Survey of Living Standards 2007 and Extension 2008

2007-2008 SURVEY FIELDWORK

The TLSLS was designed to run over a period of a full year in order to better account for any seasonal variation in different indicators. In addition, the fieldwork was designed to be more or less evenly spread throughout the country over the year. The TLSLS was launched on March 27, 2006. However, after about eight weeks of fieldwork, the survey had to be suspended due to the outbreak of conflict in the country.

The survey was resumed on January 9, 2007 and survey operations progressed without interruption since then. Fieldwork for the survey concluded on January 22, 2008.

At the time of the resumption of the survey, a decision was made to revisit the households who were interviewed in 2006 prior to the interruption of the survey. In particular, 351 households had been visited in 2006. Of these, 317 households were revisited during December 2007-January 2008. The remaining 34 households could not be found at the time of the revisits, and instead an additional 41 new households were interviewed as replacement households. In order to maintain a sample for a continuous period of a year, the final TLSLS sample thus excludes the 351 households interviewed in 2006 and instead includes the 358 revisited or replaced households.

Given the challenges of the turbulent political and security situation during some periods in 2007, the fieldwork schedule had on occasion to be slightly modified to accommodate concerns of security and feasibility of fieldwork. Despite this,the distribution of the sample by month of interview and by region and rural and urban areas indicates a sample that is well-spread through the year, which should allay any concerns of intra-year seasonality.

2008 EXTENSION SURVEY

As one of the objectives of the survey was to measure how much the head of the household knows about the financial activities of all household members, and to be able to produce gender disaggregated analytical work, it was decided that concurrent interviews would be conducted with the head of the household and his/her spouse, to prevent them from contaminating the responses of the other. To accomplish this, the interview was conducted in two parts. During the first part, one interviewer would complete the roster and agricultural modules with the household head, or the household member best able to respond to these questions. They would then schedule a time to return to the household to interview the household head and spouse separately. Two separate books were therefore used.

This survey also included the randomized assignment of the justice and vulnerability modules to the household head or spouse. This was done using a randomized number and designation printed on the cover of the questionnaire. Additionally, this designation was used to assign the household either the short (household head only) or long (all household members over the age of 15) finance module.

The pre-printed random number designated the household into an "odd" or "even" category. The randomization was generally well followed with the exception of one team. All of the interviews conducted by this team are included in the "questionable" category. Any gender disaggregated analysis should therefore exclude the "questionable" households and include the "re-interview" households, using sampling weights w2.

All households were asked sections 1 and 2 to the respondent best able to answer. Interviewers asked even households the justice module to the household head, the vulnerability module to the spouse, and the finance module to all household members. Interviewers asked odd households the vulnerability module to the household head, the justice module to the spouse, and the finance module to the household head only.
The division of the questionnaire into household head and spouse was undertaken to guarantee the separation of the two respondents. During field work, while a priority was placed on assigning female interviewers to female respondents and male interviewers to male respondents, this was not always possible due to scheduling conflicts.

To ensure that the correct households were re-surveyed during the TLSLS2-X, the covers of the questionnaires were individually pre-printed with the household location information and the designation of "odd" or "even". Additionally, on the reverse side of the cover, the roster and plot lists from the TLSLS-2 questionnaire were printed as reference.

The roster for the TLSLS2-X listed all current household members. Interviewers compared this information to the pre-printed roster information from the TLSLS2 survey to identify new members. New members were asked a series of questions relating to age, marital status, education, etc, which were skipped for existing members. All members were asked questions related to preventative health care. Interviewers are then asked to compare the new roster with the pre-printed and determine the whereabouts and reason for leaving of all household members in the TLSLS2 survey who were no longer part of the household.

Note: In the case of a female headed household, or a male headed household without a spouse, the procedures changes for the justice and vulnerability modules. The vulnerability module was asked to the most appropriate respondent of the specified gender, while the justice module was skipped. In some cases, a second choice was not available for the vulnerability module and it was also skipped. The actual respondent can be linked to the roster through their personal identification number (pid). Also note that surveys were marked as complete if the interviewer followed the correct procedures, not if all sections are actually completed. Therefore, some surveys marked complete will be missing information for the vulnerability or justice modules.

Organization and Timing of 2008 Extension Survey:

Field work for the TLSLS2-X was carried out by field teams from the Direcção Nacional de Estatística [DNE], with training and supervision provided by the World Bank, and Mekong Economics and Sistemas Integrales, private consulting firms. Each field team consisted for three interviewers, a supervisor, a data entry operator, and a driver. Data entry was concurrent with data collection, and performed in the field using laptop computers.
The questionnaire was developed by the individual topic teams within the World Bank in the fall of 2007. Pilot testing was conducted in January 2008. Unforeseen political events delayed the start of the training until May 2008. Ten days of training was conducted between May 13 and May 23 in Dili. Training was conducted by Sistemas Integrales with assistance from the World Bank, and consisted of both classroom exercises and field training. Field exercises were conducted in Dili, Alieu and Liquica districts.

Field work was originally scheduled for 20 weeks beginning in May and being completed in August. Questions as to the quality of the data arose during the compiling and cleaning process in September 2008. A "spot-check" data quality review mission was conducted in October 2008, and at that time it was determined that further interviews would be necessary in two districts. The additional interviews were conducted in November and December 2008. Data cleaning and compilation took place in January and February 2009, with the finished dataset being released to World Bank team members in February 2009. Plans are still on-going for public dissemination of the data.