© 2021 CFA Institute. All rights reserved.
Steps in Hypothesis Testing
Neil Govier, CFA
Regional Director, APAC
Professional Learning
Lesson Overview
After this lesson, you will be able to:
• Describe the steps in building and
testing a hypothesis
• Apply hypothesis testing and
interpret the outcome of a
hypothesis test
Steps in Hypothesis Testing
Hypothesis testing is, together with sampling and estimation, a branch of statistics known as
statistical inference.
A test sample will be used to test, statistically, the accuracy of a hypothesis.
Steps in Hypothesis Testing:
1) State the hypothesis
2) Identify the appropriate test statistic and its probability
3) Specify the significance level
4) State the decision rule
5) Collect the data and calculate the test statistic
6) Make the statistical decision
7) Make the economic or investment decision
Step 1 – State the Hypothesis
Definition: Null Hypothesis, H0: the hypothesis that is being tested
Definition: Alternative Hypothesis, Ha: the accepted hypothesis if H0 is rejected
Formulation of Hypotheses:
1) H0: Ө = Ө0 versus Ha: Ө ≠ Ө0 (two-tailed)
2) H0: Ө ≤ Ө0 versus Ha: Ө > Ө0 (one-tailed)
3) H0: Ө ≥ Ө0 versus Ha: Ө ˂ Ө0 (one tailed)
Ө = value of population parameter — for example, mean, µ, or variance, σ2
Ө0 = possible value of population parameter
Ex. We believe the population mean is 6%: H0 is µ = 6.0; Ha is µ ≠ 6.0
Note: Ensure H0 has the equality
Step 2 – Identify the Appropriate Test Statistic
Definition: A quantity, calculated based on a sample, whose value is the basis for deciding whether to
accept or reject the null hypothesis:
Test statistic = Sample statistic – Value of the population parameter under H0
Standard error of the sample statistic
Example:
! we are hypothesizing, H0, that the population mean is µ.
Assume, from a sample with mean 𝑋,
𝑠
Recall that the SE of the sample mean 𝑠"! = when population standard deviation is not known.
𝑛
!
"#$
Test statistic =
&⁄ '
The test statistic is how far from the hypothesized mean the sample statistic lies, as a number of
standard errors.
Step 2 – Probability Distribution
What probability distribution should we use?
• z-distribution or t-distribution
Test Concerning a Single Mean
Small Sample (n < 30) Large Sample (n ≥ 30)
Population normal with known variance z-test z-test
Population normal with unknown variance t-test t-test (z-test alternative)
Population non-normal with known variance Not Available z-test
Population non-normal with unknown variance Not Available t-test (z-test alternative)
Source: CFA Institute
Step 3 – Specify the Significance Level
Based on the test statistic, there are two
possibilities: Reject or not reject H0.
A significance level of 5% implies there is
a 5% chance of incorrectly rejecting H0.
There is a 95% chance of correctly
accepting H0.
Step 4 – State the Decision Rule
• z-test
• 5% significance
• 95% confidence
• Two-sided, Ha ≠
2.5% 2.5%
95%
-1.96 1.96
Rejection Region Acceptance Region Rejection Region
test statistic ˂ -1.96 test statistic > 1.96
Data for illustration only.
Step 4 – State the Decision Rule
• z-test
• 5% significance
• 95% confidence
• One-sided, Ha >
5.0%
95%
1.645
Acceptance Region Rejection Region
test statistic > 1.645
Data for illustration only.
Step 5 – Collect Data and Calculate Test Statistic
Collect the data…
Example:
For the period 1900 to 2017 inclusive (118 annual observations), the arithmetic mean equity risk premium
for US stocks, relative to bill returns, was 7.5% per year. The sample standard deviation was 19.5%
n = 118, 𝑋( = 7.5, s = 19.5. Hypothesized value of population mean = 0.
…and calculate the test statistic:
!
"#$
Test statistic =
&⁄ '
7.5 - 0
Test statistic = = 4.18
19.5/ √118
Please note that an error has been corrected on this slide that still exists on the
original shown in the lesson video.
Source: CFA Institute
Step 6 – Make the Statistical Decision
Compare the test statistic to the critical statistic.
For a two-tailed test at 5% significance level, the critical statistic is 1.96.
Because 4.18 > 1.96, the statistical decision is to reject H0 and accept that there is positive risk premium on
US stocks.
There is only a 5% probability that this sample came from a population with a mean of zero.
Note: The decision is whether to reject or not reject the null
hypothesis, not to accept or not accept the alternative hypothesis.
Step 7 – Make the Economic/Investment Decision
Although there is evidence to support a
risk premium, such other factors as the
economic cycle, investors’ attitude
toward risk, required portfolio liquidity,
and so on should be taken into account
when making the investment decision.
Hypothesis Testing
A company, believing that its receivables payment days average 45 or less, has taken a random sample of
50 accounts and calculated that the mean payment time of the sample is 49 days with a standard deviation
of 8 days. Test the hypothesis, at the 5% significance, that the payment time is greater than 45 days.
Solution, using the z-test:
H0 ≤ 45, Ha > 45
49 - 45
Test statistic = = 3.536
8/√50
Critical statistic (z-value) = 1.645
Because the test statistic (3.536) > critical statistic (1.645), the decision is to reject H0 and conclude that
payment days exceed 45 days.
Please note that an error has been corrected on this slide that still exists on the
original shown in the lesson video.
Summary
• Know and follow the steps
involved in a hypothesis test
• Carry out a hypothesis test and
decide whether to reject or not
reject the null hypothesis
© 2021 CFA Institute. All rights reserved.
Up Next: Review This Lesson and
Complete the Assignment before
Moving on to the Next Lesson