Retired course

This course has been retired and is no longer supported.

About this lesson

The scientific method of analysis is to create an hypothesis, develop experiments that generate applicable data, then analyze that data to prove or disprove the hypothesis. This approach allows us to confidently answer inquiry questions with data. This lesson explains the concepts of hypotheses in problem solving.

Quick reference

Concept of Hypothesis Testing

Hypothesis testing is one of the primary analytical techniques used at various stages of the Lean Six Sigma process.  This lesson introduces the basic concepts of hypothesis testing and relates it to the more general scientific method of inquiry.

When to use

The concepts of hypothesis testing are used whenever hypothesis testing is used.  Lean Six Sigma projects use hypothesis testing in the Analyze phase to identify and prioritize potential root causes of the problem.  It is also used in the Improve phase to demonstrate the effectiveness of proposed solution.

Instructions

Hypothesis testing has it roots in the scientific method.  This is an approach for organized inquiry.  A question is posed, based upon the question, a hypothesis is created.  To prove or disprove the hypothesis, an experiment is developed.  The experimental method is proved valid or it is changed.  When the experimental method is ready, data is collected and then analyzed.   If the data is inconclusive, often the hypothesis must be revised and the process repeated.  But when the data is conclusive, the conclusion will be to reject or accept the hypothesis as an answer to the question.

Hypothesis testing is an integral part of the scientific method.  It has also become associated with the Lean Six Sigma methodology as a means to demonstrate with data and statistical confidence the existence of root causes for the problem being investigated.  It is also used to demonstrate with data and statistical confidence that a proposed improvement is effective.  The hypothesis is a set of statements that essentially answer the question, “Is there a real difference between this data set and that data set?”  If the difference is real, then we have isolated a possible root cause or demonstrated that the improvement makes a difference.  Hypothesis testing allows us to use data to demonstrate with statistical confidence something that we think we already know.   We don’t need to rely upon lucky guesses or intuition, we can use data when forming our conclusions.

Within Lean Six Sigma, hypothesis testing is used in the Analyze phase to isolate root causes for the problem being investigated.  Many times the hypothesis test will use the data gathered in the Measure phase.  The hypothesis consists of two opposing statements called the Null hypothesis and the Alternative hypothesis.  Through the analysis, the statistics developed directs you to either reject the Null hypothesis or fail to reject the Null hypothesis.  If the hypothesis is supporting a Lean Six Sigma project during the Analyze phase, the Null hypothesis might state that there is no difference in the quality of products built on production line A as compared to production line B.  The Alternative hypothesis would state that there is a difference in the quality of products built on production line A as compared to production line B.  Using data from each line, the hypothesis test would determine which statement is most likely to be true. 

During the Improve phase of the Lean Six Sigma project, a similar approach is used, only now it is to demonstrate that the solution is effective.  The Null hypothesis would state that the solution makes no difference in the quality of the products produced.  And the Alternative hypothesis would state that the solution has improved the quality of the products being produced.  

Hints & tips

The hypothesis test will tell us if the changes we see in two data or more data sets is statistically significant.  While it may be obvious that one data set has a higher or lower value than the others, that may just be to random chance.  The hypothesis test tells us if the difference is statistically significant.

Clarify the question you want to answer.  This will make writing the hypotheses statements easier.

