Six Sigma Two

Start Studying!

Terms

undefined, object

copy deck

In the measure phase of DMAIC, what are the items needed?: A valid measurement system
In the measure phase, we are going to establish a defect rate, but black belts typically see the defect rate go down: false
What is one of the first important milestones that indicates that a black belt is on track?: The process map is complete
How many data points do you need to have a short-term capability?: Between 30 and 50 data points
Process mapping: is an ongoing living document used throughout the DMAIC process
What are the two key questions to ask for each step?: Why do we know this step and how do we know this step is good?
An XY diagram is a statistical tool: False
A YX diagram helps point the black belt into a direction with factual evidence: False
A FMEA describes the following...: What you want to know about a type of defect
AN FMEA is complete during the measure phase: False
In an FMEA, what is the RPN if OCC is 5, DET is 4 and SEV is 9?: OCC * SEV * DET
180
Measurement system analysis MSA is used: To validate the data used for analysis
MSA is a tool that can be omitted from the DMAIC process: False
Cp is a capability index with the units measured in: No units
If Cp is 1.0 what is the sigma value: 3
Can Cp be greater than Cpk: yes
What is the Cp and the Cpk index number when you have a six-sigma capability?: Cp=2.0 and Cpk=1.5
What is the purpose for gauge R&R?: 1. statistical analysis to evaluate measure error.
2. To understand repeatability and reproducibility of your MSA.
3. Validate what is/isn't a defect.
4. Look for variation within and between operators.
If your data is nonnormal, you are stuck in the measure phase: false
What is the layperson's description of a hypothesis test?: A tool to compare stuff
What are the reasons for nonnormality?: 1. Bimodal conditions exist
2. Different normal distributions are within the data set.
If you have a nonnormal data set does transforming the data fix the nonnormal causes of the problem?: no
What would best describe a bimodal distribution?: A multivari chart and and an X factor that has two different Y-output distributions
How does comparing factors help solve the problem: breaks down the problem to the vital Xs. Contrasts the trivial few versus the the vital few. Helps answer the hypothesis question. Deals with data facts that can be proven, and focuses the team on data, not opinion.
Tool wear can cause nonnormal distributions: true
What plot describes the many distributions in one graph in quartiles: box plot
Is it okay to remove outliers in a data set that causes an increase in standard deviation?: no - yes, if you know the cause to stop it.
What is the best way to show multimode distributions: Interval plot, dot plot, one way ANOVA,
Lowess analysis fits a robust line through the data to display a relationship between X and Y: true
In a multivari analysis, the X levels are randomly selected levels during the study: true
Different operators producing the same Y cannot cause nonsymmetrical distributions: False
In simple terms, what is meant by a p-value of less than 0.05?: You're 95% confident that there is a statistical difference.
The 95% confidence interval increases as the standard deviation increases: true
You do not need a capable measurement system for multivar analysis: false
Shift-to-shift variation can be measured on one shift: false
A hypothesis test can show the interactions of the factors: false
Sample size has no effect on the width of a distribution: false
If an X has been identified as statistically significant, do you disregard it owing to an expert telling you to ignore it?: No - ask what data the expert has to show it is worth ignoring.
What does hypothesis testing fundamentally change: turns problem into a fact based process - destroys the emotions - departure from the I think/feel culture, and data is now used to drive the decision
If you changed an X that was proven to be statistically significant and the Y was given to you with 3 months prior to the change and 1 month after, could you show a before and after hypothesis to validate the change: No
Using an Anderson-Darling normality test, normal data have a p-value of less than 0.5: true
How many runs does a 2/3 full factorial experiment consist of?: 8
In an experiment, inputs are allowed to vary randomly throughout the specification range.: false
One-factor-at-a-time experiments generater more powerful data than a full factorial experiment: false
What is an experimental factor: The input vairables for the experiment
What does orthogonal mean: A property that ensure that all experimental factors are independent of each other, no correlation exists between Xs.
Standard order is the same as run order: false
Why use factorial plots: 1. see plots of the main effects
2. alls you to see the interaciton plots
3. allows you to see the cube plots
4. shows how to set each factor to max or min the response.
What tools can be used to determine if factors have interaction: 1. balanced ANOVA
2. Standardized effects
3. Interaction plots
4. Fractional factorial fits.
What does it mean when no p-values are presented in the ANOVA output?: 1. All factors were statistically significant.
