STA255H1S: Statistical Theory

Week 12-2: Hypothesis test

Yaoming Zhen

Department of Statistical Sciences

University of Toronto

March 27, 2025

1 / 17
Roadmap

1. Hypothesis test

2 / 17
Type I and II Errors

Example
Suppose the speed limit is 120 km/hr and a device has
recorded the speed of 3 cars. We assume X ∼ N(µ, 4) and we
want to test H0 : µ = 120 vs H1 : µ > 120 with test statistic
T = X1 +X32 +X3 = X̄ .
(a). What values of t will reject H0 for α = 0.05? (z0.05 = 1.64)?

3 / 17
Type I and II Errors

(b). If the police decide to fine drivers going over 121 km/hr,
what is the probability of a tyle I error?

4 / 17
Type I and II Errors

(c). If a car passes at a true speed of 123 km/hr, what is the

probability of a type II error under critical region in (a)?

5 / 17
p-values

The type I error α is also called the significant level

If α is small (usually below 0.05), the decision to reject H0

is convincing, otherwise it is not very convincing.

The p-value is defined as P(the test statistics is at least as

extreme as what we got |H0 )

If p < α, then the experimenter will reject H0 in favor of HA .

6 / 17
p-value for Normal Hypothesis Test

We assume X1 , X2 , ..., Xn ∼ N(µ, σ 2 ), where µ is unknown

and σ is known.

Null hypothesis: µ = µ0 for some specific value µ0

X̄ −µ
Test statistic: Z = σ/ n
√0

Null distribution: Z ∼ N(0, 1)

One-sided p-value (right side): p = P(Z > z|H0 )

One-sided p-value (left side): p = P(Z < z|H0 )

Two-sided p-value: p = P(|Z | > |z|)

7 / 17
Normal Hypothesis Test

Example
Suppose that we have data that follows a normal distribution of
unknown mean µ and known variance σ 2 = 4. Let the null
hypothesis be H0 : µ = 0. Let the alternative hypothesis be H1 :
µ > 0. Suppose we collect the following data: 1, 2, 3, 6, -1. At
significance level α = 0.05, should we reject H0 ?

8 / 17
Normal Hypothesis Test with unknown σ 2

Assume X1 , X2 , ..., Xn ∼ N(µ, σ 2 ) with unknown µ and σ 2

Null hypothesis: µ = µ0 for some specific value µ0

X̄ −µ
Test statistic: T = S/ n
√0

Null distribution: T ∼ t(n − 1)

One-sided p-value (right side): p = P(T > t|H0 )

One-sided p-value (left side): p = P(T < t|H0 )

Two-sided p-value: p = P(|T | > |t|)

9 / 17
Normal Hypothesis Test

Example con’t
Suppose that we have data that follows a normal distribution of
unknown mean µ and unknown variance σ 2 . Let the null
hypothesis H0 be that µ = 0. Let the alternative hypothesis HA
be that µ > 0. Suppose we collect the following data: 1, 2, 3, 6,
-1. At significance level α = 0.05, should we reject H0 ?

10 / 17
Tests for Large Samples

For large sample sizes the distribution of the studentized mean

can be approximated by a standard normal distribution. We still
calculate
x̄ − µ0
t= √
s/ n

But the test can be approximated by P(Z ≤ t) where

Z ∼ N(0, 1)

11 / 17
Two-Sample Equal Variances

Use: Test if the population means from two populations dif-

fer by a hypothesized amount.
Data: X1 , X2 , ..., Xn and Y1 , Y2 , ..., Ym .
Assumptions: Both groups of data are independent normal
samples:

Xi ∼ N(µx , σ 2 )
Yj ∼ N(µy , σ 2 )

where both µx and µy are unknown and possibly different. The

variance σ 2 is unknown, but the same for both groups.

12 / 17
Two-Sample Equal Variances

H0 : For a specified µ0 : µx − µy = µ0

HA : For a specified µ0 : µx − µy ̸= µ0

Test-statistic:
X̄ − Ȳ − µ0
T = q
sp n1 + m1

where sP2 is the pooled sample variance.

(n − 1)sx2 + (m − 1)sy2
sp2 =
n+m−2

13 / 17
Two-Sample Equal Variances

Example
Real data from 1408 women admitted to a maternity hospital for
(i) medical reasons or through (ii) unbooked emergency
admission. The duration of pregnancy is measured in complete
weeks from the beginning of the last menstrual period. Medical:
775 obs. with x̄ = 39.08 and sx2 = 7.77 Emergency: 633 obs.
with ȳ = 39.60 and sy2 = 4.95. Set up and run a two-sample
t-test to investigate whether the duration differs for the two
groups.

14 / 17
Two-Sample Equal Variances

15 / 17
Paired Two-Sample T-Test

Use: Test if the average difference between paired values

in a population equals a hypothesized value.

Data: X1 , X2 , ..., Xn and Y1 , Y2 , ..., Yn must have the same

length.

Assumptions: The differences Wi = Xi − Yi between the

paired samples are independent follow normal distribution
N(µ, σ 2 ), where µ and σ are unknown.

NOTE: This is just a one-sample t-test using Wi .

16 / 17
Paired Two-Sample T-Test

H0 : µ = µ0 .

HA : µ ̸= µ0 .

W̄ −µ
Test statistic: T = √0
S/ n

S 2 is the sample variance:

n
1 X
2
S = (Wi − W̄ )2
n−1
i=1

Null distribution: T ∼ t(n − 1).

17 / 17