在LSMS面板测量中的应用：从纵向研究中提高调查估计的质量

Longitudinal Survey Challenges - Longitudinal surveys face challenges in maintaining accuracy over time due to sample attrition, migration, and sample fatigue, which introduce measurement errors[2] - Sample attrition, caused by deaths and relocations, along with the impact of new populations and migration flows, leads to underrepresentation in surveys[2] - Correct panel survey design and implementation require methods to address these issues at different stages: sampling design, data collection, and estimation[7] Proposed Methodology - The study proposes a weighted sharing method-based estimator that provides more accurate individual-level statistics compared to current estimators used in Uganda's national panel survey[2] - The proposed method shows higher stability when changing samples, particularly in cross-sectional estimates based on transition matrices[2] - The methodology focuses on improving survey quality by addressing sampling design, data collection, and estimation stages[8] Empirical Application - The methodology was experimentally evaluated using data from Uganda's national panel survey and the Living Standards Measurement Study (LSMS)[2] - The study applied the proposed methodology to LSMS-ISA data, using Uganda's national panel survey waves from 2009, 2013, and 2015 as a case study[10] - The empirical evaluation demonstrated the effectiveness of the proposed method in improving the quality of survey estimates[10] Data Collection and Tracking - Data collection improvements include flexible methods like phone interviews to reach non-responsive populations and minimal variable data collection for attritors[7] - Tracking rules are defined to balance the cost of tracking migrants with the potential bias introduced by their exclusion from the sample[29] - Proxy interviews and minimal variable sets are used to collect data from non-respondents and migrants, reducing sample attrition bias[53] LSMS-ISA Survey Overview - LSMS-ISA surveys are nationally representative longitudinal household surveys focusing on the relationship between living standards and agriculture[58] - The surveys track households and individuals over time, incorporating new members like immigrants and newborns to maintain representativeness[61] - Tracking methods vary across LSMS-ISA surveys, with some focusing on households and others on individuals, impacting representativeness and cost[63] Uganda National Panel Survey (UNPS) Case Study - The UNPS is a multi-purpose household panel survey in Uganda, providing data on income dynamics, consumption, and agriculture[71] - The study used UNPS data from 2009/10, 2013/14, and 2015/16 waves, with final datasets including 18,313, 17,377, and 15,905 individuals respectively[72] - The empirical evaluation of UNPS data showed that 45% of individuals from the 2009/10 wave were re-interviewed in 2013/14, with 30% of the sample being new individuals[73]