2.Factors weren't different.
3.Only one repetition was run at each treatment combination.
4. Had no center points
Why do we replicate our experimental runs?: To obtain a better estimate of the error and look at interactions
To use a center point in your experimental design, at least one factor must be able to be set at itâ€™s midpoint coded value = 0.: false
Why use center points in your experimental design?: to detect curvature
If a center point is significant, its p-value in the ANOVA table will be greater than 0.05.: false
Fractional factorial designs require more runs than full factorial designs given the same number of factors.: false
What is the main reason for using a fractional factorial design?: Allows you to test and screen a large number of factors in fewer runs
Gives you good estimates of low order interactions
Gives you relative significance of the factors
Given three factors, A, B, and C, the highest-order interaction would be ABC.: true
In a four-factor 1/2 fractionated design, the AB interaction is confounded with the CD interaction.: true
What does it mean when A is confounded with BC?: BC is contributing to the result.
In a resolution IV design, two-factor interactions are aliased with three-factor interactions.: true
In a resolution III design, single factors are not aliased with any other factors.: true
The identity expression I + ABCD is used to generate the confounding pattern.: true
Why are resolution V designs preferred over resolution III and resolution IV designs?: No main effect is confounded with any other main effect or second-order interactions.
No second-order interactions are confounded with any other second-order interaction, and second-order interactions are confounded with third-order interactions.
Allows for the differentiation of the effects down to the second order, assuming that the effects of third-=order interactions are negligible.
Why should you do a hypothesis test before running a DOE?: To identify as many the vital few factors prior to DOE
What is the mission of the Improve phase?: Find the relationships between X and Y
Validate hypothesis tests
Which inputs to control in the next phase
Run a pilot to validate experiment
Can you calculate epsilon-square percent contribution for the DOE given that the degree of freedom for each factor is different in the ANOVA?: No
The WECO rules are based on probability. We know that for a normal distribution, the probability of encountering a point outside +/-2.5 is 0.3 percent. This is a rare event. Therefore, if we observe a point outside the control limits, we conclude that th: true
Outliers usually have a significant effect on an equation derived with regression analysis: true
Using Y=f(x), do we set tolerance limits for Y?: no
What is the residual?: It indicates how well the equation fits. And is a calculation of the expected value minus the observed value
What are control charts?: Charts used to routinely monitor quality.
What does a P-chart track: na
A DOE is always needed to solve process issues: true
The purpose of performing a designed experiment is to determine what?: all
With the DOE, the easiest way to test for curvature is to include center points: true
The most common response surface design is called the central composite design: true
Control plans provide a written description of the actions that are required at each phase of the process to ensure that all process inputs and outputs will be in a state of control.: true
Control plans are only generated at the start of the life cycle of a product.: false
SPC is a statistically based graphing technique that compares current process data to a set of stable control limits established from normal process variation: true
Control limits and specification limits are the same thing: false
Regular residuals are the actual values of the residuals calculated by subtracting the expected value from teh observed value.: false
What is the best description of a data transform.: It transforms data into a more approximate normal distribution
Once the special causes of variation in a process have been discovered and eliminated, the long-term goal of anyone managing a process will be to reduce common cause variation by improving the process or system itself.: true
Mistakeproofing seeks to gain permanence by eliminating or rigidly controlling human intervention in a process: true
Six Sigma product and process design made to the product/process that eliminates the error condition are called what: mistake elimination
Systems that monitor the process and automatically adjust critical Xs to correct settings are called what: Full automation
In order for mistakeproofing systems to operate effectively, the following rules must NOT be observed: System overrides must not be used except in an emergency
The EWMA is a variable data control chart: True
An â€œout of controlâ€ situation in a production process may be signaled by a sample f output, which generates a data point outside the control limits on either the Xbarâ€“R range chart.: true
When an â€œout of controlâ€ situation is signaled on a control chart, the person using the chart will know why the data are giving the signal.: false

Start Studying!

Deck Info

Number of cards 90