You are on page 1of 4 P E  F
(
)
“and” =  (Multiply)
“or” =  (Sum)
P ( E / F ) 
P F
(
)
Are the events mutually exclusive?
P(A) & P(B) = 0
P(A) or P(B) = P(A) + P(B)
with P(F)<>0
Are the events Independent?
P(A) & P(B) = P(A) x P(B)
Example: does P(A/B) = P(A)
P(A or B or C) =
P(A) + P(B) + P(C)
-P(A X B) -P(A X C) -P(B X C)
+ P(A X B X C)
P(A) or P(B) = P(A) + P(B) – [P(A) x P(B)]
Are the events Conditional?
Sampling WITH Replacement
Sampling WITHOUT Replacement
(Is one event dependent on the other)
1) BINOMIAL
HYPERGEOMETRIC
P(A) & P(B) = P(B/A) x P(A)
2) When n is large & p is small use
P(A) & P(B) = P(A/B) x P(B)
Poisson Formula or Tables.

n! = n x (n-1) x (n-2) x
x 2
 
nD  
D   N  n 
2
n !

n !
HYPERGEOMETRIC
2
1
n
POISSON
P 
n
C 
r
r
n r
N 
N    N  1 
e
x
(
n
 r
)!
r n  r
!(
)!
D
N D
p x
(
) 
x
0,1,
   
x !
 
N
x
n
 x
p x 
(
)
x  0,1,2,
min(n, D)
= Population Size
= Size of Sample
BINOMIAL DISTRIBUTION
n
N
 
nD
 
D
= Class of Interest
n
p x
(
)    
x
= # of samples fall into
p
x
(1
p
)
n  x
x 0,1,
N
class of interest
  np  p
2 (1
 
n
x
)
  np
,n
 
Mod–Med-X bar
Xbar-Med-Mod H
n
 0.1
N
B
p  0.1
p  0.9
(The smaller
p
and larger
np   10
n
the better)
0.1
p
0.9
P
 15
(thelarger the better)
Let p’=1-p,
the smaller p’
and larger n
the better
N
 EXPONENTIAL DISTRIBUTION GAMMA DISTRIBUTION r ( f x )   e  x  0 2 1 1   x    2    1 e  x   ( f x )  (  x ) r  x  0 2 r  1   1   np    ( r )   2  P x a    a  2  np      a  2 (  )  np (1  p )       np (1  p )      WEIBULL DISTRIBUTION   1     P a  x  b      b  1 2   np    a 1   2  np   f x  ( )     x      exp         x       x  0      np (1  p )      np (1  p )       2   2        1 2         1 1   2        1    1                   P u p ˆ v            v p     u p    p (1  p n )/    p (1  p n )/   pˆ x / n
 Sample Size Variable Data: n    Zs    E  2 E      Z s   n    Z s n
  pq   Z    E  2 Sample Size Discrete Data Binomial: n Sample Size Discrete Data Poisson:  p   Z    E  2 n P x a
1
P x a
P x
 

a
P x
a
Px

a Px
a 
 Variance of the o ulation 2  ( x i  )  2 Variance    N
 Variance of a set of data   x  2 2 i  x Variance    Sigma Square  n
 Estimate Variance of a infinite population Variance  s 2  (  ') 2   ( x i  x ) 2 n  1
 Estimate Variance of a finite population 2 2   ( x i  x ) 2 ( N  1) Variance  s  (  ')  n  1 N

