Your example is a poor example because you’re choosing the entire population as your denominator for the per capita statistic, but the actual percentage of the

Your example is a poor example because you’re choosing the entire population as your denominator for the per capita statistic, but the actual percentage of the population engaging in Russian roulette and riding trains is drastically different. Whereas the percentage of the overall population engaging in and/or being affected by driving is almost 100%, so the discrepancy that makes Russian roulette seem safer than trains in your example doesn’t exist when comparing traffic deaths across countries.