Ho : u ≤ 40
Ha : u > 40
X = 40.5 n = 10 σ = 1.25 α = 0.05 df = 9
Z = x̄ - μ
--------
σ / √n
z = -1.264911 α = 0.05 => 1.83
-1.264911 < 1.83
Comment: With an alpha level of 0.05 there was not enough evidence to support that the battery’s life will not exceed 40 hours. Hence we failed to reject Ho.
1.83 => 41.22,
p( Z > 1.83) = 1 - P(Z < 1.83)
= 1 - 0.95
p( Z > 1.83) = 0.05
p( Z > -1.264911) = 1 - P(Z < -1.264911)
= 1 - 0.1029516
p( Z > -1.264911) = 0.8970484
0.8970484 > 0.05
β: P(X < 41.22 | μ = 42)
Z of 41.22 to 42 => -1.973261
P(Z < -1.973261) = 0.0242329
Comment: the probability of having a β-error if the true mean life is 42 hours is very low at only 2.42329%.
β: P(X < 41.22 | μ = 44) = 0.10
n = (Za + Zb)^2 * σ^2
-----------------
δ^2
n = 1.233686
Comment: If the true mean is 44 no matter how many the sample size is used the chances of having β-error will never exceed 0.10. As seen with the equation is that to have at least 1.233686, which would just be rounded off to 1, as a sample size to only have 0.10 chance for a β-error. If you would increase your sample size β-error will get smaller.
It is important to set a proper confidence bound so that you can minimize having an α-error and a β-error. One way that you would be able to reject Ho in part A would be changing it from a right tail to a left tail and decrease the confidence bound 0.90 which would have a z-score of -1.28.
Ho: u1 = u2
Ha: u1 < u2
Brand A
u = 19.6 n = 16
σ = 0.4 α = 0.05
df = 15
Brand B
u = 20.2 n = 16
σ = 0.6 α = 0.05
df = 15
Solution
Zo = u1 - u2
---------
σ1^2 σ2^2
√(---- + ----)
n1 n2
Zo = -3.328201
P(Z < -3.328201) = 0.0004370436
0.0004370436 < 0.05
Comment: With a significance level of 0.05 there is enough evidence to say that brand B is significantly better than that of brand A. In this scenario we reject the null hypothesis