The surveys were conducted by personal visits to the farmers in the daytime by two interviewers. The interview was face-to-face, participation was entirely voluntary, and farmers were free to deny information without further justification. In practice, no farmer objected and all questionnaires included in the final dataset were complete for the pesticide use data. To protect the rights, dignity, safety, and well-being of the respondents, ethical clearance was sought and issued by the Commission for Ethic of Health study from Dustira Hospital Cimahi, West Java. Each participant received a gift of staple food as compensation, such as instant noodles, coffee/tea, cooking oil, and sugar. The questionnaire forms were filled in by the interviewers. During the survey, interviewers did not only record the respondent answers but also performed a crosschecking to confirm his or her response to avoid misunderstanding, especially regarding the pesticide application practice. For example, farmers were asked to show the interviewer the materials and equipment they used or to demonstrate their pesticide application practices in order to avoid confusion. Interviewers checked the weight percentage or concentration of pesticide from each product, amount of application, and the brand package was also photographed for further reference. Whenever farmers used a container or spraying tank in their pesticide preparation, the container’s or tank’s dimensions or volume were measured.

2.3 Estimation of Pesticide Usage

Equation 1 was applied to calculate the pesticide use (i.e., expressed in active ingredient or a.i.) per year. Throughout the paper, the words pesticide and active ingredient are used as synonyms. We use the term “pesticide brand” to refer to a product of pesticides sold as a specific formulation.
\(Pa=\frac{C\ \times\ V\ \times\ f}{A}\) (1)
Where, Pa  is the annual amount of pesticide usage per hectare (g /ha/year), C is the concentration of the active ingredient in the product (g/l), V is the total spraying volume of pesticide brand (l/application), f is frequency of pesticide application (times/year), and A  is the size of surveyed agricultural area of each individual farmer (ha).
In case the applied pesticide was in solid form, its concentration was expressed as a weight percentage (Equation 2 ).
\(Pa=\frac{\%w\ \times\ W\ \times f\ }{A}\) (2)
With the following new parameters, i.e. %w is weight percentage of a.i. in the pesticide brand (%) and W  is the total weight of the pesticide brand used (g/application).

2.4 Comparing Prescribed vs. Actual Use

For rice, which covered almost 65% of the surveyed area, the prescribed use of pesticide was compared to the actual use. The data on the prescribed use was mostly taken from the Indonesian national guidelines (Directorate of fertilizers and pesticides, 2019). This was done per brand, since prescription instructions are brand-specific. When information on the minimum and maximum prescribed use per hectare were available, these were compared with the actual use. When only prescribed dilution ranges were available, these values were also compared with the actual dilution value from the survey result. In case the brand was not recommended for use on rice, we used the minimum and maximum prescribed use values from other crops.