GPS data on distance and travel times can be used to understand barriers to the use of particular services. Entwisle et al. (1997) examine the importance of accessibility to family planning on choice of contraceptive device, and in doing so, demonstrate two advantages of GPS over survey-based measures of access. They note that first data on family planning accessibility is often collected in surveys only for certain or political administrative boundaries, such as whether there is a facility in the village. However, facilities in neighboring administrative units may be closer. Using geo-referenced allows data more flexible specification of boundaries, which are not constrained by administrative definitions. Secondly, they note that reported travel times to health facilities in their survey are often звалили in terms of 30 minute multiples, whereas using GIS gives no злипання, allowing better specification of functional form.
Gibson et al. (2006) examine the use of different financial channels for receiving remittances in Tonga. Transactions costs on money transfers are much higher using Western Union than when the recipient withdraws funds from an Automatic Teller Machine (ATM). There are eight ATMs on the main island of Tongatapu compared to five Western Union branches, so a branches per capita measure of access suggest that would ATMs are more accessible. However, they collect GPS coordinates of the ATMs and Western Union branches, and combine this with village-level population information from the Census and a digitalized road network to measure the share of the population within different travel distances of the two competing financial channels. Figure 1 shows their results. Although Western Union has less branches, they are more dispersed, and cover 97 percent of the population within a 10km travel distance, compared to only 77 percent of the population covered by ATMs within this distance. Figure 1 also illustrates how effective the combination of GPS data collection and mapping software can be at illustrating access in a form accessible for policymakers.
Figure 1: Service Areas for ATMs (left) and Western Union Branches (right)
for Tongatapu, Tonga.
Source: Figure 4 in Gibson et al. (2006)
More recent health applications combine distance with measures of infrastructure quality. Hong, Montana and Mishra (2006) use the 2003 DHS in Egypt to look at the relationship between IUD contraceptive use and the quality of family planning services available. They link each household to the nearest family planning clinic within 10km, and then use detailed DHS survey data to measure the quality of the facility. Rosero-Bixby (2004) uses GPS data on census tracts and locations of health facilities in Costa Rica to assess the extent to which health reforms led to improvements in access, measuring access with a combination of distance and services provided by the facility. He notes that households may not necessarily use the nearest facility, particularly if it is low quality, and by using GIS one can calculate measures such as the density of services that meet a standard quality within a specified radius.
A limitation with the above set of health studies is that they only measure distance at the level of a community, whereas households on opposite sides of a village or town may be closer to different facilities from each other. A second limitation is that distance to health facilities could be correlated with a host of other unmeasured factors, such as poverty, disease environment, and other infrastructure, which could also affect health decisions. A recent innovative experiment by Thornton (2006) in rural Malawi is able to identify the causal impact of distance. She studies the decisions of individuals who had been tested for HIV to attend a voluntary counseling and testing (VCT) center to learn their test results. Using their GPS coordinates, households in villages were grouped into zones, and a location within each zone was randomly selected to place a small portable tent, which served as the temporary counseling center, with the average straight-line distance to the center 2.1 km. As Figure 2 below shows, she finds a strong negative impact of distance on the probability of learning the results of the test, particularly for those living within 1km of the center.
Figure 2: Greater Distance Lowers the Probability of
Accessing VCT Centers in Malawi
Source: Figure 3 from Thornton (2006)
4. Using GPS can improve the collection of household survey data
GPS is also starting to be used to improve the quality and cost-effectiveness of collecting household survey data. These uses occur at several phases of data collection from the development of a sample frame, to quality control, and use for follow-up surveys. More accurate and cost-effective surveying enables researchers to carry out better analysis and provide better evidence-based advice to policymakers.
Representative household surveys require an accurate sample frame. The most common approach involves using a recent Census to select enumeration areas. However, censuses may become outdated during periods of rapid urbanization, and will be of little use in drawing samples in post-conflict countries that haven’t had a census for decades. For example, Afghanistan is planning on completing a census in 2007, its first since 1979 10. while Lalasz (2006) reports 15 countries have not taken a Census since 1990. The traditional solution to this problem is to do area sampling, in which enumerators list all households in a well-defined block, such as a village or an area bounded by certain city streets. Such blocks are largely determined by the convenience in defining them and locating them, and can be expensive to enumerate.
