A thesis submitted in partial fulfillment of the requirements for the degree of Master of Engineering in Transportation Engineering

Thesis (M.Eng.) - Asian Institute of Technology, 2021

The national accident data records a large number of motorcycle accident casualties from all over the country, so it is good at showing the number and characteristics of casualties for the entire country. However, it records in just a few details and also suffers from the under-reporting problem. In contrast, the in-depth accident data records only a small number of motorcycle accident casualties but in a lot more details. Thus, it is useful at being used to study the details of motorcycle accident casualties. Nevertheless, the coverage of the in-depth accident data is still doubted. Therefore, before using the in-depth accident data to represent motorcycle accident casualties at the national level, an assessment of how representative the in-depth accident data is must be first made. Consequently, in this thesis, the capture-recapture method was used to examine the under-reporting in the national accident data and found out that 69% and 70% of fatal and non-fatal casualties were under-reported, respectively. After that, the assessment of the representativeness of the in-depth accident data was made by comparing it with the national accident data. The result showed that some variables were biased. Thereby, the post-stratification and IPF methods were applied to compensate for the bias and improve the representativeness of the in-depth accident data.