1/18
Looks like no tags are added yet.
Null Hypothesis
There is NO difference between the two population proportions.
Alternative Hypothesis
There is a difference between the two population proportions
Randomization test
Check to see if our data is caused by random chance
Pretend the null is true and repeatedly shuffle the data we already have
Null Distribution
A null distribution is a collection of simulated differences that we would expect to see if there really were NO difference between the population proportions.
Centered at 0
Count Samples cutoff
The value used to decide which simulated differences count as extreme
Actual observed result or larger in a null distribution
p-value
If the null hypothesis were true, how likely would we be to get a difference as extreme as (or more extreme than) the one we observed?
P-value less than 0.05
Reject the null hypothesis
P-value greater than 0.05
Fail ro reject null hypothesis
Conditions for mathematical approach
Independence
Success/failure (10 successes and 10 failures)
What if conditions on mathematical approach are not met?
Use randomization approach
Standard Error
how much we expect our sample difference in proportions to vary from sample to sample.
Test Statistic
tells us how far our observed difference is from what we would expect under the null hypothesis, measured in standard errors
Workflow of a problem
Read the research question
Write the null and alternative hypothesis
Check conditions to determine if using randomization or mathematical approach
Find and interpret the p-value
Calculate and interpret the confidence interval
Communicate conclusion
Paired data
each observation in one group is connected to a specific observation in another group.
Randomization test on paired data
randomly flipping signs to see what averages could happen under the null hypothesis.
When to use the nearly normal histogram criteria vs the success/failure criteria
Nearly Normal Histogram
This applies when you're working with means
for n<15, no outliers or skew in shape
for n<30, some skewness is okay, no major outliers
for n<60 Moderate skewness is okay, some outliers
for n>60 use CLT
What happens when the confidence level changes?
When the confidence level gets higher- wider interval.
What happens with a bigger sample size
Larger sample size = narrower confidence interval.
Smaller sample size = wider confidence interval.
Margin of error
amount of random sampling error in the results