Skip to main content Skip to course navigation

Hypothesis Test Selection

Learning Objective

By the end of this lesson, the learner should be able to:

  • Explain the purpose of hypothesis testing.
  • Identify the information required when selecting a statistical test.
  • Understand the importance of data type, groups, sample size and distribution.
  • Recognize appropriate hypothesis-testing tools.
  • Interpret statistical significance in the context of process improvement.

1. Introduction

Hypothesis testing helps determine whether observed differences provide sufficient statistical evidence of a difference.

For a Black Belt, hypothesis testing provides a structured method for evaluating process questions rather than relying only on visual impressions or assumptions.

The appropriate test depends on factors such as:

  • Data type
  • Number and structure of groups
  • Sample size
  • Distribution and applicable assumptions

2. Start with the Data Type

The first step is to understand the type of response data being analysed.

Common categories include:

  • Continuous data — measurements such as time, weight or temperature.
  • Categorical data — classifications such as pass/fail or defect/no defect.

The data type helps determine which statistical methods may be appropriate.

3. Consider the Groups

The number and structure of groups influence test selection.

For example:

  • Comparing two groups requires a different approach from comparing several groups.
  • The relationship between observations also matters when selecting a method.

A Black Belt should therefore clearly identify what groups or conditions are being compared before selecting a test.

4. Consider Sample Size and Distribution

Test selection should also consider:

  • Sample size
  • Distribution of the data
  • Applicable statistical assumptions

Distributional assumptions can affect whether a particular statistical method is appropriate.

The Black Belt should therefore examine the data before choosing the test rather than selecting a test simply because it is familiar.

5. Statistical Significance

A statistical test evaluates evidence concerning a null hypothesis.

A statistically significant result indicates evidence against the null hypothesis under the selected method and its assumptions.

However, statistical significance should be considered together with the practical importance of the observed difference.

A result may be statistically significant while the actual process difference is too small to have meaningful operational or business impact.

6. Tools & Techniques

The source identifies these hypothesis-testing tools:

  • t-tests
  • Chi-square tests
  • ANOVA
  • Non-parametric tests
  • Minitab

The choice depends on the characteristics of the data and the question being investigated.

7. Application Example

Comparing Two Production Shifts

A company wants to determine whether defect performance differs between two production shifts.

The Black Belt first considers:

  1. What type of data is being analysed?
  2. How many groups are being compared?
  3. What is the sample size?
  4. What distributional assumptions apply?
  5. Which statistical test is appropriate?

An appropriate two-sample test can then be selected to evaluate the observed difference.

8. Case Study

Textile Mill — Night-Shift Defects

A textile mill investigated whether the difference in defect performance between shifts was statistically meaningful.

The analysis found that the night-shift defect difference was statistically significant.

The organisation then introduced additional supervision, resulting in a 15% reduction in night-shift defects.

Black Belt Learning

The statistical analysis helped establish evidence of a difference. The improvement action then addressed the process condition associated with the observed performance.

This illustrates the connection between:

Data → Appropriate Test → Statistical Evidence → Process Action → Improvement

9. Black Belt Perspective

A Black Belt should not begin with:

“Which statistical test do I know?”

Instead, begin with:

“What question am I trying to answer, and what does my data look like?”

Then consider:

Data Type → Groups → Sample Size → Distribution/Assumptions → Appropriate Test → Interpretation

This disciplined approach helps prevent inappropriate statistical analysis.

10. Lesson Practice

  1. What information should be checked before choosing a statistical test?
  2. Why does data type influence test selection?
  3. Why does the number of groups matter?
  4. Why should distributional assumptions be considered?
  5. Why can a statistically significant result still be practically unimportant?
  6. When might a non-parametric method be considered?

11. Key Learning Points

  • Hypothesis testing evaluates whether observed differences provide sufficient statistical evidence of a difference.
  • Test selection depends on data type, groups, sample size and distribution.
  • Statistical significance should be interpreted under the selected method and its assumptions.
  • Relevant tools include t-tests, chi-square tests, ANOVA, non-parametric tests and Minitab.
  • Statistical evidence should lead to appropriate process investigation and action.
  • Statistical significance and practical importance are not necessarily the same.

12. Lesson Conclusion

Hypothesis test selection is a critical Black Belt skill because the quality of the conclusion depends on using an appropriate statistical method for the data and question.

The essential sequence is:

Define the Question → Understand the Data → Select the Appropriate Test → Evaluate Evidence → Interpret the Result → Take Appropriate Action