[Jun8]

"...even though sample size is small it rings 100% true"

I just want to point attention to the (perhaps ironic) fact this sort of reasoning is exactly the same one used by the (allegedly) racist hosts and drivers in the story, i.e. people are much more prone to accept what they are told (mostly in sample size = 1 cases or anecdotal generalizations) if it aligns well with their beliefs (e.g. "another BnB host told me a black guy thrashed his house", "we all know that girls can't code").

A (very) rough characterization of the process, I think, would be:

1. Creation: For this or that reason an a priori belief is formed, e.g. "world is inherently racist" or "blacks are much more probable to be criminals".

2. Filtering: The facts that are reported are filtered using a selection bias (usually, mostly subconsciously) so that facts that strongly agree with the belief are remembered more. (The fact that news items generally report on low-probability, high-standard deviation items, a la man biting the dog, makes this effect even stronger)

3. Update The belief is then updated by the facts with their relative weights determined in (2)

The above (well known and documented, e.g. Blink EDIT: Sorry, wrong reference, see below) process is not a bad thing! AFAIK, it's default brain operation. It takes quite a bit of control and patience to push back the default process at all thee levels.

Note that I'm not arguing that racism, etc. does not exist; however, before jumping to conclusions, like the OP and his/her friend did ("AirBnB doesn't work for black people"), we need to be a bit more careful.

[jfc]

Speaking of irony...

Doesn't Blink posit that a person's extensive experience in an area can lead them to a quick, gut reaction that is, in fact, correct, even before they are able to verify it?

It seems that Andrew's friend probably has had a fair amount of experience with being mistreated due to his skin color. His reaction to being denied a room 3 out of 3 times was perhaps hyperbolic, but certainly applicable to his experience with the service. (Andrew noted his friend's interactions with the driver, so it wasn't all his friend's imagination).

You could use the ideas noted in Blink to look at this scenario from a variety of perspectives, not just confirmation bias.

[Jun8]

Actually I meant Thinking Fast and Slow but somehow typed Blink, sorry about that.

There are two different kinds of generalization here, I think:

As you state, "Brandon" probably has had to endure many other forms of racist behavior, e.g. taxis not wanting to pick up black clients, so his generalization to the new domain Uber was perhaps justifiable, but how about generalizing to AirBnb. AirBnB host population may be very different than the general population "Brandon was used to deal with" (I don't have data to back this up, hence the "may be", but I think that is a very high probability hypothesis, esp. in SF) yet he was quick to generalize, although the refusals may have been due to other factors. In other words "Brandon" has reliable data from a different population that he is now attaching to a new population.

The OP and greendata, OTOH, are generalizing to various beliefs based on a single data point of "Brandon".

[zorpner]

Exactly correct. Additionally, his friend had previously stated that AirBnB didn't work for him, so this "small sample" we're looking at is just Andrew's experience with the issue at this point in time.

It's always disturbing to me how much commentary in threads of this type on HN is devoted to asserting that the individual who's actually experiencing the problem is misinterpreting it. I love data, but demands for data/citations in situations where the problem is experiential are too often used to dismiss people's real, upsetting experiences with privilege and prejudice.

[Jun8]

"... asserting that the individual who's actually experiencing the problem is misinterpreting it"

Note that may main point was about the generalization of the top level comment in this thread, based on this post only, this doesn't mean that the statement in that comment was not correct.

[greendata]

My point was the sample size was small but the "experiment" here is very similar to one seen with taxi cabs and other older services that are very similar to the new online "sharing economy". The discrimination in the former businesses is well documented and studied. I'm not basing my statement of "it rings true" on the data points alone as you are implying but on the large body of similar work.

"it rings 100% true" does not necessarily imply any conclusion. It just means that the initial data on the new sharing economy are inline with other older experiments which study almost the same thing (taxi cabs, cleaning services, etc).

I'm not sure I trust Malcolm Gladwell as a reliable source any more.

[pessimizer]

It should be easy to grow the sample size. Do you know any black people, and have they used AirBnB or Uber?

[tw268]

What do you call a stereotype when it is backed by statistical data? The a priori belief that is most common is that no stereotypes are true. It's the polite thing to believe. But if you are of a certain analytical mindset, you can be argued out of it with enough data and anecdotal evidence from friends.

[Jun8]

An oxymoron :-), i.e. stereotypes are by definition not backed by reliable data (http://www.merriam-webster.com/dictionary/stereotype)

[jack-r-abbit]

Not to split hairs... but your link defines Stereotype (noun): an often unfair and untrue belief that many people have about all people or things with a particular characteristic

Note the word often which does not mean always.

Often (adverb): many times : on many occasions

Always (adverb): at all times

So a stereotype, by definition, may be true... though often is not.

[omonra]

What do you call an accurate assessment of a group that may be unflattering or unpleasant?

And what if said assessment is correct in 90% of the cases - how do the 10% for whom it fails would feel?

[tw268]

Black people really do commit a lot of crime in America, it's not just a "stereotype". And particularly, young black people, and particularly young black males. Black males between the ages of 14 and 24 commit 27% of all murders, despite being 1% of the population.

It's not scrawny white twitter engineers who are shooting each other in San Francisco

[csense]

It's not entirely race. I'm certain that e.g. poverty and lack of opportunity play a role as well. Growing up without access to money, good schools, or positive role models can be crippling to success regardless of race.

[tw268]

Poor whites and Hispanics aren't nearly as dangerous as poor (young, male) blacks. There is not another comparable group that manages to commit 27% of murders as 1% of the population. That's a huge ratio.

Keep in mind, most black males will never kill anyone. But black males commit a plurality of murders in the US. And 1 in 3 will spend some time in jail in his life. 1 in 7 American black men are in jail at any moment.

Remember, there are more whites in poverty than total blacks in the United States, and blacks commit 48% of murders.

If you're in an urban environment, don't be a progressive hero and try not to avoid young black males. It can be dangerous[1][2]

[1] http://isteve.blogspot.com/2013/11/yglesias-i-wasnt-victim-o...

[2] http://isteve.blogspot.com/2013/08/all-we-have-to-fear-is-in...

[tw268]

What I'm trying to say is not all racial stereotypes describe a false belief, though it is fashionable to think so.

[csense]

> What do you call a stereotype when it is backed by statistical data?

An inconvenient truth. Or perhaps the elephant in the room.