Are female hurricanes deadlier than male hurricanes?
In an article published in the Proceedings of the National Academy of Sciences based on six behavioral studies and one archival study, Jung, Shavitt, Viswanathan, and Hilbe (2014) asserted that hurricanes were deadlier when they had (more) female names. The results of their archival study claimed to find evidence that this held true for hurricanes that made landfall in the United States between 1950 and 2012.
The criticisms in Jung et al.’s (2014) paper were varied and numerous. In a letter to the editor I pointed out that the results of their archival study were not robust to the inclusion of the one two-way interaction they had omitted from their model. In their rebuttal the authors argued that there was no theoretical rationale for the importance and thus no reason for the inclusion of this two-way interaction in their model. In a letter not published in the PNAS (because the PNAS only allow for one interaction between letter writers and authors), I responded that arguing that there is no theoretical rationale for an interaction effect of the barometric pressure of a storm and its damage toll in estimating its death toll is akin to arguing that there is no theoretical rationale for an interaction effect of body height and body circumference on body weight.
Below you find links to the original article, the letters, the authors’ data file, and a Stata code file to reproduce the authors’ results as they appear in the original article and in my letter to the editors.
Jung et al. (2014) Female hurricane are deadlier than male hurricanes, PNAS
Malter (2014) Female hurricanes are not deadlier than male hurricanes, PNAS
Malter et al. (2014) Reply to Jung et al. 2014, unpublished manuscript