This survey has been conducted starting May 2020 to collect data on the impacts of COVID on households' welfare.The data were collected every 6 to 8 weeks with 6 rounds completed as of January 2021.
Myanmar Household High-frequency phone survey (MMR HFPS) is part of the MYANMAR COVID-19 MONITORING effort the World Bank initiated at the beginning of the Covid-19 Pandemic with support from Myanmar Central Statistical Organization (CSO). The MMR COVID-19 Monitoring platform provides regular updates on households’ living conditions, enterprises’ activities and communities' welfare.
The MMR HFPS data were collected nearly monthly. As of February 2021, six rounds of data collection had been conducted on a nationally representative sample of 1500 households, some of which had been interviewed in more than one round. Since the survey was conducted over the phone, it had to be corrected for lack of coverage of households who did not own a phone. The household survey questionnaire had core questions on respondents and household heads’ labor participation and income, farm and non-farm household businesses, food security, coping mechanisms and social assistance. Depending on stakeholders’ interests, the questionnaire adopted a modular approach, which allowed extending it in some rounds to cover health and education questions, and COVID-19 knowledge.
Myanmar COVID-19 Monitoring was generously supported through the Trust Fund for Statistical Capacity Building (TFSCBIII) by the United Kingdom’s Department for International Development, the Government of Korea, and the Department of Foreign Affairs and Trade of Ireland. Additional support was provided by the governments of Australia, Denmark, Finland, and Sweden.
Kind of Data
Sample survey data [ssd]
Unit of Analysis
The household survey questionnaire had core questions on respondents and household heads’ labor participation and income, farm and non-farm household businesses, food security, coping mechanisms and social assistance. In some rounds, the survey looked at households' access to health and to educational services.
The universe for this survey is the whole population in Myanmar. The sample frame used was an existing list of phone numbers provided by the firm collecting the data.
Producers and sponsors
Central Statistical Organization
Trust Fund for statistical capacity building
Myanmar Multi-donor Trust fund
GFF Myanmar HFPS Data Collection Support Grant
The sampling frame is a list of all units of analysis within a population. In the present survey, the sample frame is all households (the unit of analysis) within the population of Myanmar. For list-based sampling, samples should be selected with simple random sampling, explicitly or implicitly stratified.
For the HFPS, two sample frames are under consideration: (1) a database of 500,000 phone respondents built by the consultancy firm implementing the survey and collected on a monthly basis to have; and (2) a clustered sample frame from Myanmar Living Conditions Survey 2017 (MLCS 2017) with names of household heads and phone numbers from 13,730 households.
The final sample size has been dictated by the available budget although sample size requirement depends on analytical objectives. We are interested in measuring changes in employment opportunities and food security when surveying households. The number of observations to detect changes over time may be more than those usually required for reliable for point estimates. We have assumed that budget would be available to collect data on 1500 households to have sufficiently precise estimates. Each round will target 1500 interviews. If the respondent fails to carry forward from round 1, then they should be replaced.
The HFPS could follow a simple random sampling using the frame provided by the firm. This consists of randomly selecting the 1,500 respondents from the list frame provided by the firm which is not clustered.
One shortcoming of the COVID-19 HFPS is its lack of national representativeness in key statistics. People who respond to phone interviews may have systematically different characteristics as compared to people who do not respond to phone interviews. Many poor households or those living in rural areas do not have a phone, while most rich households or those in urban areas do. Since phone ownership is essential for phone interviews, an unbalanced distribution of phone ownership makes the collection of nationally representative data challenging because responses are often not uniform.
To address these sampling limitations of a phone survey, we adjust sampling weights so that weighted averages of key statistics become nationally representative. The reweighting process has two major steps: (i) Propensity Score Weighting and (ii) Maxentropy or raking.
Propensity Score Weighting (PSW) is designed to adjust a phone survey's sampling weights by comparing a nationally representative household survey, called a reference survey, with a phone survey. PSW appends the phone survey to the reference and estimates each household's probability of being included in the phone survey. PSW then ranks all households in the phone survey data based on the predicted probability and creates quintiles. The weights of households in the phone survey are adjusted so that each quintile’s share of households in the phone survey exactly resembles that of the reference survey. More specifically, the weights of households in the phone survey are adjusted so that the sum of their weights in each quintile becomes identical to that of households in the reference survey.
To refine the weights further, we execute maxentropy. Even after PSW, summary statistics in the phone survey could differ largely from those in the reference survey. Such differences can be real, particularly when a long time has passed between the reference and phone surveys. Still, it is unlikely that summary statistics of time-invariant (slowly changing) indicators like household size, dependency ratios, household head’s education attainment, or population shares of districts would change significantly. Maxentropy adjusts weights to match the summary statistics of these time-invariant variables between the reference and phone survey in an exact manner.
Dates of Data Collection
Data Collection Mode
Computer Assisted Telephone Interview [cati]
Disclaimer and copyrights
The user of the data acknowledges that the original collector of the data, the authorized distributor of the data, and the relevant funding agency bear no responsibility for use of the data or for interpretations or inferences based upon such uses