TQM LECTURE 5: STATISTICAL METHODS IN QUALITY MANAGEMENT

EXAMPLE 6.1
In testing a new personal computer after assembly, a company discovered that among a
sample of 100 units, 3 failed to boot up properly because of a defect in the mother-board,
STATISTICS 4 units had a hard drive failure, and 2 units experienced both failures. Let A be the event
 The science concerned with “the collection, organization, analysis, interpretation, and “failure to boot” and B be the event “hard drive failure”. Then P (A)=3/100 and P
presentation of data.” (B)=4/100. However, these events are not mutually exclusive because both A and B
 The use of statistical methods in quality dates back to 1903 at the Bell Telephone occurred together; specifically, P (A and B)= 2/100. Therefore, the probability that one or
system. In the 1920s, Bell Labs thought that statistical tools would have applications in the other failure occurred is P (A or B)= P(A) + P(B) – P(A and B) = 3/100 + 4/100 – 2/100
the factory, and began to experiment with statistical sampling, eventually leading to = 5/100.
the development of control charts.

BASIC PROBABILITY CONCEPTS PROBABILITY IN A COIN

 An experiment is a process that results in some outcome. In a coin, we have 2 sides (tails and heads)
 The outcome of an experiment is a result that we observe Rule 1 Probability of an Outcome compose of an event
o E.g.: the number of defective parts in the sample or the length of time until the Now the probability of getting a tail in a toss is (1 means-in a coin there is only 1 tails, 2
bulb fails. means the number of possible outcome (tails and heads)
 The collection of all possible outcomes of an experiment is called the sample space. 1_ (# of tails in a coin)
 Probability is the likelihood that an outcome occurs. 2 (# of sides of a coin or this maybe the total units to be analyzed)
Now, in tossing of a coin are the Event, repeating the toss is the number of events.
The probability of getting the tails in a toss will be 1 divided by 2 is equal to 0.5 or 50%.
PROBABILITY PROPERTIES
 Probabilities are presented mathematically as fractions (1⁄2, 1⁄4, 3⁄4) or as decimals
(0.25, 0.50, 0.75, 0.90) between zero and 1. CONDITIONAL PROBABILITY
 Assigning zero (0) means that something never happens; and a probability of 1  Conditional probability is the probability of occurrence of one event A, given that
indicates that something will definite happen. another event B is known to be true or have already occurred.

 Multiplication rule of probability:

EVENTS
 An event is a collection of one or more outcomes from a sample space
o E.g.: finding 2 or fewer defectives in the sample of 10, or having a bulb burn for
more than 1000 hours.
 If A is any event, the complement of A, denoted as Ac consists of all outcomes in the EXAMPLE 6.2
sample space not in A. Diagnostic test on products or equipment are often unreliable. For example, if a test
 Two events are mutually exclusive if they have no outcomes in common. indicates a failure, it may be wrong some fraction of the time; similarly, a test that results
in a pass may also be wrong. Suppose that if a product that is defective, a diagnostic test
indicates that it is defective only 94 percent of the time, and if the product is good, the test
CALCULATING PROBABILITIES incorrectly states that it is defective 2 percent of the time. Assume that the true
 Rule 1: The probability of any event is the sum of the probabilities of the outcomes percentage of product failures is 1 percent. This situation can be illustrated by a tree
that compose that event. diagram as shown in Figure 6.1. The probability associated with any branch is conditional
 Rule 2: The probability of the complement of any event A is P(Ac) = 1 – P(A). on what has happened before. Thus, along the top right branch, 0.94 represents the
 Rule 3: If events A and B are mutually exclusive, then P(A or B) = P(A) + P(B) probability that the test indicates that the product is defective give n that the product
 Rule 4: If two events A and B are not mutually exclusive, then P(A or B) = P(A) + P(B) actually is defective.
– P(A and B)
Using the multiplication rule, the probability that the product is actually defective and the  mode (most frequent observation
test indicates that is defective can be found by multiplying the probabilities along the
 Exponential - is often concerned with the amount of time until some specific event
branches of the tree.
occurs.
Thus, P (test indicates defective and product is defective)
= P (test indicates defective | product is defective) P (product is defective)
SAMPLING METHODS
= (0.94) (0.01) = 0.0094
1. Simple Random Sampling: Every item in the population has an equal probability of
being selected.
Similarly,
2. Stratified Sampling: The population is partitioned into groups, or strata, and a
P (test indicates defective and product is not defective)
sample is selected from each group.
= P (test indicates defective | product is not defective) P (product is not defective)
3. Systematic Sampling: Every nth (4th, 5th, etc.) item is selected.
= (0.02) (0.99) = 0.0198
4. Cluster Sampling: A population is partitioned into groups (clusters) and a sample of
clusters is selected. Either all elements in the chosen clusters are included in the
sample or a random sample is taken from each of them.
INDEPENDENT EVENTS
5. Judgment Sampling: Expert opinion is used to determine the sample.
 Two events A and B are independent if P (A | B) = P(A).
 Example 6.3: Multiplication Rule for Independent Events

Suppose a process consists of two sequential steps, with the probability of a non-
defective part produced in the first step (event A) being 0.95 and the probability of a
non-defective part produced in the second step (event B) being 0.98. Clearly the two
events are independent, so the probability of producing a non-defective part in the
process in P (A and B)= P (A) P (B) = (0.95)(0.98) = 0.931. This means that if we start
with 1000 parts, only 931 will be non-defective at the end of the process. In quality
control terminology, this is often called the “rolled throughput yield”.

PROBABILITY DISTRIBUTIONS
 A random variable, X, is a numerical description of the outcome of an experiment.
Formally, a random variable is a function that assigns a numerical value to every
possible outcome in a sample space.
 A probability distribution, f(x), is a characterization of the possible values that a
random variable may assume along with the probability of assuming these values.
 The cumulative distribution function, F(x), specifies the probability that the random
variable X will assume a value less than or equal to a specified value, x, denoted as P(X
≤ x).

IMPORTANT PROBABILITY DISTRIBUTIONS

1. Discrete
 Binomial- probability distribution is one in which there is only a probability of two
outcomes.
 Poisson - is a discrete distribution that counts the frequency of occurrences as
integers, whose list {0,1, 2, ...} can be infinite.

2. Continuous
 Normal - also known as the Gaussian distribution
Properties of Normal Distribution
 mean (average)
 median (midpoint)