POLI 3000 Lecture Notes - Lecture 10: Null Hypothesis, Statistical Significance, Test Statistic
![](https://new-preview-html.oneclass.com/Exbq3r4gwdYONPXbe1zYNy1MLBo2plvz/bg1.png)
P value: probability that the data is extreme or more extreme given the null is true.
Degrees of freedom= n-1 (when you are doing sample means)
Why do we call them degrees of freedom?
-how many pieces of information do I need?
-if you have a mean of 51, and you know 27 of them, you can solve for 28.
-the minimum number of variables you need to figure out the rest of information.
Standard deviation equation divide by n-1: because of degrees of freedom
Summary:
1. If sample size is large enough: 95% CI
I. Warning about significance:
a. Important distinction
i. Statistical significance not equal to Substantive significance
ii. Statistical significance is a poor choice of wording:
1. Better: statistically distinguishable from the null
II. Why 2 tailed tests?
a. 1 tailed test leave too much room for mischief
b. 2 tailed tests protect from concerns about post-hoc theorizing
i. Making it easier to find statistical significance
c. 2 tailed tests are a higher bar
d. convention
Bivariate Statistics:
I. Relationships between two variables
a. Independence: when we are testing the relationships between two variables, we are
asking if they are independent
i. If they are independent: then knowing one variable tells you nothing about the
other
ii. Null hypothesis is always that there is no relationship
II. Getting test statistic: (probability-null hypothesis) divided by standard error
find more resources at oneclass.com
find more resources at oneclass.com
Document Summary
P value: probability that the data is extreme or more extreme given the null is true. Degrees of freedom= n-1 (when you are doing sample means) If you have a mean of 51, and you know 27 of them, you can solve for 28. The minimum number of variables you need to figure out the rest of information. Standard deviation equation divide by n-1: because of degrees of freedom. If sample size is large enough: 95% ci. Important distinction: statistical significance not equal to substantive significance, statistical significance is a poor choice of wording, better: statistically distinguishable from the null. Why 2 tailed tests: 1 tailed test leave too much room for mischief, 2 tailed tests protect from concerns about post-hoc theorizing, making it easier to find statistical significance, 2 tailed tests are a higher bar, convention.