Hurricanes, cognitive bias and “what’s in a name?” – Gender bias update

Palm trees don't have a cognitive gender bias...

Palm trees don’t have a cognitive gender bias…

Thanks to the Proceedings of the National Academy of Sciences of the United States of America we’ve got a good example of sexist cognitive bias in operation. (See essay “Think you’re not sexist? Think again: A thinking tool for fixing gender bias”). Kiju Jung, a doctoral student in marketing at the university, and marketing professor Sharon Shavitt have published a paper (here) looking at the relative destructive power of hurricanes with male and female names respectively, and found the feminine titled hurricanes are more deadly. Unpicking the research shows a number of areas where a sexist cognitive bias shows itself to be the source of faulty logic and bad decision making in the people faced with weathering the storm.

You can download the data for yourself – it’s not a huge sample (109 Mechanical Turk users) – and it’s relatively complex to work out the results for yourself, but here’s the supporting documentation and the related dataset if you’re into that sort of thing.

The upshot is, this study shows the power of unconscious sexism to alter the rational choices of ordinary people. To explain this, you need to know that hurricane names themselves are assigned from a list of alternating male and female names (assigned by the National Hurricane Centre in Miami). So it’s not likely that the people who give hurricanes their names are exercising bias in their choices (e.g. “This hurricane is a bad one, so I’ll name it after my ex-wife”).

Alas, the data seems weighted towards referencing the feminine names (of which there are more in the sample) however, the attitudes of the general public appear to be massively influenced by the name. People seem to assume masculine named hurricanes will be more severe, and tend to take better precautions than when faced with a feminine named-hurricane. From the data it appears that a female named hurricane averages about 23 deaths, whereas a male named hurricane averages about 15.

What makes this finding really interesting (and demonstrates the power of unconscious gender bias) is the fact the word hurricane is also in the name. Hurricanes and their severity are rated in terms of categories, so we have a gender neutral, scientific measure to compare them. So Hurricane Bob (1985, category 1) and Hurricane Bonnie (1985, also category 1) represent the same level of threat, but Bob killed no-one and 3 people died in Bonnie. This pattern seems to be regular. Basically, it appears that people assume a female hurricane will follow the sexist perception that women are less violent, less dangerous and somehow, more forgiving.

But there’s more cognitive bias at work here than just gender bias though… names have processing power

The problem can also be associated with the process of naming severe storms in the first place. The idea was, originally, that giving a severe storm a human name would make them more memorable, it would help people remember and relate to the risks of a severe weather event. “There’s a hurricane coming” turned into “Hurricane Irene is coming”, which triggers a lot more data to influence associated behaviours necessary to preparing for the task at hand, in the same way “A person is coming to dinner” vs “Mr Richardson my boss is coming to dinner” changes aspects of your decision making relating to that dinner event. By assigning an identity, it allows the person allocated the task to make decisions that will shape the details of what they are being asked to do. Dinner for a generic person is hard to imagine (because we don’t know any generic people, they’re an artificial construct, you never have a complete stranger round for dinner without knowing something about them). If your boss is coming to dinner, you probably try to make a different impression from your buddy or your parents, for example.

In the hurricane example, giving it a name was supposed to help people prepare for a discrete, focused threat presented by an anthropomorphised weather event rather than a more generic kind of weather event. And it worked too, because people did relate to the event differently once it had a name – obviously from these results – but not quite as anticipated. Where the logic of naming hurricanes failed was to disregard that the desired response would be affected by other cognitive decision making biases in the audience, in this case, assuming you would come away better from a category 1 hurricane if it was a woman than a category 1 hurricane if it was a man. The fact they’re BOTH FUCKING HURRICANES and therefore the gender of the nickname they’re given is irrelevant, appears both obvious and yet, at the same time, not.

Names aren’t just names. They become, with memory and experience, means for compressing a whole array of data into a nice little bite sized chunk that makes decision making easier. A dog called Fluffy and a dog called Brutus, before you even see the animals, conjure a whole load of associated information that effects your predictions and decisions about the kind of dog to expect, and the animal’s likely behaviour and disposition. A car called an “Interceptor” or a car called a “Friendy” again, cause your unconscious mind to process a whole load of data to help you predict the car’s appearance, the experience of driving it, and the kind of customer that might buy one… maybe even where you’re likely to see one parked. Your brain uses names as compressed snippets of code to help you make decisions and predict complex sets of interactions between people, places and activities.

When names are combined with gender specific data, this changes how we weight the data we process when making decisions relating to the name data. However, when you’ve grown up in a world where there is a bias in terms of gender role and gender behaviours, that doesn’t make for good, rational decision making. No, it means you’ll see a weather warning for similar weather events and judge, based mostly on unfiltered, unchecked, unscientific bias that the category 2 hurricane called “Floyd” will somehow be more dangerous than the category 2 hurricane called “Sandy” and make yourself significantly more likely to drown or be crushed by debris during Hurricane Sandy as a result.

If ever there was a good reason to try and correct for cognitive bias in your decision making, that’s about as good an example as I can think of. It’s also a good reason to keep reading ManVsBrain, too.