Standard Deviation for a Distribution of averages, Ns= # samples,

n= samples size (
x  x
)
2
i
x
N
s

Standard Error

or

s n ( x  x )
Z 
)
 X
s
2
(
x
i 
x
i
Z 
t
 
F 
1
s
s
s
s
2
n
2
n
 N
  
2
(
2
)
2N 2
 
 
 2)/ 2
e
2
2  (
 2   !
2
2
 
2

2 (

n

1)

s

2

f (

2 )

where

2

 95 % Confidence Interval with known Sigma Confidence Interval with UN-known Sigma X 1.96 n   1.96 X n X  t  2 s n  X  t  2 s n
 Null Hypothesis TRUE FALSE Decision Made: Null Hypothesis (Ho) true Null hypothesis (Ho) false Do not reject Ho No error (1- ) Type II error () Reject Ho Type error () No error (1-) Z
 Z 
H
VARIANCE KNOWN
VARIANCE UNKNOWN
0
/ 2
H
0
0
0
0
Reject Ho if
H
: : 
   
x 
H
: : 
   
Reject Ho if
1
0
Z 
 0
1
0
0
H 1 : :
x

0
if Z 0  Z 
n
1
0
0
0
,
n
1
t 
0
Ho: 0
If If
H
 
 
, , reject reject H H
if if
t t
  t
t
0
S
n
1
1
0
0
0
,
n
1
Reject Ho
0 0
 
P
typeI error
P
reject
H
H
is true
0
0
Reject Ho if Z 0  Z 
Ho:
 
P
typeIIerror
P
fail toreject
H
H
is false
0
0 n  S
1
2
2
The null hypothesis is rejected if:
0
2
H
2
2
0
2
2
or if
2
2
0
0
/2,
n
1
0
1

/2,
n
1
H
0 : :  
2
2
corresponding to the upper /2 and lower1- ( /2) percentage points
of theChi -square distribution with n -1df.
1
0
H 1 : :
2
2
2
 
2
H
 
0
0
 
1
 n
,
1
2
2
The null hypothesis is rejected if:
2
2
1
0
0
 n
,
1
The null hypothesis is rejected if:
  
The Probability of Type II error
 
x
Under H 0 Under H 1
  0
1
0
Z 
n 
n 
/
n
0
0
 
 
Z
   
Z
0
/2
 
 
/2
 Z  / 2
 n /
 
Z  / 2

Clasificacion

Nivel Confianza Z

 Maradona 90% 1.645 Michel Jordan 95% 1.96 Yao Li 99% 2.58 Sigma Percentage In PPM Defective 1 68.26 317,300.0 2 95.44 45,500.0 3 99.73 2,700.0 4 99.9937 63.0 5 99.999943 0.57 6 99.999660 0.002  21 
Z
for
a two- tailed test:
H
0 :
 
H
:
0
0
1
0
P   1

(
Z
)
for
an upper- tailed test:
H
0 :
 
H
:
0
0
1
0
( Z
)
for
a lower - tailed test:
H
 
H
:
0 :
0
0
1
0
 Confidence Interval on the Variance Normal Dist. ( 1) n   2  /2, n S  1 2   2  (  n  2 1   1) /2, S n  2 1 One sided Confidence Interval on the Variance  2  ( n  1) 2 1    , n S  1 2 ( n  2   , 1) S n  1 2   2 INFERENCE ON A POPULATION PROPORTION H H : 1 : p p   p p 0 0 0 Reject if: Z  0  x (  0.5)  np 0  np (1  p )  0 0  if x  np 0  ( x  0.5)  np 0 Z  Z if x  np 0 0  /2  np (1  p )  0 0 Confidence Interval on a Population Proportion ˆ  Z p  /2 p ˆ(1 ˆ)  p n  p   p ˆ Z  /2 p ˆ(1 ˆ)  p n
 Inference for a Difference in Means, Variance Known H 0 : 1 2 0 Alternative Hypothesis Rejection Criterion Z  0 H 1 :   1   2 0 Z Z orZ 0   /2 0  Z  /2 x  x  1 2 0  2  2 1  2 n n 1 2 H H 1 1 :   1   2 :   1   2 0 0 Z 0  Z  Z  Z 0  Confidence Interval on a Difference in Means, Variances Known x 1  x 2  Z  /2  2  2 1  2 n n 1 2   1   2  x 1  x 2  Z  /2  2  2 1  2 n n 1 2

