• let N-value stand for the size of the population
• let n stand for the size of the sample
• if we have a cage full of 100 rats, and pick a sample of 10

• then N-value = 100, n = 10
• the ratio n¸ N-value is called the sampling rate, usually specified as a %
• for above sample, sampling rate is:   ··%

parameter: a value computed by examining a whole population

• e.g. the average purchase price of all homes sold in Austin in 1996
• a parameter is a "true" value

• e.g. take a sample of just 100 home purchases and compute the average price for that sample
• a statistic is a guess or estimate of the true value

sampling error: a measure of the difference between a statistic and the parameter it is estimating

Example:

• if on the basis of a sample,
• we estimate the average home price to be $100,000 (statistic)
• and the true average is $90,000 (parameter)
• the raw sampling error  is ··
• the percent sampling error =  ·· %

Example:

• big bag of say, 1000 M&M’s.   Yumm!!

• parameter: # yellows in bag: let’s say there are 200
• want to estimate the # of yellow M&Ms (without counting them all)
• mix thoroughly, close your eyes, and grab a sample of 100
• count the yellows = 25
• statistic: estimate # of yellow M&M’s in bag:
• that's ·· out of ··, or ·· % yellow
• which would be ·· out of 1000
• so our estimate is: ·· yellow M&M's
• sampling error (estimate is off by): ·· out of 200 = ··%

• BUT, if we were very unlucky, our sample could be all yellows! (it’s possible, isn’t it?)
• in which case, statistic: ··
• sampling error: ·· out of 200 = ··%!!

Example (extreme):

• Little Suzy just loves green M&M’s!
• we let Little Suzy choose our sample, a - n - d . . .
• we let her see what she is picking!
• will we get an unbiased sample?
• tune in next week for the next exciting episode of:

Little Suzy and the Mystery of the Green Tongue!