[lit-ideas] Re: Calling all women and someone who understands statistics

  • From: "Andreas Ramos" <andreas@xxxxxxxxxxx>
  • To: <lit-ideas@xxxxxxxxxxxxx>
  • Date: Thu, 6 Oct 2005 11:43:56 -0700

Calling all women and someone who understands statistics? Does this imply the "someone" isn't a woman?

What the hell. I'll answer.

The size of the population (the sample size in your study) determines the confidence of your results. The larger the population, the more reliable that your results will be.

Andy and Eric decide to solve their arguments with a coin toss.

Andy flips his coin four times. He gets 3 heads / 1 tails.

So... does this mean that a coin will land heads 75% of the time?

No. The problem is that his sample set (four flips) is so small that the results aren't representative of the total possibilities.

You need a "large-enough set" that gives you a reliable number.

Erin wants me to cut the blah-blah and shortcut to the answer, so I will:

400 events gives you a confidence within a 5% margin of error.
2000 events will produce results within a 2% margin of error.
10,000 events will produce results with a 1% margin of error.

So, in the study with 10,000 people, the results are very reliable. The vaccine 
works.

yrs,
andreas
www.andreas.com


------------------------------------------------------------------
To change your Lit-Ideas settings (subscribe/unsub, vacation on/off,
digest on/off), visit www.andreas.com/faq-lit-ideas.html

Other related posts:

  1. Home
  2. lit-ideas
  3. Archive
  4. 10-2005