Case 1:

2

2

2

t 0 1
1
S
p n
n
1
2
H
: 
1
1
H
: 
1
1
H
: 
1
1

Inference for a Difference in Means, Variance Unknown

0

0

0

t

0

n 1 n

2

t

0



t

/ 2,

n

1

t

0

t

,

n

1

n

2

2

t

0



t

,

n

1

n

2

2

n

2

2

or

2 2

S p

2

2

2







H 0 : 1 2 0 n 1 S

2

2

2

Alternative Hypothesis

Rejection Criterion

Case 2:

1

2

2

2     x
 x 
1
2
0
S
2
S
2
1
2
n
n
1
2

*

t 0      2
 S
2
1 2 S 
2
n n
 
1  2
2
(
S
2
/
n
)
2
(
S
2
/
n
)
2
1
1
2
2
n  1
n  1
1
2

1

2 x 1

x 

2

0

n 1

1

n n 2

1

2  2

S

1

n

1

2

n

2   
Confidence Interval on a Difference in Means, Variances Unknown
2
2
2
1
2
1
1
1
1
x
x
t
S
x
x
t
S
1
 
2
1
2
/ 2,
n
n
2
p
1
2
/ 2,
n
n
2
p
1
2
n
n
1
2
n
n
2
2
1
2
1
2
1
2
S
2
S
2
S
2
S
2
x
 x  t
1
2
x
x
t
1
2
1
2
/2,
2
1
n
n
1
2
/2,
n 1 n
1
2
2
 Inference for paired data d t 0  S d / n H  0 : d  0 Reject if: t 0  t  /2, n  1 Average difference between data. Sd is the STD of differences
 Inference on the Variances of Two Normal Alternative Hypothesis / Test Statistic / Rejection Criterion 2 H :  1 1 2 2 F  0 S S 2 1 2 F 0 F 0  F  or  F 1 /2, 1, n  n  1 2 2 1 2  (  / 2), n  1 1, n  2 S 2 2 1 H 1 :  2 F 0  2 F 0  F  , n 2  1, n 1  1 2 S 2 1 H : 2  2 S 2 F 0  F 1 F 1  , n  1 1, n 2  1 1 2 0  S 2 2 Confidence Interval on the Ratio of the Variances of Two Normal Distributions S S 2 1 2 F 1   /2, n 2  1, n 1  1    2 1 2  S S 2 1 2 F  /2, n 2  1, n 1  1  2 1 2  S S 2 1 2 F  , n 2  1, n 1  1 S S 2 1 2 F 1   , 1, n  n  2 1 1   2 1 2 2 2 2  2 2 2  2
 Inference on Two Population Proportions p ˆ 1  p ˆ 2 GOODNESS OF FIT Z  0 H H 0 : The shape of the distribution is the proposed 1  p ˆ(1  ˆ)   p   n 1   n   1 2 0 : p  p 1 2  k ( O E i  i ) Reject Ho if: 2  2 2  0 E i   Alternate Hypothesis Rejection criteria i  1 0   k p   , 1 H 1 1 : p  p Z  Z or Z  Z Where k = class intervals, p = number of parameters 2 2 H 1 p  H 1 Z Z  : p : p 1 1 p 2 0 0 0   Z   Z /2  0  /2 Confidence Interval on the Difference in Two Population Proportions ˆ  p ˆ p 1 2  Z  /2 p ˆ (1  ˆ p ) p ˆ (1  ˆ p ) 1 1  2 2 n n 1 2  p 1  p 2  p ˆ 1  p ˆ 2  Z  /2 p ˆ (1  ˆ p ) p ˆ (1  ˆ p ) 1 1  2 2 n n 1 2             If n different in between cells, then the Error degree of freedom is:

a*b*(n-1)*(n-1)*(n-1)… 