Page 248 - Six Sigma Demystified

228 Six SigMa DemystifieD

Distributions

When process and population data are fit by assumed distributions, broad pre-
dictions can be made with minimal data. Popular statistical distributions in-
clude the binomial and Poisson distributions for discrete (count) data and the
normal, exponential, and robust Johnson and Pearson distributions for continu-
ous (measurement) data.

When to Use

Measure to Control Stages

• To estimate properties of an existing or potential process or population,
including its failure rate or sigma level

• To generate random data for process modeling

Binomial Distribution
The binomial distribution is used to estimate the number of units meeting one
of two possible conditions in a process or population. For example, if the pop-
ulation is the total number of orders shipped in July, the condition of interest
might be the number of units shipped on time. Since there is only one other
possible condition (the order is not shipped on time), the binomial distribution
is appropriate for modeling the number of units shipped on time. It may be
applied when the number of samples is fixed and trials are independent with
equal probability of success.

Binomial Distributions

Minitab

Use Calc\Random Data\Binomial to generate random numbers using a fixed
sample size and p value.

Excel

Use Data\Data Analysis\Random Number Generation.

Set Distribution = Binomial using a fixed sample size and p value.