13 - 15: Statistical Inference - Single Population (2)∗

Dr. Bhanu Pratap Singh, IIM Rohtak, 2022

Learning Objectives
• Hypotheses Testing
• Testing Hypothesis about a Population µ using z Statistics (σ Known)
• Testing Hypothesis about a Population µ using t Statistics (σ Unknown)
• Testing Hypothesis about a Population Proportion
• Testing Hypothesis about a Population Variance
• Solving for Type II Errors

Introduction
• A foremost statistical mechanism for decision making is the hypothesis test.

• The use of statistics to “prove”or “disprove” claims hinges on it.

Introduction
• Hypothesis Tests (HT): For One Sample

• Mean
– σ Known: z HT for µ.
– σ Unknown: t HT for µ

• Proportion
– z HT for p

• Variance
– χ2 HT for σ 2

Hypotheses Testing
• Hypotheses are tentative explanations of a principle operating in nature.
– Types of Hypotheses
– HTAB System to Test Hypotheses
– Rejection and Non Rejection regions
– Type I & Type II Errors
∗ Reference: KB, Chapter 9

1
Hypotheses Testing: Types of Hypotheses

• Research Hypotheses
• Statistical Hypotheses
• Substantive Hypotheses

Hypotheses Testing: Types of Hypotheses

• Research Hypotheses
– A statement of what the researcher believes will be the outcome of an experiment or a
study.
– Before studies are undertaken, business researchers often have some idea or theory
regarding the outcome.

Hypotheses Testing: Types of Hypotheses

• Research Hypotheses: Examples

– Older workers are more loyal to a company.
– The implementation of a Six Sigma quality approach in manufacturing will result in
greater productivity.
– Airline company stock prices are positively correlated with the volume of OPEC oil
production.

• However, to formally test research hypotheses, it is generally best to state them as statistical
hypotheses.

Hypotheses Testing: Types of Hypotheses

• Statistical Hypotheses
– A more formal hypothesis structure.
– To prove or disprove research hypotheses, researchers convert their research hypotheses
to statistical hypotheses.
– And then test the statistical hypotheses using standard procedures.
– Consist of two parts, a null hypothesis and an alternative hypothesis.
– These two parts are constructed to contain all possible outcomes of the experiment or
study.
– Null Hypothesis: The “null” condition exists; that is, there is nothing new happening.
– Alternative Hypothesis: The new theory is true, something new is happening. Symbol:
H0 = Null, Ha : Alternate.

Hypotheses Testing: Types of Hypotheses

• Statistical Hypotheses: Example

– Suppose flour by a manufacturer is sold by weight; and a particular size of package is
supposed to average 40 ounces.
– Test to determine whether packaging process is out of control as determined by the
weight of the flour packages.
– The Null hypothesis: The average weight of the flour packages is 40 ounces (no problem).
– The Alternative hypothesis: The average is not 40 ounces (process is out of control).

2
Hypotheses Testing: Types of Hypotheses
• Statistical Hypotheses: Example
– Suppose a company has held an 18% share of the market.
– However, company officials believe that it is now greater than 18%, and would like to
prove it.
– The null hypothesis is that the market share is still 18% or perhaps it has even dropped
below 18%.
– Converting the 18% to a proportion and using p
– H0 : p = 0.18 & Ha : p > 0.18.
– Even though the “less than” sign is not included in the null hypothesis, it is implied
that it is there.
– Thus equal sign, generally, only for Null Hypothesis.

Hypotheses Testing: Types of Hypotheses

• Statistical Hypotheses
– One could say that the null and alternative hypotheses are mutually exclusive (no
overlap) and collectively exhaustive (all cases included).
– Statistical hypotheses are written so that they will produce either a one-tailed or a two-
tailed test.

