[GCA10]

Actually, if you gathered all that data, your sample would split pretty quickly into

a) tightly controlled societies (some = totalitarian) where the government has a strong interest in keeping the official number low, and

b) open societies (some with a sensationalist media culture), where news outlets gain the most attention by quoting the "experts" with the highest initial counts.

I submit the 1948 Ashgabat earthquake in what was then the Soviet Union and is now Turkmenistan. Initial reports from Tass (the USSR news agency of the time) talked of a relatively modest 6,000 injuries and 600 children orphaned, without providing an actual death toll. More modern estimates say that 110,000 to 176,000 people died.

By contrast, Japan's earthquake/tsunami of 2011 produced estimates of as many as 18,000 deaths in the weeks immediately after the disaster. The official tally now stands at a lower 15,890.