Scribd
Upload
73 views25 pages

Mas S Mohktar Email: Mas - Dayana@um - Edu.my Phone (Office) : 0379677681

The document discusses two types of errors that can occur in hypothesis testing: 1. Type I error (rejecting the null hypothesis when it is true) which is determined by the significance level α. 2. Type II error (failing to reject the null hypothesis when it is false) which is represented by β. It provides an example calculating the type II error β = 0.042 for a one-sided hypothesis test comparing male cholesterol levels to a null hypothesis. The power of a test is 1 - β, the probability of avoiding a type II error. Sample size and significance level can be adjusted to balance type I and type II errors.

Uploaded by

Sam
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
73 views25 pages

Mas S Mohktar Email: Mas - Dayana@um - Edu.my Phone (Office) : 0379677681

The document discusses two types of errors that can occur in hypothesis testing: 1. Type I error (rejecting the null hypothesis when it is true) which is determined by the significance level α. 2. Type II error (failing to reject the null hypothesis when it is false) which is represented by β. It provides an example calculating the type II error β = 0.042 for a one-sided hypothesis test comparing male cholesterol levels to a null hypothesis. The power of a test is 1 - β, the probability of avoiding a type II error. Sample size and significance level can be adjusted to balance type I and type II errors.

Uploaded by

Sam
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 25

Mas S Mohktar

Email: [email protected]
Phone (office): 0379677681
 Types of Error
 Two kinds of errors can be made when we conduct a test of
hypothesis
 The first one is called a type I error
 Happens when H0 is true and we reject the null hypothesis, where
H0 : μ = μ0
 The probability of making a type I error is determined by the
significance level of the test; recall that α = P(reject H0|H0 is
true)
 If we were to conduct repeated and independent tests of
hypotheses setting the significance level at 0.05, we would
erroneously reject a true null hypothesis 5% of the time
 The second kind of error that can be made during a hypothesis
test is a type II error
 A type II error is made if we fail to reject the null hypothesis H0 :
μ = μ0, when H0 is false
 The probability of committing a type II error is represented by
the Greek letter β, where
 β = P(do not reject H0|H0 is false).
 If β = 0.10, for instance, the probability that we do not reject the
null hypothesis when μ ≠ μ0 is 10%
 The two types of errors that can be made are summarized
below
 The mean serum cholesteric levels for all 20- to 74-yr-old
males in US is μ = 211 mg/100 ml and the standard deviation is
σ = 46 mg/100 ml
 If we do not know the true population mean but we know the
mean serum cholesteric levels for the subpopulation of 20- to
24-yr-old males is 180 mg/100 ml
 If we were to conduct a one-sided test:
 H0 : μ ≤ 180 mg/100 ml v.s. HA : μ > 180 mg/100 ml, at the α =
0.05 level of significance
 What is the probability of the type II error associated with such
a test, assuming that we select a sample of size 25?
 To determine this, we first need to find the mean serum
cholesterol level our sample must have for H0 to be rejected
x−μ0 x−180
TS: = = 1.645
σ n 46 25

 Solving for x, we have


1.645 × 46
x = 180 + = 195.1
25
 As shown in the figure, the area to the right of x = 195.1
corresponds to the upper 5% of the sampling population of
means of samples of size 25 when μ = 180
 Recall that the probability of making a type II error (i.e. β) is
the probability rejecting the null hypothesis given that H0 is
false
 Therefore, it is the chance of obtaining a sample mean that is
less than 195.1 mg/100 ml given that the true population mean
is not 180 mg/100 ml but is instead 211 mg/100 ml
 Since a sample mean less than x = 195.1 mg/100 ml
195.1 − 211
z= = −1.73
46 25
 The area under the standard normal curve that lies to the left of
z = -1.73 is 0.042
 Therefore, β = 0.042 when the true population mean is 211
mg/100 ml
 While the type I error α is determined by looking at the case in
which H0 is true (i.e. μ = μ0), the type II error β is found by
considering the situation in which H0 is false (i.e. μ ≠ μ0)
 If μ is not equal to μ0, however, there are an infinite number of
possible values that μ could assume
 The type II error is calculated for a single such value, μ1
 If we had chosen a different alternative population mean, we
would have computed a different value for β
 The closer μ1 is to μ0, the more difficult it is to reject the null
hypothesis
 The power of the test of hypothesis is the probability of
rejecting the null hypothesis when H0 is false, which is 1 – β
 In other words, it is the probability of avoiding a type II error:
power = P(reject H0|H0 is false) = 1 − β
 Like β, the power must be computed for a particular alternative
population mean μ1
In the serum cholesterol example in the previous section, the
power of the one-sided of hypothesis is

Power = P(reject H0 : μ ≤ 180|μ = 211)


= P(X ≥ 195.1|μ = 211)
= P(Z ≥ −1.73)
= 1 − P(Z > 1.73)
= 1 − 0.042
= 0.958
 If we were to plot the values of 1 - β against all possible alternative
population means, we would end up with what is known as a power
curve as illustrated
 The power of the test approaches 1 as the alternative mean moves
further and further away from the null value
 In most practical applications, a power less than 80% is considered
insufficient
 One way to increase the power of a test is to raise the significance
level α
 If α had been equal to 0.10 for the test of the null hypothesis
 H0 : μ ≤ 180 mg/100 ml, the power becomes 0.982 as illustrated
 The trade-off between α and β is similar to that observed to exist
between the sensitivity and the specificity of a diagnostic test
 The only way to diminish α and β simultaneously is to reduce
the amount of overlap in the two distributions – the one
centered at μ0 and the one centered at μ1
 An alternative is to increase the sample size n
 By increasing n, we decrease the standard error 𝜎 𝑛
 This causes the two distributions to become more narrow,
which, in turn, lessens the amount of overlap
 Another options that we have not yet mentioned is to find a
“more powerful” test statistic
 If we conduct a one-sided test of the null hypothesis
 H0 : μ ≤ μ0 v.s. HA : μ > μ0
 at the α level of significance, H0 would be rejected for any test
statistic that takes a value z ≥ zα
 Similarly, considering the desired power of the test 1 - β, the
generic value of z that corresponds to a probability β is z = −zβ
 The two different expressions for x are
σ
x = μ0 + zα
n
and
σ
x = μ1 + zβ
n
and setting them equal to each other gives us
2
σ(zα + zβ )
n=
μ1 − μ0
 This is the sample size necessary to achieve a power of 1 - β
when we conduct a one-sided test at the α level
 Several factors influence the size of n
 If we reduce the type I error or type II error, this would produce
a larger value of n
 The difference μ1 −μ0 decreases and the sample size increases
 Finally, the larger the variability of the underlying population σ,
the larger the sample size required
 If we conduct a two-sided test, the sample size necessary to
achieve a power of 1 - β at the α level is
2
σ(zα 2 + zβ )
n=
μ1 − μ0

 Note that the sample size for a two-sided test is always larger
than the sample size for the corresponding one-sided test

You might also like