Area frame construction was used to develop the frame, for this survey. It consisted of eight strata, of which those considered for the survey were no. 1 to 4, including: 1) coastal, low and level land (reporting a highly intensive agriculture); 2) river valleys and foothills (reporting an intensive agriculture); 3) upland hills (reporting a diversified agriculture); 4) low mountain areas (reporting a limited agriculture). Strata 5 to 8, including military locations, lakes and other bodies of water and major cities, were not taken into consideration. Weighted segment estimation was the selected method, used as follows: once the area frame had been completed, a three stage process was used to select the sample areas for visiting and interviewing of farmers. At the first stage, a total of 1551 Primary Sampling Units (PSUs) were identified with their geographic boundaries, and measured. At the second stage, PSUs were selected in each of the four strata with probability proportional to size, and then split into segments. At the third stage, two sample segments per PSU were randomly selected without replacement. Subsequently, once the sample segments were identified, all of them were visited to ascertain whether or not they had farming (cropping) activities. All farm households performing these activities in the selected segments were listed and, among them, a sub-sample of 1177 farms was selected for interviews and data collection.
