there is a bias

[keannu]

[FONT=&#54620]37)Here’s a fascinating social experiment. Select one hundred people at random in New York City and ask them each to list all their friends, so you can figure out their average number of friends. Then, in turn, ask their friends how many friends they have. You will find that the latter’s average number of friends is higher. Sociologist Scott Feld of Purdue University drew attention to this apparent paradox. The explanation lies in the realization that there is a bias in the question being asked, since you are more likely to know popular people and less likely to know unpopular ones. That is also why people at your local gym tend to be fitter than you, because you do not encounter the relatively out*-of-*shape ones who rarely turn up.

I can't get how the underlined is related to less friends of yours compared to your friend's more friends. I can't get the whole paragraph.



[/FONT]

 

[+Technist Warp-]

For example, many people know Justin Beiber but Justin Beiber doesn't know as many people.

You might go to Justin Beiber and ask him who his friends are. Then you ask Justin Beiber's friend who his friends are. Justin Beiber's friend would probably have more friends than Justin Beiber because Justin Beiber is more popular so might be busy while his friend is just Justin Beiber's friend and is unpopular. Justin Beiber is popular, so you know him. You might not know his friend, who might be unpopular.

The underlined section you pointed out means that the question and survey is somehow unfair because you would probably choose 100 people from you circle of friends, popular people you know, or acquaintances.

 

[SoothingDave]

For example, many people know Justin Beiber but Justin Beiber doesn't know as many people.

You might go to Justin Beiber and ask him who his friends are. Then you ask Justin Beiber's friend who his friends are. Justin Beiber's friend would probably have more friends than Justin Beiber because Justin Beiber is more popular so might be busy while his friend is just Justin Beiber's friend and is unpopular. Justin Beiber is popular, so you know him. You might not know his friend, who might be unpopular.

The underlined section you pointed out means that the question and survey is somehow unfair because you would probably choose 100 people from you circle of friends, popular people you know, or acquaintances.

That's completely wrong.

Justin Bieber will know more people because he is popular. It is the popular people who have more friends. You may have 5 friends, but one of them is real popular, so that person will have 25 friends.

Ask 100 people and they will have 5 friends on average. But they will know 1 or 2 popular people who have 25 friends each. Of the original 100 people, the average number of friends is 5. But of their friends, the average number of friends they have will be higher.

Your explanation of the "bias" in the question is wrong, too. It clearly states that the 100 people are chosen at random. It has nothing to do with choosing 100 people "from your circle."

The "bias" is that when you pick 100 random people you are not going to catch a lot of popular people in that sample. But when you ask anybody for a list of friends, that will include popular people. Which will drive the average number of friends up.

 

[keannu]

Thanks SD!
Then how is this fact related to the argument? Does it mean that you meet more popular people at the gym, who have more friends? Bias seems to refer to asking the surveyed for their popular friends, but I can't get it completely. Your first part is quite comprehensible!

That is also why people at your local gym tend to be fitter than you, because you do not encounter the relatively out*-of-*shape ones who rarely turn up.

 

[SoothingDave]

The gym is a completely different example. Not about popularity, but about fitness. You will meet more "fit" people (percentage wise) at a gym than you will in a random sample of people.

The bias is in the selection of the people to sample. Bias meaning there is an influence in the result you get.

Consider if I ask 100 random people on the street if they think we should have higher liquor taxes to pay for education or medicine for senior citizens. Then consider that I ask 100 people who are just coming out of a bar the same question.

Do you think I would get different answers?

 

[MikeNewYork]

Good points.

 

[emsr2d2]

I'm guessing TechnistWarp and Justin Bieber aren't friends, or TW would know how to spell his surname correctly. ;-)

 

[Raymott]

Also, some of the first 100 you ask may have no friends. But each of the "friends" you ask must have at least one friend (the original referrer) - assuming that friendship is reciprocal.

You could also ask a random 100 people how many alcoholics they know; and then ask those alcoholics how many alcoholics they know. I think the number will go up for a similar reason to the original. But here you're replacing friendliness with alcoholism.