Review Sheet for Test on Sections 6.2, 6.3, and 7.1
D period: Friday, 2/26/1999
F period: Thursday, 2/25/1999
Switching between tests is permitted only with written permission from your other teacher.
Terminology (approx. 25 points):
alternative hypothesis
confidence intervals
critical value (upper p critical value) for z and t tests
degrees of freedom
distribution-free procedure
matched pairs procedure (especially reread paragraph on p.506)
nonparametric procedure
null hypothesis
one-sided alternative / one-sided tests
power
P-value
robust procedure
sign test (i.e., the form given on p.516)
significance (statistical significance vs. practical significance)
significance level (if level is fixed, we denote it by a
)
standard error (sometimes called "estimated standard error")
t distribution
t test for significance
two-sided alternative / two-sided tests
Type I error
Type II error
z test for significance
looking up critical values in t or z tables to estimate probabilities
using calculator to compute binomial, t, or z probabilities
binomial coefficients (yes, you should still remember them)
estimating "inverse t" or "inverse z" value from the tables (i.e., given a percentage, estimate t or z)
computing P-value
computing standard error
performing a sign test
making a normal quantile plot (non-TI 83 users may make a stemplot instead)
Problem Solving (approx. 30 points):
You will need to be able to set up and work one or two entire problems of the type represented in the homework. In fact, one of the problems will probably be taken virtually unchanged from the homework.
You must know the rules of thumb on p.510. You won’t need to write them out word for word, but you’ll need to demonstrate thorough knowledge by applying them (and mentioning that you’re applying them) as you solve the problems.
Write your conclusion to each problem as a complete sentence using accurate wording.
You may be asked to calculate power (against a given instance of Ha) and/or assess the impact of changing the sample size. More important than the calculations themselves, however, will be a qualitative assessment of power (or Type II error) and how it changes.
In-Depth Understanding (approx. 20 points):
Reread pages 472-477. In fact, if you were to give a layman 5 pages from the text to read to help make him a more intelligent consumer of statistics, these would be the pages to choose. (Maybe you’ll bookmark these pages for use later in life…?) You will have one or two essay questions based on this passage and/or similar types of "thought" questions.
You should be able to use graphs and words to explain each of the answers to problem 6.69 (p.494).
Some Final Words of Advice
1. Work the entire test. Do not get hung up on any one problem for too long.
2. Always give your answers a "reasonableness check." If you have time, try an alternate method as a check.
3. It is almost always a good idea to draw a diagram.
4. Remember that "AP-style" criteria will be enforced as follows:
a) Exposition must be linear, top to bottom. No jumping around or fragmentary thoughts in the final presentation.
b) Cross out anything you wish to be ignored (preliminary work that led nowhere, for example). However, anything that is crossed out will be truly ignored (i.e., not graded). No need to erase.
c) Document any assumptions or rules of thumb used (e.g., np > 10, n(1-p) > 10).
d) Use "AP-style" wording. For example, avoid using the word "it," and be able to make professional, suitable-for-the-layman statements about confidence intervals that avoid the words "probability" or "chance."
e) In cases where there is no single right or wrong answer, be sure to provide adequate justification. By the same token, however, do not write extra words that add nothing of value. Keep your answers focused and brief.
f) Use standard notation wherever possible. It is not necessary to document, for example, that x with a bar over it denotes the sample mean, because that is standard statistical notation. Others include H0 for null hypothesis, s/Ö n for standard error, and m 0 for the null hypothesis mean. If you use symbols that may be subject to some confusion, be sure to define what they mean.
g) Starting with this test, misuse of the equal sign (=) will receive a point penalty. The typical place where people are tempted to do this is when computing a z or t statistic and then asserting that the statistic "equals" the probability associated with it.
h) Unless otherwise stated, all numeric answers should be rounded to 3 places after the decimal point.