Hypotheses Testing: Types of Hypotheses

• Statistical Hypotheses: One Tailed Tests
– Always directional, and the alternative hypothesis uses either the > or the < sign.
– A one-tailed test should only be used when the researcher knows for certain that the
outcome of an experiment is going to occur only in one direction
– Or the researcher is only interested in one direction of the experiment as in the case of
the market share problem.
– In one-tailed problems, the researcher is trying to “prove” that something is older,
younger, higher, lower, more, less, greater, and so on.

• Statistical Hypotheses: Two Tailed Tests

– Always use = and ̸= signs in the statistical hypotheses and are directionless.

Hypotheses Testing: Types of Hypotheses

• Substantive Hypotheses
– Null rejected: Common to say that a statistically significant result has been obtained.

• Statistical Significance
– Market Share: Suppose a large sample is taken, and a sample market share of 18.2% is
obtained.
– Suppose further that a statistical analysis of these data results in statistical significance.
– We would conclude statistically that the market share is significantly higher than 18%.
– It actually means that it is unlikely that the difference is because of chance.
– However, to the business decision maker, a market share of 18.2% might not be signifi-
cantly higher than 18%.
– Hence, Statistical Significance does not always translate to Economic Significance.
– Substantive Result: When the outcome produces results that are important to the
decision maker.

3
Hypotheses Testing: HTAB System

• Hypothesize
• Test
• Take Statistical Action
• Determine the Business Implication

Hypotheses Testing: HTAB System

• Hypothesize: 1. Establish Null and Alternate hypothesis.

– Identify, whether the hypotheses are one tailed or two tailed.
– In hypothesis testing, it is always assumed that the null hypothesis is true at the be-
ginning of the study.

Hypotheses Testing: HTAB System

• Test: 2. Determine the appropriate statistical test.

– Consider the type, number, and level of data being used.
– Statistic used in the analysis (µ, p, σ 2 , etc.).
– Consider the assumptions underlying statistical tests.

• Test: 3. Set the value of α, the Type I error rate.

– α is the probability of committing a Type I error.
– Common values of alpha include 0.05, 0.01, 0.10, and 0.001.

Hypotheses Testing: HTAB System

• Test: 4. Establish a decision rule.

– Using α and the test statistic, critical values can be determined.
– These critical values are used to determine whether the null hypothesis is rejected or
not.
– If the p-value method is used, the value of α is used as a critical probability value.

Hypotheses Testing: HTAB System

• Test: 5. Gather sample data.

– Might include survey, random sampling, or even sampling from secondary data sources.
– Care should be taken in establishing a frame, determining the sampling technique, and
constructing the measurement device.
– A strong effort should be made to avoid all non sampling errors.

• Test: 6. Analyze the data.

– After the data are sampled, the test statistic can be calculated.

4
Hypotheses Testing: HTAB System

• Take Statistical Action: 7. Reach a statistical conclusion.

– Using decision rule and the value of the test statistic, draw a statistical conclusion.
– Conclude whether the null hypothesis is rejected or is not rejected.

• Determine Business Implication: 8. Make a Business Decision.

– Determine Business significance (Economic significance).

Hypotheses Testing: Rejection & Non Rejection Region

• Rejection Region: Statistical outcomes that result in the rejection of the null hypothesis.

• Non Rejection Region: Statistical outcomes that fail to reject the null hypothesis.

• When is the sample mean so far away from the population mean that the null hypothesis is
rejected?

• The critical values are used to answer this question.

Hypotheses Testing: Rejection & Non Rejection Region

Hypotheses Testing: Type I & Type II Errors

• Type I Error
– Rejecting a true null hypothesis.
– Because of possibility of selecting a extreme sample by chance.
– The rejection regions represent the possibility of committing a Type I error.
– If the null hypothesis is true, any mean that falls in a rejection region will result in a
decision that produces a Type I error.
– The probability of committing a Type I error is called α or level of significance.
– The value of α is always set before the experiment or study is undertaken.

5
Hypotheses Testing: Type I & Type II Errors
• Type II Error
– Fail to reject a false null hypothesis.
– The probability of committing a Type II error is β.
– Unlike α, β is not usually stated at the beginning of the hypothesis testing procedure.
– Because β occurs only when the null hypothesis is not true, the computation of β varies
with the many possible alternative parameters that might occur.

