MS_NotesWeek12

# MS_NotesWeek12 - 114 CHAPTER 2 ELEMENTS OF STATISTICAL...

• Notes
• 16

This preview shows pages 1–4. Sign up to view the full content.

114 CHAPTER 2. ELEMENTS OF STATISTICAL INFERENCE 2.5 Hypothesis Testing We assume that Y 1 , . . . , Y n have a joint distribution which depends on the un- known parameters ϑ = ( ϑ 1 , . . . , ϑ p ) T . The set of all possible values of ϑ is called the parameter space Θ . Example 2.31 . Parameter spaces for various distributions: 1. Y i iid Bin( m, p ) , where m is known, i = 1 , 2 , . . . , n . Then ϑ = p and Θ = (0 , 1) . 2. Y i iid Poisson( λ ) , i = 1 , 2 , . . . , n . Then ϑ = λ and Θ = R + . 3. Y i iid N ( μ, σ 2 ) , i = 1 , 2 , . . . , n . Then ϑ = ( μ, σ 2 ) T and Θ = R × R + . 4. Y i ∼ N ( β 0 + β 1 x i , σ 2 ) independently for i = 1 , 2 , . . . , n . Then ϑ = ( β 0 , β 1 , σ 2 ) T and Θ = R 2 × R + . A hypothesis restricts ϑ to lie in Θ star , where Θ star Θ . If we wish to test whether ϑ Θ star , then we test the null hypothesis H 0 : ϑ Θ star against the alternative hypothesis H 1 : ϑ Θ \ Θ star . Example 2.32 . Examples of sets Θ star restricted by null hypotheses: 1. Y i iid Poisson( λ ) , i = 1 , 2 , . . . , n . If we test H 0 : λ = λ 0 against H 1 : λ negationslash = λ 0 then Θ star = { λ 0 } . 2. Y i iid N ( μ, σ 2 ) , i = 1 , 2 , . . . , n . If we test H 0 : μ = μ 0 against H 1 : μ negationslash = μ 0 then Θ star = { μ 0 } × R + . 3. Y i ∼ N ( β 0 + β 1 x i , σ 2 ) independently for i = 1 , 2 , . . . , n . For testing H 0 : β 1 = 0 against H 1 : β 1 negationslash = 0 we have Θ star = R × { 0 } × R + . If Θ star is a single point, the hypothesis is simple . Otherwise, the hypothesis is composite . The set of all possible values y of the random sample Y is called the sample space Y .

This preview has intentionally blurred sections. Sign up to view the full version.

2.5. HYPOTHESIS TESTING 115 Example 2.33 . Examples of sample spaces: 1. Y i iid Poisson( λ ) , i = 1 , 2 , . . . , n . Then Y = [ Z + ∪ { 0 } ] n . 2. Y i iid N ( μ, σ 2 ) , i = 1 , 2 , . . . , n . Then Y = R n . The rejection region is the subset R ⊆ Y such that we reject H 0 if y ∈ R . 2.5.1 Type I and II errors There are two errors in hypothesis testing. A type I error is to reject H 0 when H 0 is true and a type II error is to accept H 0 when H 0 is false. When H 0 is simple, that is, we have H 0 : ϑ = ϑ 0 , the significance level or size of the test is defined to be α = P ( type I error ) = P ( Y ∈ R| H 0 true ) = P ( Y ∈ R| ϑ = ϑ 0 ) . When H 0 is composite, that is, we have H 0 : ϑ Θ star , where Θ star is not a single point, then we have a significance function α ( ϑ ) = P ( Y ∈ R| ϑ ) which depends on the particular value in Θ star taken by ϑ . In this case, the size of the test or significance level is defined to be α = max ϑ Θ star α ( ϑ ) . When H 1 is simple, that is, we have H 1 : ϑ = ϑ 1 , the power of the test is defined to be β = 1 - P ( type II error ) = 1 - P ( Y ∈ R c | H 1 true ) = P ( Y ∈ R| H 1 true ) = P ( Y ∈ R| ϑ = ϑ 1 ) , where R c is a complement of R (not a rejection region).
116 CHAPTER 2. ELEMENTS OF STATISTICAL INFERENCE When H 1 is composite, that is, we have H 1 : ϑ Θ \ Θ star , then we have a power function β ( ϑ ) = P ( Y ∈ R| ϑ ) which depends on the value in Θ \ Θ star taken by ϑ . 2.5.2 Neyman-Pearson lemma Suppose that H 0 and H 1 are both simple hypotheses, so that we wish to test H 0 : ϑ = ϑ 0 against H 1 : ϑ = ϑ 1 . Then we can compare many different tests with the same significance level α . We will choose the one with the largest power, that is, for a fixed α = P ( type I error ) we choose the test that maximizes β = 1 - P ( type II error ) . The following famous result gives a way of doing this.

This preview has intentionally blurred sections. Sign up to view the full version.

This is the end of the preview. Sign up to access the rest of the document.
• Fall '15
• Rogers
• Science, Null hypothesis, Statistical hypothesis testing, significance level, Statistical power

{[ snackBarMessage ]}

### What students are saying

• As a current student on this bumpy collegiate pathway, I stumbled upon Course Hero, where I can find study resources for nearly all my courses, get online help from tutors 24/7, and even share my old projects, papers, and lecture notes with other students.

Kiran Temple University Fox School of Business ‘17, Course Hero Intern

• I cannot even describe how much Course Hero helped me this summer. It’s truly become something I can always rely on and help me. In the end, I was not only able to survive summer classes, but I was able to thrive thanks to Course Hero.

Dana University of Pennsylvania ‘17, Course Hero Intern

• The ability to access any university’s resources through Course Hero proved invaluable in my case. I was behind on Tulane coursework and actually used UCLA’s materials to help me move forward and get everything together on time.

Jill Tulane University ‘16, Course Hero Intern