Landry and Shen (2005) show how GPS can be used to do area-based sampling quickly and cheaply, since enumeration areas can be defined in terms of spatial coordinates, and made arbitrarily small. They apply this to the problem of surveying in China, where household registration lists are widely used as sample frames. Widespread migration from rural areas however means that many households are unlikely to be found on these registration lists. They use GPS to survey randomly chosen 54 ? 54 meter squares (approximately square one second), and find that 45 percent of the households reached were not on household registration lists.
However, one potential problem with this approach is that the sample size is not known until data collection has occurred, since the number of households within a spatial block is not known ex ante. Landry and Shen use existing population data to create a rough population model of Beijing, but found that the number of dwellings within their spatial units was four times as large as they had budgeted for, so they only administered their questionnaire to one-fourth of the units. It appears likely that aerial photography will alleviate such problems in the future. For example, Cowen and Jensen (1998) extracted individual dwelling unit information in a 32 census block area in South Carolina from aircraft multispectral data. They found the correlation of dwelling data unit derived from remote sensing with similar data derived from the census to have a correlation of 0.91. As the resolution of satellite imagery continues to improve and fall in price, it appears likely that the combination of remote sensing and spatial sampling will become the standard for constructing sample frames in situations where reliable census or registrar data are not available.
Another example of combining remote sensing and GPS for drawing samples is provided by Kumar (2007) in a survey of 1600 households spread across different air pollution zones in Delhi, India. The study area was partitioned into different strata characterized by air pollution levels (obtained by remote sensing) and proximity to main point sources of air pollution. Random points were then simulated using GIS techniques (weighting by the size of the residential area in each strata) і GPS was then used to navigate to the households located at each selected point. These households were then asked to participate in the survey. This method of creating a frame drawing and a sample should be more efficient than simply imposing a regular grid across the study area, since air pollution is irregularly distributed over space.
Visualization of the locations at which sampling has occurred can provide a useful form of quality control to ensure that interviewers conduct of surveys where they are supposed to, and to check whether any dwellings are inadvertently missed. In 2004, Timor-Leste became the first country to use GPS units to record the locations of all households in their Census. USAID Timor-Leste (2004) reports that survey managers checked the GPS points відвідали by the Census teams against detailed aerial photograph maps, and used this to detect areas in the missed enumeration, sending enumerators back to complete the surveys. Population counts from the Census are used in many countries for a variety of policy purposes, including the division of federal money and for defining political representation. Undercount can be particularly high in developing countries – Lalasz (2006) reports that the 1991 Census is thought to have undercounted Nigeria’s population (officially put at 89 million) by perhaps 20 million people. The use of GPS can help show where such undercount has occurred, and help survey managers reduce it.
Another potential use of GPS is through reducing the cost and time taken to re-locate the same dwelling for follow-up surveys. Two reasons for follow-ups are to allow field managers to check errors made by enumerators, and for the collection of panel data. A recurrent problem in many developing countries is the lack of street addresses, making re-locating the same dwelling time consuming, especially in densely populated urban areas. A pilot study conducted by Dwolatsky et al. (2006) with the aim of tracing patients who left a tuberculosis control program in South Africa shows the potential for using GPS to re-locate dwellings. They compared the time taken to re-find a home given residential addresses with the time with a customized personal digital assistant (PDA) linked to GPS. The time taken to find the dwelling was found to be 20-50 percent less using the PDA/GPS device. The main limitation of this study was that it was a small pilot of only 20 houses, so further experiments are needed to confirm the promising results found here.
When panel surveys attempt the more difficult (but conceptually correct) task tracking of individuals rather than dwellings, GPS can be very useful for tracking people who had previously been co-residents in the same household. For example, the Kagera Health and Development Survey 2004 (KHDS 2004) in Tanzania used GPS to record the locations of 2700 households that contain members who had been in the baseline sample of 900 households first interviewed in 1991-94 (Beegle, de Weerdt and Dercon, 2006). Measures such as how far people have moved from either their baseline village center or from households with members who had been co-residents in the baseline surveys can be related to various socioeconomic characteristics.
Finally, collecting GPS data for households allows the possibility of linking the household data set to other surveys and other datasets. There is considerable option value in doing this, since many potential uses of the data will not be known at the time of the collecting survey.
Короткий опис статті: test pda gps Size: 125.85 Kb.; Using the Global Positioning System (gps) in Household Surveys документи реферати