Analysis - Statistics

24 important questions on Analysis - Statistics

Why do we use statistics in the Analysis phase?

The graphs indicate a possible outcome, but how do you know if that outcome is correct?  It is checked by using statistics like Confidence Interval.

What is an Confidence interval?

A specified range within which the true process statistic will fall

Translate to six sigma speak:
How confident do you want to be in your decision?

Set your alpha level
  • Higher grades + faster learning
  • Never study anything twice
  • 100% sure, 100% understanding
Discover Study Smart

Translate to six sigma speak:
Select and run the test

Calculate P value based on your data

Translate to six sigma speak:
Decide if your theory was right or not

Accept or reject null hypothesis

How does the Alternative Hypothesis always start?

With the sentence: There IS a difference

Hypothesis test for averages - routemap
Are the samples normally distributed? What do you do when you are not sure?

Use the Tukey's quick test

Which test for one sample?

1 Sample T test

Does the data need to be normally distributed for a T test?

In theory yes, in a practical sense the data can be roughly normal

Do the two samples have to be the same size?

No

What do we do when P value is lower than 0.05?

We reject Null hypothesis, because there is a difference.

When do you use a Paired T test?

When you want to compare two samples that has data linked in pairs.

How does the 1 sample Sign test work?

It allows you to compare the median of just one sample against a known median value, such as an industry benchmark or well established historical median.

How does the Moods Median Test work?

It compares the medians (central position) of different samples of data, where the samples are not Normally distributed and where there are obvious outliers in the data samples.

What is Chi Square similar to?

One way ANOVA, basically the same. Chi Square is used for proportions and percentages.

With one process output, what do you use to graph the data?

Scatter plot

With many process inputs, what do you use to check correlations?

Use Pearson Coefficient, and decide which factors to include in the regression. IF you are in doubt include them.

What do you use for Simple Line Regression?

Minitab: Fitted Line Plot

Pearson Coefficient: Interpreting P values with 95% confidence level
What happens when the P value is higher than 0.05?

It is possible that NO correlations exists.

What is Linear Regression?

Linear means in a straight line, diagonal.

What is Quadratic regression?

Shaped like a parabola. Showing a relation that moves up and down again.

What is Cubic regression?

Rare situation where process relation rises, falls and rises again.

What are the main differences between Regression and DOE?

Regression techniques are generally used to analyse historical data that is taken from the the process in its "normal mode"

Designed experiments are used to create and analyse real time data that is taken from the process in "experimental mode"

In DOE, what does Full design mean?

It means that every possible combination of the input factors (at their 2 levels) is used in the experiment.

The question on the page originate from the summary of the following study material:

  • A unique study and practice tool
  • Never study anything twice again
  • Get the grades you hope for
  • 100% sure, 100% understanding
Remember faster, study better. Scientifically proven.
Trustpilot Logo