The study was conducted in close cooperation with the PLDP between September and December 1999 in Nawabganj Sadar and Gomastapur thanas, where the PLDP was first started. Six villages, three in which the project was being implemented (PLDP-adopting), and three in which the PLDP had not yet started (PLDP nonadopting), were selected. All villages had similar geographical characteristics, and the PLDP nonadopting villages were situated at least 1.5 km from the PLDP-adopting villages. In the PLDP-adopting villages, between 6% and 25% of the households were end producers of chicken and eggs. This was the most common enterprise as it required the least technical skills and loans. Census and registration lists used by the NGO, facilitated the selection of the study villages and households. Participants from the PLDP-adopting villages, 35 PLDP-adopting households were selected. Likewise, 35 non-PLDP–adopting households were selected from the nonadopting villages. From each household, one woman,,3 mo pregnant or at least 6 mo postpartum, and one daughter aged 5–12 y, were selected to participate in the study. One criterion for participating in the PLDP is that households must own,6000 m2 of land. Data collection Quantitative data on food intakes of the participants were collected using a 24-h recall method. A structured questionnaire was used to collect data on family composition, socioeconomic status, poultry production, resources, living conditions and food preferences. The questionnaire was pretested and revised before the study. With respect to food preferences, the women were asked to prioritize the first four foods that they would buy if they had more money to spend on food. The foods mentioned were scored from 4 to 1, with the first food mentioned having a score of 4. One investigator, assisted by a specially trained female interpreter, collected all data. Each interview lasted;1.5 h, and all information was given was written down during the interview by the investigator. The heights of the participants were measured with a measuring tape and a ruler. Weights were measured with a calibrated digital bathroom scale to the nearest 100 g, and age was reported by the women. Dietary assessment An open-ended–24 h dietary recall, combined with a weighting of food models representing the recalled portions consumed, was used for collecting semiquantitative data on the food intakes of the participants. The quantity of food eaten was recalled by using cooked rice as a model for rice, thick curries and fried or mashed vegetables. Similarly, water was used to estimate the amounts of drinks, soups and very thin curries. Plates and glasses used by the participants, and the serving spoons and cooking pots used in the households, were used for estimating portion sizes. The recalled portions consumed were weighed using an electronic measuring scale with a precision of 1 g and a capacity of 5,000 g. In addition, a table of weights of raw foods was constructed for common foods bought from the local market. This was used to estimate the amounts of raw foods used in the dishes. The total amount of each cooked dish was estimated using the food models. Data analysis Data from 35 women and 35 girls from the PLDP-adopting households were included in the quantitative analysis, whereas data for three women and two girls from the nonadopting households were excluded. Two women and their daughters were excluded because the land holding of the household exceeded the PLDP inclusion criterion, whereas the third woman had eaten irregularly the day before the interview, due to illness. All data were checked and analyzed by the investigator using SPSS for WINDOWS (version 7.5; SPSS Inc., Chicago, IL). All food items recorded were categorized in one of eight food groups: cereals, nonstaple plant foods, oil and fats, fish, chicken, eggs and milk. Food items such as water, tea leaves, spices, and vitamin and mineral supplements were not included. The contributions of energy and nutrient intakes from foods were calculated using a food composition table of Bangladeshi foods. A factor of 0.36, derived from three cooking trials of local parboiled rice, was used to convert the weight of cooked rice to that of raw rice. Protein intake was adjusted by a factor0.88 for digestibility in rice (8). Micronutrients were adjusted for a loss during cooking, frying and baking . Data for poultry production and food intakes were compared between the PLDP adopters and non-adopters. All variables were tested for normal distribution using the Shapiro-Wilk test. Associations between the PLDP adoption factor and normally distributed variables were analyzed using univariate analysis of variance and the General Linear Model procedure. Associations between the PLDP adoption factor and nonnormal distributed variables were analyzed using the Mann-Whitney U test. Energy and nutrient contributions from food groups,as well as priority scoring of foods, were calculated for the PLDP adopters and non-adopters combined.