2

In 2019 a research paper was published measuring the association between cougar attacks on humans using the Fisher's Exact test between 1924 to 2018 [1]. There were a total of 74 attacks reported from 16 different states."The most common activities victims were doing at the time of attack were hiking (n=20, 35%), performing peridomestic activities (n=9, 16%), hunting (n=8, 16%), and jogging (n=5, 9%)." and "Half of all the victims were alone when they were attacked: 100% of joggers, 56% of persons engaged in peridomestic activities, and 38% of hikers were alone at the time of the attack, although these percentages were not statistically significant (P=0.3)" (pg 245 paragraph 1).

The paper stated that those events were primarily due to predatory reasons and a main assumption for the Fisher's exact test requires that the margins are fixed. Wouldn't the research need to include other predatory animals as well in their sample to meet the main assumption?

References

  1. Wang YY, Weiser TG, Forrester JD. Cougar (Puma concolor) Injury in the United States. Wilderness Environ Med 2019;30:244-250. Accessed December, 2021
cn838
  • 19
  • 1
    The wording around the percentages seems awkward to me. For example, "100 % of joggers" sounds like all joggers are attacked by cougars. I know I have been jogging (in the woods) and I have never been attacked by a cougar. Can you clarify the conditioning of those frequencies, please? – Galen Mar 24 '22 at 15:56
  • The OP quoted the paper verbatim when presenting the proportions attacked alone. However, he failed to provide the denominators. These are stated in the paper as "The most common activities victims were doing at the time of attack were hiking (n=20, 35%), performing peridomestic activities (n=9, 16%), hunting (n=8, 16%), and jogging (n=5, 9%)." – Toffee-nosed Hypothesis Mar 24 '22 at 21:44
  • "a main assumption for the The Fishers Exact test requires that the margins are fixed" -- this is a commonly seen statement but it's an error. It's only necessary that you condition on the margins (which may be perfectly reasonable in a number of contexts), not that they must be inherently fixed. – Glen_b Mar 24 '22 at 23:15
  • @Glen_b please correct me if i am wrong. wouldnt make more sense to asses prey selection based on activities, which can be generalized between ungulate and human populations? ref:Cougar Kill Rates and Prey Selection in a Multiple-Prey System in Northeast Oregon – cn838 Mar 26 '22 at 04:22
  • I'm not sure I correctly follow the point you're making. If you're saying that it makes sense to either condition or not condition on the margins, or if you're saying that it makes sense to require the marginal totals to be fixed (or not to require it), please note that my comment was purely about what it is necessary to assume for Fisher's exact test (disputing the direct claim), not what it would make sense to assume in this case. – Glen_b Mar 26 '22 at 06:30

0 Answers0