Hypotheses Testing: Type I & Type II Errors

• How α and β are related.
– α can only be committed when the null hypothesis is rejected.
– β can only be committed when the null hypothesis is not rejected.
– Hence both can’t be committed at the same time on the same hypothesis test.
– Generally, α and β are inversely related.
– One way to reduce both errors is to increase the sample size.
– Power: 1 − β. Probability of a statistical test rejecting the null hypothesis when the
null hypothesis is false.

Hypotheses Testing: Type I & Type II Errors

• How α and β are related.

Testing Hypothesis: µ using z (σ Known)

• z test for Single Mean.

• Using p − value to Test Hypothesis.

• Using Critical Value Method to Test Hypothesis

Testing Hypothesis: µ using z (σ Known)

• z test for Single Mean.
x̄−µ
– z = (σ/ √
n)
– Example: A survey of CPAs found that the average net income is $74,914.
– By taking a random sample of 112, determine whether the net income figure changed.
– The researcher could use the eight steps of hypothesis testing to do so.
– Assume the population standard deviation is $14,530.

6
Testing Hypothesis: µ using z (σ Known)

• z test for Single Mean.

• Hypothesize: H0 : µ = 74, 914, Ha : µ ̸= 74, 914.

• Determine the appropriate statistical test.

– Because σ is known and the researcher is using the sample mean as the statistic, z
statistic is appropriate.

• Pre specify the Type I error rate (α): 0.05.

– Two tailed test. Thus, the rejection region is 2.5% of the area in each end.
– Hence, the critical z value is zα/2 = ±1.96.

• Decision Rule
– If the data gathered produce a z value ≥ 1.96 or ≤ - 1.96, reject the Null Hypothesis.

• Gather Data & Compute z.

– x̄ = 78, 695, n = 112, hypothesized µ = 74,914.
– z = 2.75.

Testing Hypothesis: µ using z (σ Known)

• Action
– Because calculated test statistic (2.75) is greater than the critical value (1.96), reject
the Null hypothesis.
– The calculated test statistic is often referred to as the observed value.

• Business Implication
– The evidence gathered indicates that the national average may have increased.
– May mean that CPAs will be more expensive to hire either as full-time employees or as
consultants.

Testing Hypothesis: µ using z (σ Known)

• For Finite Population

– Use of Finite Correction Factor

Testing Hypothesis: µ using z (σ Known)

• Using p − value to Test Hypothesis.

– p − value: Sometimes referred to as observed significance level.
– No preset value of α is given.
– The p − value defines the smallest value of α for which the null hypothesis can be
rejected.
– For example, if the p − value of a test is .038, the null hypothesis cannot be rejected at
α = .01.
– However, the null hypothesis can be rejected for α = .05.

7
Testing Hypothesis: µ using z (σ Known)

• Using Critical Value Method to Test Hypothesis.

– Determines the critical mean value required for z to be in the rejection region and uses
it to test the hypotheses.
x¯c −µ
– zc = (σ/ √
n)
– zc = ± 1.96, µ = 74,914, n = 112. Calculate critical mean values.

Testing Hypothesis: µ using z (σ Known)

• Using Critical Value Method to Test Hypothesis.

– With the critical value method, most of the computational work is done ahead of time.
– Before the sample means are computed, the analyst knows the range of sample mean value to
reject Null.

Testing Hypothesis: µ using t (σ Unknown)

• Very often the value of the population standard deviation is unknown.

• Use the sample standard deviation as an estimate of it.

x̄−µ
• t test for µ: t = √
(s/ n)

Testing Hypothesis: µ using t (σ Unknown)

• Example
– The machine that produces plates is set to yield plates that average 25 pounds.
– Test to determine whether the machine is out of control.
– H0 : µ = 25 pounds, Ha : µ ̸= 25 pounds.
– Collect sample & calculate sample statistics. x̄ = 25.51, s = 2.1933, n = 20.
– Set α = 0.05. Calculate critical t values.
– The t distribution table is a one-tailed table but the test for this problem is two tailed,
so alpha must be split.
– t = ± 2.093
– Calculate observed t value: 1.04.

