Chapter 3 Study background and data
3.3 Source of data and background information about the survey
The Government of Bangladesh considers agricultural development a major priority alongside food and nutrition security. With a view to investigating issues concerning food security and agricultural development, the Bangladesh Policy Research and Strategy Support Programme (PRSSP) was launched in October 2010. It is funded by the United States Agency for International Development (USAID) and implemented by the International Food Policy Research Institute (IFPRI). For the purpose of meeting the data requirements of the IFPRI- PRSSP’s research, the IFPRI researchers designed the Bangladesh Integrated Household Survey (BIHS). This survey also serves as a baseline for the United States (US) Government’s global initiative to address the root causes of poverty, hunger, and under- nutrition (A. U. Ahmed et al., 2013). The IFPRI-PRSSP research plan includes three rounds of the BIHS. The first round of the BIHS, which was conducted in 2011-2012, is used as a reference point to measure progress through repeat surveys. The second round was carried out from January to June in 2015, and the third round is planned to be conducted from November 2017 to March 2018 (A. U. Ahmed, 2016). Since the second round of the BIHS became available for non-IFPRI researchers in December 2016 (Nilam, 2017), by when the majority of this dissertation’s analysis was completed, we are unable to utilise the panel nature of the BIHS.
We obtain our data from the first round of the Bangladesh Integrated Household Survey (BIHS) 2012 (A. U. Ahmed, 2013). It is worth mentioning that the BIHS is the most comprehensive household survey dataset available for Bangladesh to date (A. U. Ahmed et al., 2013), and it has some notable strengths over the other datasets available for Bangladesh.
Two potential alternative datasets include the Bangladesh Demographic and Health Survey (BDHS) and the Household Income and Expenditure Survey (HIES). The former is a particularly rich dataset for health-related information of children e.g. immunisation and anthropometric indicators. Also, it has data on women’s participation in household decision- making relating to domestic affairs. However this dataset has no information on food security and subjective well-being. Thus the BDHS has not been used in our dissertation.
The other available national household survey is the Bangladesh Household Income and Expenditure Survey implemented by the State agency—Bangladesh Bureau of Statistics (BBS)—which collects food consumption data but at household rather than individual level. Moreover it has no information on anthropometric indicators and subjective well-being measures. Most importantly, none of the aforementioned datasets would enable us to measure the empowerment of women engaged in agriculture in a multidimensional framework.
The BIHS, which is used in this thesis, is the only nationwide household survey that has collected data on (i) individual-level dietary intake of all household members; (ii) anthropometric measurement of all household members; (iii) a wide range of empowerment measures from the primary male and primary female member of the same household; and (iv) life satisfaction. In addition, the BIHS carried out a community survey to complement information on area-specific contextual factors. Given its richness, the BIHS stands out to be the most appropriate dataset for this dissertation. However utilising the dataset essentially confines the scope of our analysis to rural households, as the BIHS has covered rural households only.
3.3.1Sampling
The BIHS sample is statistically representative of rural areas of each of the seven administrative divisions of Bangladesh: Barisal, Chittagong, Dhaka, Khulna, Rajshahi, Rangpur, and Sylhet. The survey followed stratified sampling in two stages: first, selecting the primary sampling unit (PSU) i.e. villages; and second, selecting households within each PSU using the sampling frame developed from the community series of the 2001 population census of Bangladesh. The sampling technique required the BIHS to survey 6,503 households from 325 primary sampling units (PSU) across Bangladesh. Figure 3.1 shows the country map of Bangladesh in which each black dot represents a survey area.
Figure 3.1 Distribution of the primary sampling units across Bangladesh
3.3.2Survey instrument
The BIHS used a two-part questionnaire – one part for female respondents and the other for male respondents. For implementing the survey the IFPRI contracted Data Analysis and Technical Assistance (DATA) Limited— a consulting firm in Bangladesh. The firm trained 60 female and 60 male enumerators as well as 3 female and 17 male supervisors to conduct
the survey. In addition, there were 4 female and 6 male editors to edit the completed questionnaire in the field during the survey.
3.3.3Survey administration
The female enumerators interviewed an adult female household member (generally wife of the head of the household) of a household, and the male enumerator interviewed an adult male member of the household who is generally the household head. It took about four hours for an enumerator to complete the questionnaire i.e. eight hours to survey a household. The surveyed household received some financial gift in appreciation of the time given for the survey. The first round of the BIHS survey started on October 26, 2011 and ended on March 15, 2012. On completion of the data entry and cleaning, the DATA delivered the complete dataset to the IFPRI-PRSSP at the end of June 2012 (A. U. Ahmed et al., 2013).
The IFPRI and DATA took extensive care to ensure the quality of the survey data. For instance, at the place of the survey the supervisors oversaw the interviews conducted by the enumerators routinely on a daily basis and verified whether enumerators had completed all the questionnaires. If the supervisors detected any inconsistent responses in the completed questionnaires, they visited the concerned respondents to find out the reasons and corrected the responses as needed. In addition, the supervisors made random checks of about 10 percent of the completed questionnaires by revisiting the sample households. The IFPRI researchers also made frequent field visits to supervise the fieldwork (ibid.).