Six Sigma – Day 9

Tracy Owens
Exercise 2 – the 5-second exercise
In your group, choose one person to be the timer
No other timing devices are permitted
Each person will take these steps:
1. Nod to the timer and say “Start”
2. At that instant, the timer will start the timing device
3. Count silently until you reach 5 seconds
4. Nod to the timer and stay “Stop”
5. The timer will stop the timing device and type the time
into the worksheet, rounded to tenths of a second
i.e., 4.7 or 5.2, not 4.71221
6. Timers, do not reveal the times to the team
7. Each person will continue in sequence until 12 total
numbers have been recorded
 Data Types

3
What are the two types of data?

Continuous data – measured on a continuum

Discrete data – not measured on a continuous scale
- PASS/FAIL binomial
- Counted whole numbers, not subdivided

3
 Data Types
TYPE OF DATA HOW YOU DESCRIBE THE DATA HOW YOU EVALUATE CAPABILITY STATISTICAL SCORING HOW YOU MONITOR
center and spread of the data ability to meet the customer's specifications comparing and predicting performance statistical process control chart

CENTER - Mean, x or µ Cp, Cpk, Cpm Short Term Z-score x-mr chart for individuals
CONTINUOUS SPREAD - Standard Deviation, σ Pp, Ppk Long term Z = (x-µ)/σ x-R chart for small subgroups
or Variance, σ2 Sigma value converted from Cpk (Cpk*3) where: x = individual value x-s chart for large subgroups
Note: lower case s is often used for the µ = population mean use 6-10 for the cutoff
standard deviation of a sample σ = population standard deviation
CENTER - Median, ~x Yield (% first-time quality) Binomial Distribution p chart for variable
DISCRETE SPREAD - Margin for Error, d or D Sigma value converted from Yield P(r) = [n!/((n-r)!*r!)] * (p r*qn-r) subgroup sizes
PASS/FAIL Degree of Precision or Margin for Error: where: n = total number of trials np chart for constant
The maximum expected difference between r = number of successful trials subgroup sizes
our estimate and the actual value p = probability of success
q = probability of failure = 1 - p
CENTER - Median, ~x Defects per Unit (DPU) Poisson Distribution u chart for variable
DISCRETE SPREAD - Range, R or 5/95 Span Defects per Million Opportunities (DPMO)
-m
P(r) = (e *m )/r!
r
subgroup sizes
Range: max value minus min value -np r
COUNTS Sigma value converted from DPMO or P(r) = (e *np )/r! . c chart for constant
Span: 5th to 95th percentile only where: r = hypothesized defect count subgroup sizes
or 1st to 99th, or 10th to 90th Note: DPU also equals -ln(Y) µ, l, np = average number of defects

The analysis you perform will depend on

the type of data you have
4
Sampling Methods
• Data at Rest
At the end of a time period, select samples from the
completed tasks or units
Random and Stratified Random sampling
• Data in Motion
While the process is running, select samples during the
flow to catch changes as they happen
Sequential sampling
• Rational Subgrouping
Catches variation within and between groups
The Graphical Analysis Roadmap

Create a Create a Create Create

Histogram and Time-Series Box-Plots, Scatter Plots to
check, if a plot and check to look for see correlation
normal look for trends changes in to one or more
distribution is over-time. group variance process
existent. or mean. parameters

Histogram of Filler1 Time Series Plot of Filler1 Boxplot of Filler2 Scatterplot of Lab pH vs New pH
Normal
235
9.4
25
Mean 219.9 225.0
StDev 2.107 9.3
N 120
230 9.2
20 222.5
9.1
Frequency

Lab pH
9.0
Filler1

15
Filler2

225
220.0
8.9
10
8.8
217.5 220
8.7
5
8.6
215.0 215
0 8.5
216 218 220 222 224 226 1 2 4 6 8 10 12 14 16 18 20 1 2 3 4 5 6 8.6 8.8 9.0 9.2 9.4 9.6
Filler1 Study hour Head New pH

Histogram Time-Series Plot Box-Plot Scatter plot

Histogram and Boxplot
PROJECT METRIC
Project Metric “Y” should be:
• Aligned with the organization’s strategic goals
• Clear description of what should be measured (unit level)
• Cascaded to a project level
• Indication of the source for the measurement data (if
available)

Y = f (x)
Measurement Scales

Borrowed from Statology.org

https://www.statology.org/levels-of-measurement-nominal-ordinal-interval-and-ratio/
Histogram Progression
from Dotplot to Histogram

number of bins = square root of sample size, rounded up

“I keep answering the same question…
…over and over again.”

Born from sales and legal teams sparring over delays

and incomplete information

The Response Router

 Response Router Construction

ALWAYS YES NEGOTIATE ALWAYS NO REASON ALTERNATIVE

Left column: This column is “No” column: If possible, Further, give
items that you for items that options that provide the the requestor
don’t even don’t fit and are not possible reason(s) an alternative
have to ask anything new or available for rejection if you can

Any number of items that can be sorted into ALWAYS

YES or ALWAYS NO will save time
 Response Router Sample
ALWAYS YES NEGOTIATE ALWAYS NO REASON ALTERNATIVES
Requests for Reduced Indemnification The city’s Suggest an
on-site corporate taxes insurance arbitration clause
corporate for XX months prohibits this using Radisonville
wellness clause. city code/laws.
program for
one year
Access to Employee Deviations from The city’s Remind applicant
city’s fitness training by city code for new strategic plan that approval is
facility for all staff; opening construction includes vision fast-tracked if all
company position is “not for new documentary
employees for offered”; construction and requirements are
the first three escalate to city’s is strictly met upon initial
months organizational enforced. application.
development
staff if needed
Donations by Free fitness The city’s Negotiate a
company of facility use for insurance reduced rate for
less than $500 all employees prohibits fee- months 3-12, max
for city events for more than free use of discount is 50%.
three months athletic facilities
beyond 90 days.
Anything new!