You have come across an example of an article ("Quiz", rather) that asks participants "What Is More Likely to Kill You? Some of the comparisons are cheetahs (1) to falling out of bed (600); vending machines (2) to volcanoes (200); falls (8.8/100,000) to accidental poisonings (11.6/ 100,000), and icicles (100/ year) to lightning (51/ year).

This article does use a few reputable sources- the falls to accidental poisoning ratios reference the CDC (which is an acceptable source for the American population. Of note, this article was obtained on the internet with the question "What is more likely to kill you?" The article caters to a worldwide audience, not just to Americans. The world hosts many different interfering independent variables that are not considered).

In some cases, the author makes a generalization about (the whole of) what is likely to kill you, by comparing 2 completely different populations.

For example, the comparison between icicles and lightning cites an article in the British newspaper Telegraph regarding Russian icicle fatalities (reported at 5 in one year; not 100 per year as the "quiz" purports) compared with lightning fatality statistics reported by the (US) NOAA. Not only are the sources statistically incompatible (regarding reputability and sampling techniques used), but the separate populations are not further compared for all variables.

For example, statistics are not presented on lightning strikes in Russia, or icicle-related deaths in America (and from an Alaskan, trust me, they occur); nor does the article obtain a sample of any other population in the world. However, the story makes generalizations regarding the larger population from isolated occurrences and samples, and presents flawed "statistics" to a world-wide-web audience.

Speaking of audiences, Wong & Hockenberry et. al (2005) stated that falls and trauma-related accidents kill the most toddlers and pre-schoolers per year; not accidental poisonings as the author Perry maintains (Perry cites the CDC, for adult American deaths). As a geriatric nurse, I also believe that falls kill more elderly people than poisonings.

I'm assuming that the article makes assumptions that the audience is American and middle-aged, judging by the way that "facts" are presented that are relevant to this audience only. It fails to account for the population or audience as a whole. Furthermore, generalizations that are made do not hold true for all age-groups (or even geographical) categories of the "population", or audience.