8
Testing Hypothesis: µ using t (σ Unknown)

Testing Hypothesis about a Population Proportion

• z test: z = √ p̂−p
(p.q)/n

• Example
– A manufacturer believes exactly 8% of its products contain at least one minor flaw.
– Suppose a company researcher wants to test this belief.
– The null and alternative hypotheses are H0 : p = 0.08 & Ha : p ̸= 0.08.
– Set α = 0.10, z = ± 1.645.
– Select sample and calculate observed z.
– Randomly selects a sample of 200 products, 33 items have at least one minor flaw.
– Observed z = 4.43.

Testing Hypothesis about a Population Variance

• Assumes a normally distributed population.

• The chi-square test of a population variance is extremely sensitive to violations of this

assumption.

(n−1)s2
• χ2 = σ2

• Example
– Determine whether the σ 2 is greater than 4. One Tailed.
– H0 : σ 2 = 4 & Ha : σ 2 > 4.
– α = 0.05, Sample size = 8. Critical values.
– χ20.05,7 = 14.0671.
– s2 = 20.9821.
– Observed χ2 value: 36.72

9
Solving for Type II Errors

• Fail to reject the null hypothesis: Either a correct decision or a Type II error.

• If the null hypothesis is true: Correct decision. If the null hypothesis is false: Type II error.

• A Type II error, β, varies with possible values of the alternative parameter.

Solving for Type II Errors: Example

• Assume H0 : µ = 12 ounces & Ha : µ < 12 ounces.

• A Type II error: Fails to reject the null hypothesis and the null hypothesis is false.

• If the null hypothesis, µ = 12 ounces, is false, what is the true value for the population
mean?

• Is the mean really 11.99 or 11.90 or 11.5 or 10 ounces?

• For each, the probability of committing a Type II error can be calculated.

• Suppose x̄ = 11.985, n = 60, α = 0.05, and σ = 0.10. Critical value: z = - 1.645.

• Observed z value = - 1.16. Do not reject.

• Either a correct decision or a Type II error.

Solving for Type II Errors: Example

• What is the probability of committing a Type II error in this problem if the population
mean actually is 11.99?
– The first step is to calculate a critical value for the sample mean.
– x¯c = 11.979. If the sample mean is below this value, null will be rejected.
– If µ actually is 11.99, what is the probability of failing to reject µ = 12 when 11.979
ounces is the critical value?
– Calculate z1 value using x¯c = 11.979 and µ = 11.99.
– z1 = -0.85. This value yields an area of .3023.
– The probability of committing a Type II error is all the area to the right of xc = 11.979
in distribution with µ = 11.99.
– .3023 + .5000 = .8023.

10
Solving for Type II Errors: Example

Solving for Type II Errors: Example

• Recompute the probability of committing a Type II error if the alternative mean is 11.96 ounces.

11
Solving for Type II Errors: Example

Solving for Type II Errors

• If the alternative value is relatively far from the hypothesized value, the probability of
committing a Type II error is smaller than it is when the alternative value is close to the
hypothesized value.

• This situation is shown graphically in operating characteristic curves and power curves.

Solving for Type II Errors

• Operating characteristic curves and power curves.

12
Solving for Type II Errors

• Operating Characteristic Curve

Solving for Type II Errors

• Power Curve

Solving for Type II Errors

• Effect of Increasing Sample size.

– Change in critical values.
– Previous Example: x̄ = 11.985, n = 60, α = 0.05, and σ = 0.10. Critical value: z = -
1.645.
– Change n = 100. Calculate new x¯c .
– x¯c = 11.984.
– Calculate z1 value using x¯c = 11.984 and µ = 11.99.
– z1 = -0.60.

13
Solving for Type II Errors

• Effect of Increasing Sample size.

14