Structured questionnaire and Global Positioning System (GPS) were used to collect data from the farmers of Narsingdi district, an intensive vegetable growing region of Bangladesh. By following a multistage sampling technique, twelve villages in Belabo upazila (sub-district) of Narsingdi district were selected for the survey. These villages were selected based on vegetable cultivation intensity, frequency of pesticide application and the existence of IPM program. A total number of 1926 vegetable farmers of the selected villages were obtained from the respective upazila agriculture office considered as population of the study. By considering time and other resources, a total of 331 vegetable farmers were selected as sample, comprising 17% from the total population. These farmers were selected randomly for the interview.
The vegetable farmers nineteen characteristics under five broad categories namely; social, economic, institutional, management and spatial were considered in the study. An in-depth interview based on structured questionnaire was held with the sample farmers to collect the data related to non-spatial factors or characteristics. The questionnaire was divided into four parts. The first part was related to farmers’ social characteristics measured by the variables of age, education, household size and perception towards IPM. The second part highlighted farmers’ economic characteristics in terms of the variables such as farm size, area under vegetable cultivation, annual income and land ownership status. The third part evaluated farmers’ participation towards institutional support like farmer field school training, contact with extension agent, field day demonstration and membership of IPM club. The fourth and last part measured some of the farmers managerial activities such as the use of improved variety, time spent on the farm per day and the number or type of vegetable grown in a year. On the other hand, spatial data were collected by using a handheld GPS. GPS coordinates were recorded from each of the survey respondent’s place of residence and four key locations such as upazila agriculture office, nearest market, pesticide store and national highway. Later by using Arc GIS (version 9.3) software, the distance between the farmers’ house and these four focal points were measured.
The farmers’ nineteen characteristics were treated as independent variables or predictors in the study. Some of these are binary while others are continuous. The dependent variable adoption of IPM is binary in nature. The type and Map of Belabo upazila, Narsingdi district, Bangladesh showing vegetable growers’ house and four key locations (agriculture office, market, pesticide store and highway). measuring techniques of both independent and dependent variable. As the dependent variable may have any one option of the two; either the adoption or rejection, then to justify the impact of independent variables on IPM adoption, binary logistic regression model is considered the most appropriate. The model was applied in the study twice. In the first stage, as independent variables, social, economic, institutional and management factors were considered. In the second stage, spatial factors in addition to these were entered. That means the difference between two stages of the logistic regression model is spatial factors where in the first stage these factors are absent and present in the second stage. This was done for better understanding about the effect of spatial factors in the adoption of IPM.
Prior to entering the independent variables into the model, they were checked to identify the colinearity problem. In the first stage, there was no colinearity among the independent variables. As a result all (15) factors were entered in the model. In the second stage, the independent variable (distance of the farmers’ house to national highway) was excluded from the model as it shows a high degree of correlation with another independent variable (distance to agriculture office) and a low degree of correlation with dependent variable. Therefore, eighteen independent variables were finally considered to be included into the second stage of the model.