Sexism means that female-named hurricanes are more deadly than male-named ones.
That’s the headline finding from a study published in the Proceedings of the National Academy of Sciences, which analysed 94 hurricanes which made landfall on the contiguous US between 1950 and 2012. The results are striking. Of the most severe hurricanes, those with female names on average killed 45 people, compared to 23 deaths in those with male names.
“[The] model suggests that changing a severe hurricane’s name from Charley to Eloise could nearly triple its death toll,” the study concludes.
(Before we go any further, if you’re reading this with fingers a-twitch because you want to shout-comment “correlation does not imply causation” - please, wait. We know. We all know. It’s going to be dealt with.)
The study - by academics from the University of Illinois at Urbana-Champaign and the University of Arizona - is primarily one about science communication. Meteorologists give hurricanes human names because it makes them more memorable, and it makes it easier to communicate their dangers to the public. If there’s a situation with a Hurricane Tom already making landfall while Hurricanes Dick and Harry are simultaneously forming out over the ocean, it makes it easier to communicate the danger to the public than by pointing at unnamed pressure regions on a map.
There’s an element of tradition, too. At first the names were all taken from the standard Anglo-American Second World War phonetic alphabet (Tango, Delta, etc.), but in 1953 the US National Hurricane Center switched to all-female name lists to avoid confusion with aircraft identification radio communications. Since 1979, the current system of alternating male-female names taken from English, French and Spanish dictionaries was put into place.
One of the study’s strengths is that the authors were insightful enough not to merely rely on “female” and “male” as binary categories, instead recognising that there are degrees of implied femininity and masculinity in all names. Nine people were asked to rank the names of past hurricanes from 1 (very masculine) to 11 (very feminine), and the researchers found that there was a correlation between a name being seen as being more feminine and both a higher cost of the damage caused and larger number of people being killed.
Curious to see if this pattern repeated itself in other ways, they gathered six distinct volunteer groups - three of volunteer students from Urbana-Champaign, three sourced via Amazon’s Mechanical Turk - and asked them a series of questions. When given a list of names and nothing else, male hurricanes were rated as seeming more dangerous; when given information about an incoming hurricane, it still was seen as more dangerous if it was called Alexander than Alexandra. The more masculine a name, the more likely a volunteer would say they’d either evacuate their home voluntarily or comply with a government evacuation order.
Yet asking them directly if they found male or female hurricanes more dangerous gave no preference. There are “gender-based expectations about severity” that the average person on the street has without even realising it, and that, the researchers write, means “important implications for policymakers, media practitioners, and the general public concerning hurricane communication and preparedness”.
It’s a finding that confirms what many people know to be true - that even if an individual believes themselves to be unbiased, or unprejudiced, they’re always affected by social conditioning and make implicit judgements that can be sexist. Intent isn't necessary for prejudice. Americans are, it seems, sexist about hurricanes.
But! We’ve got to be careful - and get ready, correlation-is-not-causation folks, because this is the bit you’ve been waiting for - not to extrapolate too hastily from this.
The best article on this study I’ve read comes from Ed Yong, who quotes a sceptical Jeff Lazo from the US National Center for Atmospheric Research. I recommend you go read it for the full detail (and then read some more of Ed’s stuff, because he’s one of the best science writers out there and his blog is consistently fantastic), but, in short, Lazo has four main objections:
- Hurricanes have been getting, on average, less deadly over time (thanks largely to improved preparedness), and, since all hurricanes from 1953 to 1979 had female names, it makes sense that hurricanes with female names killed, on average, more people.
- The fatality counts include “indirect deaths”, such as someone “ killed by fallen electrical lines in the clean-up after a storm”. What do such deaths have to do with gender, or disaster preparedness?
- It’s unclear if the participants of the six sample groups are representative of communities living in hurricane-threatened areas.
- Gender isn’t a big enough factor to influence disaster preparedness, or at least relative to “social situation, vulnerability, culture, prior experience, sources of information, when the hurricane makes land, and so on”.
In response to the first point, she emphasises that the study looked at feminine/masculine rather than female/male names, and that it helps control for the change in naming practice in 1979.
She writes: “Even during the female-only years, the names differed in degree of femininity (compare two female names: Fern, which is less feminine to Camille, a rather feminine name). Although it is true that if we model the data using only hurricanes since 1979 (n=54) this is too small a sample to obtain a significant interaction, when we model the fatalities of all hurricanes since 1950 using their degree of femininity, the interaction between name-femininity and damage is statistically significant. That is a key result. Specifically, for storms that did a lot of damage, the femininity of their names significantly predicted their death toll.”
Also, indirect and direct deaths due to a storm are often conflated by the National Oceanic and Atmospheric Administration (which provides the figures used in the study), and Shavitt believes that indirect deaths are “appropriately included in the dataset” as they “may reflect preparedness”.
She’s also keen to stress that “we cannot claim (nor did we claim) that gendered naming is more important than the other factors that Lazo mentions”. Think to the study last week which found that “global warming” is a more effective phrase for getting across the dangers of climate change than actually calling it “climate change” - hurricane naming is a similar issue. Shavitt’s probably correct to say that, since the issue of communicating the dangers of weather is so important, it’s worth bearing in mind that “implicit biases represent an understudied factor that makes a difference”.
Unfortunately, this still leaves some rather problematic holes. Since we’re accepting that gender isn’t the only factor at play, it’s difficult to believe that the six groups of volunteers are representative of the kinds of people who live in coastal communities at threat from hurricanes - there are simply too many other social factors at play to be sure we can trust these people as only being biased with regards to gender. It’s not as if university students have a reputation for debating gender issues, after all.
That, too, is without considering cultural changes over time. Who’s to say that a destructive hurricane doesn’t influence perception of how masculine or feminine its name is? Is someone called Katrina considered to be a stronger or more dangerous person in the decade since their namesake made landfall in New Orleans? In 1979 the top US names for babies were Michael and Jennifer; in 2012, Jacob and Sophia. We're not given any way to judge the change in perception of masculinity or femininity in names over the time period of the study.
Furthermore, as at least two people have pointed out - Eric Holthaus at Slate and Harold Brooks of the NOAA in a comment under Yong's piece - the study leaves out hurricanes Katrina (death toll: 1,833) and Audrey (death toll: 416) since they were both so much deadlier than other hurricanes in the NOAA records, and thus could be considered outliers. Yet Hurricane Sandy (death toll: 286) is, arguably, just as much of an outlier. Remove it from the dataset and not only does the pattern of feminine-named storms being more destructive disappear, it reverses.
That's not good. The margins involved mean that any further Sandy-sized hurricane could skew things just as much - this study may have come to a very different conclusion if it was written five years ago, or five years from now. Maybe, instead of a theory that Americans underestimate hurricanes with female names, causing greater damage and higher fatalities, we'd be hearing that Americans are actually sent into a counter-productive panic by aggressive, male-named hurricanes, and end up undermining their own disaster preparedness. The dataset is too small for what the study purports to be. It's flimsy.
However, that doesn't mean that communicating the dangers of weather isn't an important issue, and maybe the National Hurricane Center would be open to giving us some data to really test the name theory out. Let's petition them to alternate storms every year between cute and terrifying: Hurricane Cuddly, Hurricane Polio, Hurricane Hugs'n'Kisses, Hurricane Beastiality, Hurricane Finding A Fiver In The Pocket Of A Jacket You Haven't Worn In Ages, Hurricane Cancer. That's real science, right there.