Login to download
  • 00:04 Hi, I'm Ray Sheen.
  • 00:06 I'd like to welcome you to this course on hypothesis testing.
  • 00:09 These tests allow us to use data to demonstrate statistical confidence
  • 00:14 whether or not a statement is true.
  • 00:16 But let's start by considering some of the concepts that are embedded
  • 00:20 in hypothesis testing.
  • 00:23 The first concept is the scientific method of deduction.
  • 00:26 We don't just rely on hunches or lucky guesses.
  • 00:29 Instead, we go through a process to answer a question or inquiry.
  • 00:34 So this is the first step, ask a question.
  • 00:37 On the lean six sigma project, the question often involves
  • 00:39 determining if something is a contributing cause to a problem or whether we can
  • 00:44 demonstrate that a change has made a real difference in process performance.
  • 00:48 Based upon the question, we write a hypothesis, which is a statement
  • 00:52 about the data that we intend to prove or disprove in order to answer the question.
  • 00:57 With a clear hypothesis, it becomes easy to determine
  • 01:00 what type of data is needed to prove or disprove the hypothesis.
  • 01:04 This then dictates the type of experiment or data collection that is needed.
  • 01:09 Once the experiment is developed,
  • 01:10 you should always check that it's working as expected and provides valid data.
  • 01:15 If not, change the experiment.
  • 01:17 Next, collect the data.
  • 01:19 Depending upon the experiment or the data collection process,
  • 01:22 this is either trivial or the longest and most difficult portion of the analysis.
  • 01:26 Now analyze the data using the appropriate hypothesis test.
  • 01:30 We'll discuss later how to choose which test to use.
  • 01:33 Based upon the results of the analysis you may need to revise your hypothesis and
  • 01:38 repeat the process.
  • 01:39 Eventually you're able to draw a conclusion and
  • 01:41 then can answer the question with confidence.
  • 01:45 Let's take a moment to consider those first two steps, the question and
  • 01:48 hypothesis.
  • 01:50 The question is the goal or purpose of the analysis.
  • 01:52 If there is no question there is no need to do a hypothesis test,
  • 01:56 unless it's just to show your statistical skills.
  • 01:59 I like to write the question in a cause and effect format.
  • 02:02 Often I'm pulling this right from my fishbone or Ishikawa diagram.
  • 02:06 As I ask the question,
  • 02:08 does this potential root cause have any effect on the problem I'm observing?
  • 02:12 Or in the improve phase, the question maybe, did the change to the process make
  • 02:17 any real difference to the process performance?
  • 02:20 The hypothesis is then written to be able to answer the question.
  • 02:23 The hypothesis normally has two parts.
  • 02:26 There are statements which are typically opposites.
  • 02:29 One statement, the null hypothesis, says that the process response or the item
  • 02:34 being investigated is not being impacted by the independent variable or factor.
  • 02:39 The alternative statement is essentially the opposite.
  • 02:41 The independent factor does have an impact on the process response and
  • 02:46 this is the type of impact it has.
  • 02:48 This of cause begs the question, how do we test hypothesis to know which is true?
  • 02:53 The hypothesis testing approach gives us position that we can use
  • 02:57 to test a claim about the characteristics of a data set.
  • 03:01 We take a portion, or a sample, of the days points in the data set, and
  • 03:05 test the assertion or claim against that subset of data.
  • 03:08 Based upon the results,
  • 03:09 we can make inferences about that claim on the entire data set.
  • 03:13 The nature of the statistical hypothesis test
  • 03:16 is that they allow us to answer practical questions relating to our process or
  • 03:20 problem with statistical confidence.
  • 03:23 We might observe a difference in the data, but
  • 03:25 that difference may just be due to random effects.
  • 03:28 With these tests, we can know if there's a real difference between this condition and
  • 03:32 that condition.
  • 03:33 Hypothesis testing is normally used in the analysis based to confirm our suspicions
  • 03:39 or in the improved phase to demonstrate the impact of an improvement.
  • 03:42 It's not often used for discovery.
  • 03:45 That is because I'm gonna start with the assertion that there is tested.
  • 03:48 If I don't know what to assert, I don't know how to get started.
  • 03:52 Another way of saying this is that we use hypothesis testing to demonstrate what we
  • 03:56 think we know.
  • 03:58 It's no longer a hunch.
  • 03:59 It's now a demonstrated statistical fact, which brings us to when in the problem
  • 04:05 solving process is the appropriate time to use hypothesis testing.
  • 04:09 As I've already alluded,
  • 04:10 it can be used in several phases of a Lean Six Sigma project.
  • 04:14 During the analysis phase it can be used to demonstrate
  • 04:17 whether a potential root cause is significant or not.
  • 04:20 Also during the analysis phase,
  • 04:21 it can be used to stratify different subsets of data and
  • 04:24 isolate the likely root cause condition to determine where action is needed.
  • 04:29 It can also be used in the improve phase to demonstrate
  • 04:32 whether a change made a significant impact on the process performance.
  • 04:37 We use hypothesis testing to allow us to ask a question and
  • 04:41 then confidently answer it with data.
  • 04:44 We use the data to demonstrate what we think we already know.

Lesson notes are only available for subscribers.

PMI, PMP, CAPM and PMBOK are registered marks of the Project Management Institute, Inc.

Gift this course