About Hypotheses
Ho: p = 0.7
HA: p > 0.7
How could we write a pair of hypotheses to test whether the evidence suggests a proportion is larger than 0.7?
Control, randomize, replicate.
What are the principles of experimental design?
A test with hypotheses of the form:
H0: p = p0
HA: p not equal p0 OR HA: p < p0 OR HA: p > p0
where p0 is some given proportion and p is an unknown population proportion.
What is a one-sample t-test?
The condition that means a sample, meant to draw conclusions about a population, must be less than 10% of the total population.
What is the 10% Condition?
Ho: mu = 3.5
HA: mu does not equal 3.5
How could we write hypotheses to test whether the true mean is different from 3.5?
A group to which a known or null treatment is applied, so that we can draw comparisons.
What is a control group?
A test we'd use to determine if there is statistically significant evidence suggesting a population mean is lower than a proposed value
What is a one-sample t-test?
When the scenario involves a randomized experiment not meant to draw conclusions about a population.
When is the 10% Condition not required?
DA: To test whether there is evidence to suggest the mean is not zero.
HDA: To test whether there is evidence to suggest there is a difference in proportions.
DA: When would we use the hypotheses: H0: mu = 0 and HA: mu is not 0?
DA: When would we use the hypotheses: H0: p1 = p2 and HA: p1 is not p2?
For example, by giving one group a sugar pill, and not allowing the psychiatrists interviewing the subjects to learn which group has been given each treatment.
How could you design a medicinal study to be double-blind?
DA: One example would be to determine if there's evidence that a factory's machinery is malfunctioning so that candy bars are smaller than they should be.
HDA: One example would be to determine if there's evidence suggesting that candy bars made in Factory A are larger than those in Factory B.
DA: What is an example of a situation in which we'd use a one-sample t-test?
HDA: What is an example of a situation in which we'd use a two-sample t-test?
DA: By having a sample size of at least 30 or having a sample that appears symmetric with no outliers.
HDA: It is required for a two-proportion z-test or a two-sample t-test.
DA: How do we satisfy the Large Enough Sample/Nearly Normal Condition?
HDA:When is the Independent Groups Condition needed?
DA: This occurs when we fail to reject a false null hypothesis.
HDA: The number of standard errors away from the null value.
DA: What is a Type II error?
HDA: What is a standardized test statistic?
We should use a randomized block design in this case.
If we cannot control variation in a variable we do not want to study, what can we do to minimize its effects?
DA: The test we'd use to determine if evidence suggests dog's average attention span increased after training.
HDA: The first considers the mean difference of paired data, the second the difference of the means of independent samples.
DA: What is a paired t-test?
HDA: What is the difference between a paired t-test and a two-sample t-test?
DA: It is the proportion used to check the Success/Failure Condition in a one-proportion z-test.
HDA: They are the proportions used to check the Success/Failure Condition in a two-proportion z-test.
DA: The null value given in the null hypothesis.
HDA: The sample proportions, given by the actual data.
Accept the null hypothesis.
What is a conclusion can never be the result of a significance test?
It is necessary to make any cause-and-effect claims.
Why is it so important to have randomly assigned treatments in an experiment?
For example, to determine, by asking a random sample of people if they support a politician, if a politician has popular support.
When would you use a one-proportion z-test?
DA: It is the only test that we've used that requires four conditions.
HDA: It is the only condition that we've only seen attached to one test.
DA: What's special about the paired t-test's conditions?
HDA: What's special about the Paired Data Condition?