The study used secondary data on household income mainly from two household surveys of the Bangladesh Bureau of Statistics (BBS) including Household Income and Expenditure Survey(HHIES) 2005 (BBS, 2007b), and Household Expenditure Surveys (HHES) 1985-86 (BBS, 1988). It has selected 1985-86 as a the base year because of the availability of data as well as the substantial agricultural trade liberalization in the late 1980s. Similarly, it has selected 2005 as the current year due to the availability of the latest household survey data. Therefore, changes in household income is measured using data of HHES 1985-86 as the base year and data of HHIES 2005 as the current year. The study encountered limitations in the use of secondary data due to a lack of disaggregation. The aggregate data approach uses summaries and thus cuts out much variation, resulting in higher correlations than with disaggregated data. In HHIES 2005, all households were aggregated under 19 income or expenditure groups. For the purpose of regression and poverty analyses, this study overcame this limitation by disaggregating household data into 100 observations using respective household groups’ weight (percentage share) as the basis for disaggregation. For instance, in HHIES 2005, households having income between TK3000 and TK3999 represented 14.87 percent of the total households (BBS, 2007b) and they were disaggregated into 15 observations (households) having similar distance of income between two observations. This disaggregation is based on the assumption that keeping the same average income distance between two observations will not change the original characteristics of the data. The study has also conducted a Data Exploratory Analysis to identify outliers. Two outliers were found in the data set of HHES 1985-86 and these outliers were dropped from this data set. However, no outlier was found with the data set of HHIES 2005. The study also used primary data (Household Survey 2010, conducted by the authors) as complementary to secondary data. It applied a mixed-method research design in primary data collection. Questionnaire and face-to-face interview techniques were used for collecting primary data. A structured survey questionnaire was designed with both closed-ended and open-ended questions. Therefore, the datasets included both quantitative (closed-ended) information through using a closed-ended checklist and qualitative (open-ended) information through interviews with participants. The choice of this method was warranted to achieve the objectives of the study. The household head or a senior person of the household who had access to information of all household members answered this structured interview questionnaire. I conducted this structured interview through asking participants questions and writing their answers. If a participant did not have information about all members of the household, the participant was not requested to participate in the survey. The study used both probability and non-probability sampling methods for field surveys to collect primary data. Using convenience and judgment sampling, non-probability sampling methods (Bartlett-II et al., 2008: 47), it selected Comilla amongst the sixty-four districts of Bangladesh for conducting the field survey. According to the Bangladesh Bureau of Statistics (BBS, 2007a), there are thirteen Upazilas (sub-districts) in the Comilla district. They are: 1) Barura, 2) Brahmanpara, 3) Burichang, 4) Chandina, 5) Chauddagram, 6) Daudkandi, 7) Debidwar, 8) Homna, 9) Comilla Sadar, 10) Laksam, 11) Meghna, 12) Muradnagar, and 13) Nangalkot. The study selected Comilla Sadar Upazila, then Chouara Union from that Upazila and finally Shrimontapur village from that union for conducting the field survey. Based on cluster sampling, the households of the selected village were divided into three clusters (A, B and C) and then, using the random sampling technique, cluster C was selected for the field survey. The study surveyed all 60 households from this cluster. Therefore, the sample size of this survey was 60 households of that village. The details of observations are presented. If a participant did not have information about all members of the household, the participant was not requested to participate in the survey. Therefore, all 60 observations for all questions were found correct/valid and no sample was dropped from the original data set. The study also conducted a Data Exploratory Analysis to identify outliers and no outlier was found in this data set. The study considered rice as the representative of agriculture, thereby, considering changes in the rice price for analyzing the impact of agricultural trade liberalization on the real income of rural households for two main reasons. Firstly, agricultural trade liberalization influenced rice production significantly: agricultural trade liberalization directly impacted on new technology for rice production (such as irrigation, fertilizers, and high-yielding varieties seeds). Secondly, rice is the major agricultural product in Bangladesh, capturing the largest share of the agricultural sector. It accounted for 75 percent of the total crop production value, 63 percent of total crop sales, and 75 percent of total cultivated area of the country in 2005 (Klytchnikova and Diop, 2006: 13). In addition, rice is the staple food in the economy. Therefore, any change in rice production and the price of rice impacts directly on the livelihoods and welfare of most households in the country. The literature review showed that agricultural trade liberalization could produce diverse welfare impacts across rural households. Some households might have experienced benefits and others might have experienced losses. This is because agricultural trade liberalization affects both goods and factor prices, which in turn affect household welfare in different ways, depending on their different characteristics (Nicita, 2009: 19). All rural household groups were divided into five quintiles on the basis of income: 1. Bottom 20 percent (Quintile 1), 2. Lower middle 20 percent (Quintile 2), 3. Middle 20 percent (Quintile 3), 4. Upper middle 20 percent (Quintile 4), and 5. Top 20 percent (Quintile 5). They were classified into two main groups on the basis of their involvement in farming activities, namely: a. Farm households, and b. Non-farm households. Other classifications included: 1. Farmers, who owned farmland, and 2. Agricultural laborers. Farmers were further divided into three sub-groups based on their farm size (as used by the BBS during the Household Income and Expenditure Survey 2005, and Agricultural Sample Survey 2005): a. Small Farmers (0.05-2.49 acres), b. Medium farmers (2.50-7.49 acres), and c. Large farmers (7.5 acres and above). Finally, households were classified on the basis of their participation in the rice market either as 1. Net buyers or 2. Net sellers. The study applied the Deaton methodology to identify net seller and net buyer households. Deaton (1989) formalized the concept of net benefit ratio (NBR), which is a proxy for the net-trading position of a household, to estimate the first-order impacts of price changes on household welfare.