More fun with Math

In high school I took an intro statistics class and we touched upon the limitations of simple random sampling. One of these limitations is making sure you're giving everyone in the population an equal probability of being chosen, or you run the risk of getting screwy results.

Coincidentally, there was a newspaper article that provided a terrific example of messing up with simple random samples.

Researchers were studying sexual behavior in adults and asked in a survey one simple question: how many times do you have sex?

They found that on average men were having a lot more sex than women. Which doesn't really make sense, unless either: 1.) Women were lying about how many times they had sex to avoid looking like sluts, or 2.) Men were lying about how many time they had sex to look like studs.

While the researchers knew that some lying was taking place, it didn't seem like enough to make up for th differences they were experiencing.

Then the researchers realized that they were leaving out a key population out of their statistics: prostitutes. After correcting for that fact, the numbers were much more in-line and they also found that the average prostitute had sex about 300 times a year.

All this provided a lesson to us all: make sure you account for the right population when trying to develop your sample.

Tags:

4 comments:

mapgirl said...

Those are tags that'll get your blog denied at my work! ;-)

franky said...

lol, sorry mapgirl

Scott said...

With statistics, garbage in, garbage out. With a Bachelor's degree in statistics, I know you can prove anything with the right inputs.

franky said...

scott,
That reminds me of that statistics joke where they ask three different people a mathematician, a computer scientist, and a statistician what's 2+2. They get to the statistician and he says: "What do you want it to be".

SAMEDAYMUSIC

Designed by Posicionamiento Web | Bloggerized by GosuBlogger | Blue Business Blogger