Download as docx, pdf, or txt
Download as docx, pdf, or txt
You are on page 1of 1

- Tw types of run tests: Up/down test, Above/below tests

- Each point except the first one has been labeled with either a u or a d.
- U represents that the line coming from the last point sloped up
- D represents the line coming from the last point sloped down
- The first point does not have a u or a d because there’s no prior point
- Putting those letters together we have U U D U D U D U U D
- We now count the number of line segments
- 1 to 3 is UP; 3 to 4 is DOWN; 4 to 5 is UP; 5 to 6 is DOWN; 6 to 7 is UP; 7 to 8 is DOWN; 8 to10 is
UP; 10 to 11 is DOWN
- So there are 8 up down line segments for 11 samples (8 runs), that will be the basis of our up
down test

 Above below test is based on the median


 Here each series that is completely above or below the median has been circled
 Point 1 is BELOW so one stands alone; point 2 and 3 are all ABOVE the median so they’re all in
one circle; point 4 is BELOW so two stands alone; point 5 is ABOVE so three stands alone; point
6,7, and 8 are BELOW; point 9 and 10 are ABOVE; point 11 is BELOW so four stands alone
 We have 7 circles on the chart (7 runs)

 Once we know how many up down and above below values we have
 We need to test to see if the number we have is something reasonable for random chance or if
it is not something we would expect from random chance
 To perform this test: We need to know 3 things
 1: the number we would expect from random chance, this is the Expected Value
 2: how much variability we would expect in the expected value, this is the standard deviation
 3: how to handle this probabilistically
 both distributions are normally distributed so we will convert the actual observations to a z
value, if it is between negative and plus 2 we will conclude that there is no pattern
 outside this range we will conclude a pattern exists

- the expected value of the above below test


- N is the number points on the chart and not the size of the sample used to compute the points
- That is why one is a capital n and not a lower case n
- The expected value of the up down test
- The standard deviation of the above below test
- The numbers used in these formulas are constants and do not change
- We compute the z value for either test as the actual number of runs counted earlier minus the
number of expected runs computed earlier
- Quantity divided by the standard deviation of the number of runs computed earlier
- Within plus or minus 2 so we conclude the process is in control, this shows that a process can
appear to have a pattern but the statistical test not find one

You might also like