2.1. Profile of the study area The study was conducted in Patharghata Upazila (subdistrict, administrative entity) of Barguna District in southwestern Bangladesh. Barguna District is a part of the Barisal Division—a coast district in Bangladesh, with an area of 1831.31 sq. km. of which 399.74 sq. km. was riverine and 97.18 sq. km. was under forest. It lies between 211480 and 221290N and between 891520 and 901220E. Barguna is bounded on the north by Barisal and Patuakhali Districts, on the east by Patuakhali District, on the south by the Bay of Bengal and on the west by Pirojpur and Khulna Districts. The study area—Patharghata Upazila—occupies an area of 387.36 sq. km. with 37.29 sq. km. of forest. It is located between 211580 and 221140N and between 891530 and 901050E. The main rivers of the area are Bishkhali and Haringhata. Patharghata Upazila comprises of seven Unions (the lowest administrative entity); among those Patharghata Union was chosen as the study area as the union is situated on the coastline of southern Bangladesh. Various disasters like cyclone, storm surge, coastal floods and erosion, drought, salinity intrusion, etc. are common which affect the area frequently and severely.
2.2. Methods Data were obtained through a series of field visits using survey and various participatory approaches. Documents, studies and reports related to climate change induced disasters—and women role in disaster management were reviewed and necessary data were collected.
2.2.1. Selection of the studied villages Based on the preliminary idea gathered from the secondary sources and consultation with the local NGOs' officials and persons involved in administrations, out of nine wards three wards (spread over four villages) were selected for survey. From the three wards three villages namely Boro Tengra, Char Padma and Char Lathimara were selected purposively for the survey, which were reportedly most affected by the natural disasters. The occupations of these villagers were fishing, dry fish processing, crop production, seasonal vegetables cultivation, day labouring, rickshaw, rickshaw-van pulling and motorcycle driving; shop-keeping and service-providing.
2.2.2. Field survey techniques and data collection Primary information were collected by using checklist and semi-structured questionnaire along with field observations to know people's views about the disaster risk, women's vulnerability, response, identify the similarities and differences in gender perceptions about disaster response, etc. Under FGDs and KIIs poor men and women, representatives from local government, elites and NGO officials were interviewed. Public consultations were carried out at the local tea stalls where local people gathered spontaneously in the afternoon after their daily works. During the HHs survey, household head (male or female) along with family members (if possible) were interviewed during the daytime. The HHs survey and KIIs output were validated during the FGDs. In order to achieve the objectives, 105 HHs, ten KIIs, four FGDs, six case studies (two from each village) and a series of public consultations were carried out (Table 1). Surveys were conducted in the month of February to April 2013.
2.2.3. Data analysis Collected data were processed and analyzed using computer aided programs like MS Excel and SPSS (Statistical Package for Social Sciences, version 17.0). Before reaching to any conclusion different variables were checked and crosschecked; and then processed data were analyzed using both qualitative and quantitative approaches. A five point assessment scale (i.e. 5 for very high, 4 for high, 3 for moderate, 2 for low and 1 for very low) for problem matrix was developed which was used by the FGD participants. At the same time problem frequency was collected; and score was calculated by multiplying the problem values with problem frequency Women vulnerability and capacity analysis was assessed through Anderson's approach. Chi-squared tests (χ2 ) with a 5% level of significance were used to examine whether respondents' responses were uniformly distributed or more leaned toward particular categories of answers. Cramer's phi (φc ) statistic was used as a measure of strength of association between rows and columns in contingency tables and applied to goodness of fit of chi-squared test giving a value between 0 and þ1.