What is the Red X?
The input variable that has the largest effect on the Green Y- Example: In one of my projects, the red x was the diameter of the IP harness where it was wrapped in electrical tape had the largest effect on the green Y
What is the Green Y?
- The Output variable that matters the most to the customer- Example: In one of my projects the green Y was the side badge alignment because that is the variable that matters the most to the customer
What is Y to X thinking?
- Convergent, always reduces the area to look at. Uses distinctions to converge on the red X- Effect to cause, rather than cause to effect- Example: In one of my projects I questioned to the void and used a contrast to find the red x in the life cycle
What is X to Y thinking?
- Divergent, cause to effect. Does not guarantee that you will solve the problem. Does not guarantee that you will think of the correct solution- Example: Guess and check, fishbone diagrams
What are BOBs and WOWs?
Draw on white board- BOBs and WOWs are always on the opposite ends of the distribution- The ends are 2.5% tails
Why do we use BOBs and WOWs?
- It's for Contrast because it really helps for problem solving- Example: In one of my projects I needed to see the contrast between the smallest and biggest diameter of a IP wiring harness in order to see if the diameter had an effect on the green y
Which tail is BOB which tail is WOW?
It depends on the problem, BOBs and WOWs could be on either end
Draw the specifications, now where do BOBs and WOWs fall?
Draw on the board but don't move BOBs and WOWs- Specifications do not matter at all, they are still both at either ends
What do BOBs and WOWs mean?
- They just mean completely opposite ends of the distribution. THEY ARE BOTH BAD- Example: If a camera distance specification is 2 - 3 mm, 0 mm and 5 mm is still terrible. But you want the largest contrast
How do you turn a BOB into a WOW?
- You can't EVER turn a BOB into a WOW. You would be guessing what the problem is and then only confirming something that you know.- YOU CAN TURN WOWs into BOBS tho
How do you do an isoplot?
Go through steps
How do you know that your isoplot has passed?
If your discrimination ratio is greater than or equal to 6 and if your graph looks like a GOOD SAUSAGE, your isoplot has passed
How do you know if your isoplot has failed
If your discrimination ratios is less than 6 and you have a meatball or fat sausage then your isoplot has failed
What do you do if you fail your isoplot?
- If you fault your isoplot you either have to increase your delta P or decrease delta M- To increase delta P you need to find better BOBs and worse WOWs- To decrease delta M you can either improve measurement method, find a better Green Y, increase the resolution (YOU ARE BASICALLY LOOKING TO MAKE YOUR MEASUREMENT MORE REPEATABLE)
In 2 words, what is your delta M, what does it represent?
Measurement variation
In 2 words what is the delta P
Part variation
Where else can your delta P live?
You can get your delta P from both axes, but you use the smaller one
What could cause the delta P to be different between the axes?
- Measurement bias between trials- Comes down to the difference between the measurements you took
What is it telling you when your isoplot passes?
- Your measurement system is valid and can reliably tell the difference between BOBs and WOWs- Example: My isoplot passed for check the diameter of the IP harness, so I knew my system was valid
What is the purpose of a confirmation test?
To prove the relationship between the green y and the red x
What is the risk?
Risk is the chance that the results from your test happen because of coincidence
What is confidence?
Confidence is the opposite of risk, the confidence you have that your test is correct
Why are confirmation tests powerful?
Confirmation tests are so powerful because instead of random samples they use BOBs and WOWs to get contrast between
In a 6 pack can you measure the Y first instead of the X?
No, it's because a 6 pack is meant to be predictive. It is to see if the X can predict the Y. It establishes causation instead of correlation
Can you use the same data from a test can you use the same data in a different test?
No you have to redo the test
What are the 2 different kinds of 6 packs
1. Better vs current (is more for a process issue)2. BOB vs WOW
In the confirmation tab, what goes in the root cause box?
The red x
In the confirmation tab, what is the systemic root cause?
How the company messed up on a higher level
In the confirmation tab, what goes in the lessons learned?
What we need to put in place as a company to prevent the systemic root cause from happening