You are on page 1of 3

HW6

Q1)
What we know
𝐻! : θ = 0.3 vs H" : 𝜃 = 0.5
𝑅 = {𝑥: 𝑥̅ > 0.469}
𝛼 = 0.05
Type 2 error rate = 𝛽 = 𝑃(𝑑𝑜 𝑛𝑜𝑡 𝑟𝑒𝑗𝑒𝑐𝑡 𝐻! | 𝐻" 𝑖𝑠 𝑡𝑟𝑢𝑒)

Since we know that 𝑃(𝑧 > 𝑐|𝐻! ) = 0.05 = 𝛼, 𝑤𝑒 𝑐𝑎𝑛 𝑓𝑖𝑛𝑑 𝛽 𝑢𝑠𝑖𝑛𝑔

𝑃(𝑧 ≤ 𝑐|𝐻" )

𝑐 − 𝜃" 0.469 − 0.5


𝑐|𝐻" = = = −0.27727
P𝜃" (1 − 𝜃" ) P0.5(0.5)
𝑛 20
𝑃(𝑧 ≤ −0.27727)
b) we can use MLE based pivotal quantity
c)

New hypothesis says 𝐻" : 𝜃 > 0.3


And since the null hypothesis for this will be 𝐻! : 𝜃 = 0.3, the testing procedure for this is not
the same

We can then find the rejection region the same way in the last homework which is

𝜃! (1 − 𝜃! ) 0.3(0.7)
𝑐 = 𝜃! + 1.645U = 0.3 + 1.645U = 0.469
𝑛 20
𝑅 = {𝑥: 𝑥̅ > 0.469}
Q2)
a)
it is a good choice because if the number of 6s is closer to 5 (which is the expected number of
6s with a fair die) then the test statistic supports 𝐻! else if it is closer to 10 then the test statistic
favors 𝐻"
b)
𝑃(𝑟𝑒𝑗𝑒𝑐𝑡 𝐻! |𝐻! 𝑖𝑠 𝑡𝑟𝑢𝑒)
𝑃(𝑋 ≥ 10|𝑤ℎ𝑒𝑛 𝜃 = 1/6)
We can rewrite >= to > by saying X>9 since X only contains integers.

𝑃(𝑋 > 9|𝑤ℎ𝑒𝑛 𝜃 = 1/6)


Now since this is a binomial distribution we can use R to find the probability using
pbinom(9,30,1/6)

c)

to find the power we can use 𝑃(𝑟𝑒𝑗𝑒𝑐𝑡 𝐻! |𝐻" 𝑡𝑟𝑢𝑒)


"
which is 𝑃(𝑋 > 9 |𝜃 = #)
again we can use R and binomial distribution to find this probability.

𝑝𝑏𝑖𝑛𝑜𝑚(9,30, 1/3)

You might also like