Political Science 5: Lecture

1
Political Science 5Lecture 12 3/4/2004

• Midterm and Homework 2 back on Tuesday
• Homework Project 3 posted, due next Thursday
• Only have to do worksheets this time, next
  homework will be tougher (tests of significance)

2
The Magic of the Normal Curve

• Normal Curves
• The family of normal curves
• The rule of 68-95-99.7
• The Central Limit Theorem
• Confidence Intervals
• Around a Mean
• Around a Proportion

3
Normal Curves (or distributions)

• Normal curves are a special family of density
  curves, which are graphs that answer the
  question what proportion of my cases take on
  values that fall within a certain range?
• Many things in nature, such as sizes of animals
  and errors in astronomic calculations, happen to
  be normally distributed.

4
Normal Curves
5
Normal Curves

• What do all normal curves have in common?
• Symmetric
• Mean Median
• Bell-shaped, with most of their density in center
  and little in the tails

• How can we tell one normal curve from another?
• Mean tells you where it is centered
• Standard deviation tells you how thick or narrow
  the curve will be

6
Normal Curves

• The 68-95-99.7 Rule.
• 68 of cases will take on a value that is plus or
  minus one standard deviation of the mean
• 95 of cases will take on a value that is plus or
  minus two standard deviations
• 99.7 of cases will take on a value that is plus
  or minus three standard deviations

7
The Central Limit Theorem

• If we take repeated samples from a population,
  the sample means will be (approximately) normally
  distributed.
• The mean of the sampling distribution will
  equal the true population mean.

8
The Central Limit Theorem

• A sampling distribution of a statistic tells us
  what values the statistic takes in repeated
  samples from the same population and how often it
  takes them.

9
Confidence Intervals

• We use the statistical properties of a
  distribution of many samples to see how confident
  we are that a sample statistic is close to the
  population parameter
• We can compute a confidence interval around a
  sample mean or a proportion
• We can pick how confident we want to be
• Usually choose 95, or two standard errors

10
Confidence Intervals

• The 95 confidence interval around a sample mean
  is

11
Confidence Intervals

• The 95 confidence interval around a sample
  proportion is
• And the 99.7 confidence interval would be

12
Confidence Intervals

• If my sample of 100 donors finds a mean
  contribution level of 15,600 and I compute a
  confidence interval that is
• 15,600 or - 600
• I can make the statement I can say at the 95
  confidence level that mean contribution for all
  donors is between 15,000 and 16,200.