Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
After obtained the numerical result from a sample, we can not say that the population mean falls between 65.071g and 79.129g with 95% chance.
If many repeated samples with the same sample size were taken from the same population and the confidence intervals were constructed, the proportion of intervals containing the population mean would be approximately 0.95.
11 October, 2011
Flow Chart for determining the distributions no Is sample size Is population distribution normal? sufficiently large (n >=30), such that CLT yes applied? Is population standard deviation yes no given?
yes no
Normal tables
11 October, 2011
t-distribution tables
Normal tables
2
11 October, 2011
11 October, 2011
11 October, 2011
Question 1
0 10 20 30 40 50 60 70
11 October, 2011
11 October, 2011
the principles of hypothesis testing Hypothesis testing of the population mean Hypothesis testing of the binomial population
11 October, 2011
Overview
In Statistics, a hypothesis is a claim or statement about one specific property of a population, e.g. Population mean Population proportion A hypothesis test (or test of significance) is a standard procedure for testing such claim and is also a formal objective decision-making procedure. Under the given assumption, if the probability of a particular observed event or more extreme is exceptionally small, we conclude that the assumption is probably not correct.
STAT 101 -- Part VIII 9
11 October, 2011
Two mutually exclusive hypotheses, called null and alternative hypotheses, are defined. Null hypothesis is denoted as while alternative hypothesis is Thus when one hypothesis is true the other is false, and vice versa. The null hypothesis is the presumed condition that will be accepted unless there is strong evidence against it. The alternative hypothesis is the claim that the researcher would like to establish based on the data, sometimes called a research hypothesis. The researcher would like to prove the claim under by rejecting However, the decision of not rejecting does not prove that is true
STAT 101 -- Part VIII 10
11 October, 2011
If you are conducting a study and want to use a hypothesis testing to support your claim, the claim must be worded so that it becomes the alternative hypothesis.
There are two possible decisions: reject null hypothesis and do not reject null hypothesis There are two possible truths: null hypothesis is true and alternative hypothesis is true Therefore, there are four possible outcomes in hypothesis testing Two of the possible outcomes are correct: do not reject null hypothesis when null hypothesis is true or reject null hypothesis when alternative hypothesis is true If we reject null hypothesis when null hypothesis is true, we have committed a Type I error If we do not reject null hypothesis when alternative hypothesis is true, we have committed a Type II error
STAT 101 -- Part VIII 12
11 October, 2011
11 October, 2011
13
A general aim in hypothesis testing is to use statistical tests that make and as small as possible. As increases, will decrease or vice versa The general strategy is to control at some specific level (for example, 0.10, 0.05, 0.01, .) and use the test that minimizes , or equivalently, maximizes the power.
11 October, 2011
14
http://www.intuitor.com/statistics/T1T2Errors.html
In Singapore court of law, the fundamental principle is that a defendant is presumed innocent until proven guilty. Because innocence is the initial assumption, we have The defendant is innocent The defendant is guilty The job of the prosecutor (analogous to researcher) is to present evidence (analogous to the sample data) so compelling that the judge is persuaded to reject the null hypothesis. In Singapore legal system, defendants are found to be either guilty or not guilty, but they are never found to be innocent. A verdict of not guilty means the evidence is not sufficient to establish guilt, but it does not prove innocence. By controlling the probability of making type I error, Singapore legal system tries to control the probability that an innocent person is convicted.
11 October, 2011 STAT 101 -- Part VIII 15
11 October, 2011
16
Two-sided alternatives
11 October, 2011
17
11 October, 2011
18
11 October, 2011
19
11 October, 2011
20
p-value method
11 October, 2011
21
11 October, 2011
22
p-value
Test statistic t
11 October, 2011 STAT 101 -- Part VIII 23
11 October, 2011
24
11 October, 2011
25
1.
2. 3. 4. 5. 6.
Assumptions of population distribution if necessary State null and alternative hypotheses Calculate test statistic Rejection rule: based on critical value or pvalue approaches Decision of the test: reject null hypothesis or do not reject null hypothesis Conclusion statement in terms of alternative hypothesis
STAT 101 -- Part VIII 26
11 October, 2011
Example: pollution
The Public Health Service (US) publishes the Annual Data Tabulation, Continuous Air Monitoring Projects, which recently indicated that a large mid-western city had an annual mean level of sulfur dioxide of 0.12 (concentration per parts per million). To change this concentration, many steel mills and other manufacturers installed antipollution equipment. Plans are to make about 36 random checks during the year to determine if there has been a change in the sulfur dioxide level. The 0.05 significance level is used. Thirty-six random checks were made throughout the year. It was found that sample mean was 0.10 and sample standard deviation is 0.03. It is assumed that the level of sulfur dioxide is normally distributed. This assumption is not necessary!
11 October, 2011 STAT 101 -- Part VIII 27
11 October, 2011
29
11 October, 2011
30
Other cases
11 October, 2011
31
Example
In a recent article (USA Today, June 19, 2002) it was claimed that the average supermarket trip takes 22 minutes. Suppose that, in an effort to test this claim, a sample of 50 shoppers at a local supermarket were studied. The mean shopping time for the sample of 50 shoppers was 25.36 minutes with a standard deviation of 7.24 minutes. Using the 0.05 level of significance, is there evidence that the mean shopping time at the local supermarket is different from the claimed value of 22 minutes?
11 October, 2011
32
11 October, 2011
33
yes
Is population standard deviation given? yes no
t-distribution tables
STAT 101 -- Part VIII
Normal tables
34
11 October, 2011
35
11 October, 2011
36
http://bcs.whfreeman.com/pbs/cat_050/pbs /pvalue_pbs.html
P-value calculation
11 October, 2011
38
Mid-term Test Result 2010 Stem-and-Leaf Display Stem unit: Statistics Sample Size Mean Median Std. Deviation Minimum Maximum 99 74.41414 80 17.82741 29 99 10 2 9 3 11223357 4 5 223445668 6 113344556688 7 1113334455566677779 8 0000112233333444445566677778889 9 0001113333455566789
11 October, 2011
39