REPORT ON INVESTIGATED KILLINGS IN RELATION TO THE ANTI-ILLEGAL DRUG CAMPAIGN 1800 1557 1600 1400 1200 1000 800 600 504 571 661 400 279 233 2020 2021 200 0 2016 2017 2018 2019 Graph 2. Number of investigations of alleged drug-related cases initiated per year The Commission conducted 3,805 investigations of alleged drug-related killings from 2016 through 2021. This number reflects cases which have just been initiated and have yet to be assigned docket numbers. One thousand five hundred fifty-seven (1,557) investigations were launched in 2019, which represents 41% of total investigations on alleged drug-related killings in the period. In 2016, 504 cases were initiated while 571, 661, 279, and 233 cases were opened in 2017, 2018, 2020, and 2021 respectively. 2.2. Methodology The records and reports collected from the CHR Regional Offices and the Investigation Office varied in form, structure, and content. The initial data processing entailed the cataloguing of available records and sorting data that were incomplete, duplicated, or inconsistently formatted. Thereafter, the data was compiled and coded to build a consolidated database that serves as the database of all available information for analysis. Sampling was used to minimize other errors in analysis as this would allow more indepth examination of the selected case dockets given the time constraints in developing an analysis report. For determining the sample size and scope of the analysis, the master list from the Investigation Office was used as basis. As mentioned in Part 2.1, investigations for 2,053 case dockets have already been closed. This served as the population in determining the sample size of the cases to be analyzed. Cases closed for investigation contain the necessary documents and information that can yield the most data to be used for analysis. A probability sample of size 882 case dockets that yields estimates with 98% confidence level and three percent margin of error was obtained. Case dockets were selected using stratified sampling, with regions as strata. Number of case dockets sampled were allocated to each region proportionally. 4