340 likes | 372 Views
Student’s t Distribution. Lecture 36 Section 10.2 Fri, Nov 11, 2005. What if is Unknown?. It is more realistic to assume that the value of is unknown. (If we don’t know the value of , then we probably don’t know the value of ). In this case, we use s to estimate .
E N D
Student’s t Distribution Lecture 36 Section 10.2 Fri, Nov 11, 2005
What if is Unknown? • It is more realistic to assume that the value of is unknown. • (If we don’t know the value of , then we probably don’t know the value of ). • In this case, we use s to estimate .
What if is Unknown? • Let us assume that the population is normal or nearly normal. • Then the distribution ofx is normal. • That is, • However, for small n,
What if is Unknown? • If it is notN(0, 1) , then what is it?
Student’s t Distribution • It has a distribution called Student’s t distribution. • The t distribution was discovered by W. S. Gosset in 1908. • See http://mathworld.wolfram.com/Studentst-Distribution.html
The t Distribution • The shape of the t distribution is very similar to the shape of the standard normal distribution. • It is • symmetric • unimodal • centered at 0. • But it is wider than the standard normal. • That is because of the additional variability introduced by using s instead of .
The t Distribution • Furthermore, the t distribution has a (slightly) different shape for each possible sample size. • As n gets larger and larger, s exhibits less and less variability, so the shape of the t distribution approaches the standard normal. • In fact, if n 30, then the t distribution is approximately standard normal.
Degrees of Freedom • If the sample size is n, then t is said to have n – 1 degrees of freedom. • We use df to denote “degrees of freedom.” • We will use the notation t(df) to represent the t distribution with df degrees of freedom. • For example, t(5) is the t distribution with 5 degrees of freedom (i.e., sample size 6).
Standard Normal vs. t Distribution • The distributions t(2), t(30), and N(0, 1). t(2) t(30) N(0, 1)
Is known? yes no Is the population normal? TBA yes no Is n 30? yes no Give up Decision Tree
Is known? yes no Is the population normal? Is the population normal? yes no yes no Is n 30? yes no Give up Decision Tree
Is known? yes no Is the population normal? Is the population normal? yes no yes no Is n 30? Is n 30? yes no yes no Give up Decision Tree
Decision Tree Is known? yes no Is the population normal? Is the population normal? yes no yes no Is n 30? Is n 30? yes no yes no Give up
Is known? yes no Is the population normal? Is the population normal? yes no yes no Is n 30? Is n 30? yes no yes no Give up Decision Tree
Is known? yes no Is the population normal? Is the population normal? yes no yes no Is n 30? Is n 30? Is n 30? yes no yes no yes no Give up Decision Tree
Is known? yes no Is the population normal? Is the population normal? yes no yes no Is n 30? Is n 30? Is n 30? yes no yes no yes no Give up Decision Tree
Is known? yes no Is the population normal? Is the population normal? yes no yes no Is n 30? Is n 30? Is n 30? yes no yes no yes no Give up Give up Decision Tree
Is known? yes no Is the population normal? Is the population normal? yes no yes no Is n 30? Is n 30? Is n 30? yes no yes no yes no Give up Give up Decision Tree
Is known? yes no Is the population normal? Is the population normal? yes no yes no Is n 30? Is n 30? Is n 30? yes no yes no yes no Give up Give up Decision Tree
Table IV – t Percentiles • Table IV gives certain percentiles of t for certain degrees of freedom. • Specific percentiles for upper-tail areas: • 0.40, 0.30, 0.20, 0.10, 0.05, 0.025, 0.01, 0.005. • Specific degrees of freedom: • 1, 2, 3, …, 30, 40, 60, 120.
Table IV – t Percentiles • The table tells us, for example, that • P(t > 1.812) = 0.05, when df = 10. • Since the t distribution is symmetric, we can also use the table for lower tails by making the t values negative. • So, what is P(t < -1.812), when df = 10?
Table IV – t Percentiles • The table allows us to look up certain percentiles, but it will not allow us to find probabilities in general.
TI-83 – Student’s t Distribution • The TI-83 will find probabilities for the t distribution (but not percentiles, in general). • Press DISTR. • Select tcdf and press ENTER. tcdf( appears in the display. • Enter the lower endpoint. • Enter the upper endpoint.
TI-83 – Student’s t Distribution • Enter the number of degrees of freedom (n – 1). • Press ENTER. • The result is the probability.
Example • Enter tcdf(1.812, E99, 10). • The result is 0.0500. • Enter tcdf(-E99, -1.812, 10). • The result is 0.0500. • Thus, P(t > 1.812) = 0.05 when there are 10 degrees of freedom (n = 11).
Hypothesis Testing with t • We should use the t distribution if • The population is normal (or nearly normal), and • is unknown (so we use s in its place), and • The sample size is small (n < 30). • Otherwise, we should not use t. • Either use Z or “give up.”
Hypothesis Testing with t • The hypothesis testing procedure is the same except for two steps. • Step 3: Find the value of the test statistic. • The test statistic is now • Step 4: Find the p-value. • We must look it up in the t table, or use tcdf on the TI-83.
Example • Re-do Example 10.1, p. 616, (by hand) under the assumption that is unknown.
TI-83 – Hypothesis Testing When is Unknown • Press STAT. • Select TESTS. • Select T-Test. • A window appears requesting information. • Choose Data or Stats.
TI-83 – Hypothesis Testing When is Unknown • Assuming we selected Stats, • Enter 0. • Enterx. • Enter s. (Remember, is unknown.) • Enter n. • Select the alternative hypothesis and press ENTER. • Select Calculate and press ENTER.
TI-83 – Hypothesis Testing When is Unknown • A window appears with the following information. • The title “T-Test” • The alternative hypothesis. • The value of the test statistic t. • The p-value. • The sample mean. • The sample standard deviation. • The sample size.
Example • Re-do Example 10.1, p. 616, on the TI-83 under the assumption that is unknown.
Let’s Do It! • Let’s Do It! 10.3, p. 630 – Study Time. • Let’s Do It! 10.4, p. 631 – pH Levels.
Homework • Use the TI-83 to find the following probabilities for the t distribution. • P(T > 2.5), df = 14. • P(-3 < T < 3), df = 7. • P(T < 3.5), df = 5. • Use the TI-83 to find the following t-distribution percentiles. • 95th percentile, df = 25. • 20th percentile, df = 21. • Endpoints of the middle 80%, df = 12.