www.ditl.org

stitch626 wrote:

used the happiness levels of a million individual US citizens to discover which are the best and worst states in which to live in the United States.

So this is a bogus test because that is not a valid sample data.

There are more people than that in New York City.

Huh? When you're doing this kind of research you use a random, representative sample. You can't survey all 300 million people.

39 Nevada

39? 39??? Whores and gambling are legal! Who wouldn't be happy with whores and gambling?

Nutso wrote: 39? 39??? Whores and gambling are legal! Who wouldn't be happy with whores and gambling?

Mormons.

4 Tennessee

Way to hold it down Tennessee!

stitch626 wrote: So this is a bogus test because that is not a valid sample data.

There are more people than that in New York City.

Test validity is an involved field of knowledge, with many methods of analysis.

However, presuming the sample was properly randomized, testing one million people out of three hundred million people is "good", unless you're looking for extremely rare outliers or something like that.

sunnyside wrote:

stitch626 wrote: So this is a bogus test because that is not a valid sample data.

There are more people than that in New York City.

Test validity is an involved field of knowledge, with many methods of analysis.

However, presuming the sample was properly randomized, testing one million people out of three hundred million people is "good", unless you're looking for extremely rare outliers or something like that.

The "proper" sample size is 10% to remove bias. This is only 1%.

Also, does it say how they selected the sample? Whether it was an equal amount per state, or if it was a ratio based on the states population?

Oh ya...#31 Iowa...I'm happier than Deep, Mikey, and all the Switzers!

stitch626 wrote:The "proper" sample size is 10% to remove bias.

That's nonsense. There's no particular minimum sample size that you need to use, it just needs to be selected without systematic error. 10% is a ridiculously huge sample size if you're studying a large population. Even 1% is huge.

Sorry, I made a comment to this, but for some reason it did not post and I don't remember all of what I said. So I'll just drop it.

Lt. Staplic wrote:Oh ya...#31 Iowa...I'm happier than Deep, Mikey, and all the Switzers!

I can fix that.

Lazar wrote:

stitch626 wrote:The "proper" sample size is 10% to remove bias.

That's nonsense. There's no particular minimum sample size that you need to use, it just needs to be selected without systematic error. 10% is a ridiculously huge sample size if you're studying a large population. Even 1% is huge.

It may be huge, but it's still what I learned in my stats class, 10% at least for a valid test.

Nickswitz wrote:It may be huge, but it's still what I learned in my stats class, 10% at least for a valid test.

What is the source on this number? Your teacher, or a book? Did they say why?

Lazar wrote:

Nickswitz wrote:It may be huge, but it's still what I learned in my stats class, 10% at least for a valid test.

What is the source on this number? Your teacher, or a book? Did they say why?

Book, and no, I don't remember the reasoning behind it...

I rember, vaguely, that a 10% sample size is often recommended in order to get a more accurate sample of the population. the larger numbers will help temper any unusual conglomerations of sub-groups within your sample.

Lt. Staplic wrote:Oh ya...#31 Iowa...I'm happier than Deep, Mikey, and all the Switzers!

Deepcrush wrote:

4 Tennessee

Way to hold it down Tennessee!

You think so???