2017年-世界发展银行全球_Prospects_of_Estimating_Poverty_with_Phone_Surveys___Experimental_Results_from_Serbia_40页_2mb

报告摘要

Summary of "Prospects of Estimating Poverty with Phone Surveys: Experimental Results from Serbia"

Core Content

This paper explores the feasibility of using telephone surveys to estimate poverty and inequality statistics, focusing on the Serbia context. It evaluates the Survey-to-Survey (S2S) imputation method, which allows for the estimation of household expenditure or income from a small set of simple questions, thereby reducing the time and cost associated with traditional face-to-face interviews.

Main Viewpoints

Telephone surveys are a cost-effective and timely method for data collection, but they are not typically used for estimating poverty indicators due to challenges like low response rates, lack of representativeness, and response bias.
The S2S imputation approach is proposed as a viable solution, as it requires only 10–20 simple questions to estimate poverty rates, making it suitable for telephone interviews.
A pilot study was conducted in Serbia, the first of its kind, to test whether the S2S method works with telephone interviews and whether interview mode affects poverty estimates.
The study found that reporting patterns between face-to-face (F2F) and telephone interviews (CATI) did not change significantly, and the bias in poverty estimates due to interview mode was statistically insignificant.
These findings suggest that telephone surveys can be used for poverty monitoring if designed properly, and they offer promising results for countries with limited or no household survey data.

Key Information

Challenges of Telephone Surveys

High attrition and non-response rates: In some cases, over 50% of households may not respond, and these non-responses are not random.
Lack of representativeness: Telephone ownership is skewed toward wealthier households, leading to potential bias in the sample.
Response bias: Telephone interviews may result in different reporting behaviors compared to face-to-face interviews, especially with SMS and IVR methods.

S2S Imputation Method

The method uses regression models to estimate household expenditure or income based on non-consumption variables such as asset ownership, demographics, employment status, and location indicators.
It requires fewer than 20 questions, which can be collected in 5–10 minutes, significantly reducing the burden on interviewees.
The model is developed using past household survey data, such as the 2009 Serbia Household Budget Survey (HBS).

Pilot Study Design

The pilot was conducted in Serbia, where the Statistical Office of the Republic of Serbia (SORS) had an established telephone call center and experience with telephone data collection.
A nationally representative sample was selected, and an official letter was used to inform households about the survey's purpose to reduce non-response rates.
Computer-Assisted Telephone Interviewing (CATI) was used to collect data, as it was found to have less reporting bias than SMS or IVR.
The study compared poverty estimates from the S2S method with those from traditional household surveys.

Results

Poverty estimates from the S2S method were very close to those from the traditional method, indicating the reliability of the approach.
Response patterns between F2F and CATI interviews were similar, suggesting minimal bias due to interview mode.
The study encourages further experiments in other country contexts to confirm the generalizability of the findings.

Conclusion

The study concludes that telephone surveys can be a promising tool for poverty monitoring, especially in countries with limited or irregular household survey data. The S2S imputation method appears to be effective and reliable when implemented with careful sampling, CATI technology, and an official survey framework. However, more research is needed to confirm its applicability in different settings.

Key Authors and Affiliations

Vladan Boznic: Statistical Office of the Republic of Serbia
Roy Katayama, Rodrigo Munoz, Shinya Takamatsu, Nobuo Yoshida: World Bank Group