Design of Engineering Experiments - Experiments with Random Factors
description
Transcript of Design of Engineering Experiments - Experiments with Random Factors
Chapter 13 Design & Analysis of Experiments 7E 2009 Montgomery
1
Design of Engineering Experiments - Experiments with Random Factors
• Text reference, Chapter 13• Previous chapters have considered fixed factors
– A specific set of factor levels is chosen for the experiment– Inference confined to those levels– Often quantitative factors are fixed (why?)
• When factor levels are chosen at random from a larger population of potential levels, the factor is random– Inference is about the entire population of levels– Industrial applications include measurement system
studies
Chapter 13 Design & Analysis of Experiments 7E 2009 Montgomery
2
Random Effects Models• Example 13.1 – weaving fabric on looms• Response variable is strength• Interest focuses on determining if there is difference in
strength due to the different looms• However, the weave room contains many (100s) looms• Solution – select a (random) sample of the looms,
obtain fabric from each• Consequently, “looms” is a random factor• See data, Table 13.1; looks like standard single-factor
experiment with a = 4 & n = 4
Chapter 13 Design & Analysis of Experiments 7E 2009 Montgomery
3
Chapter 13 Design & Analysis of Experiments 7E 2009 Montgomery
4
Random Effects Models
• The usual single factor ANOVA model is
• Now both the error term and the treatment effects are random variables:
• Variance components:
1,2,...,
1, 2,...,ij i ij
i ay
j n
2 2 is (0, ) and is (0, )ij iNID NID
2 2( )ijV y
Chapter 13 Design & Analysis of Experiments 7E 2009 Montgomery
5
Relevant Hypotheses in the Random Effects (or Components of Variance)
Model• In the fixed effects model we test equality of
treatment means• This is no longer appropriate because the
treatments are randomly selected – the individual ones we happen to have are not of
specific interest
– we are interested in the population of treatments
• The appropriate hypotheses are2
0
21
: 0
: 0
H
H
Chapter 13 Design & Analysis of Experiments 7E 2009 Montgomery
6
Testing Hypotheses - Random Effects Model
• The standard ANOVA partition of the total sum of squares still works; leads to usual ANOVA display
• Form of the hypothesis test depends on the expected mean squares
• Therefore, the appropriate test statistic is
2 2 2( ) and ( )E TreatmentsE MS E MS n
0 /Treatments EF MS MS
Chapter 13 Design & Analysis of Experiments 7E 2009 Montgomery
7
Estimating the Variance Components• Use the ANOVA method; equate expected mean squares to
their observed values:
• Potential problems with these estimators– Negative estimates (woops!)– They are moment estimators & don’t have the best statistical
properties
2 2 2
2
2
ˆ ˆ ˆ and
ˆ
ˆ
E Treatments
Treatments E
E
MS n MS
MS MS
n
MS
Chapter 13 Design & Analysis of Experiments 7E 2009 Montgomery
8
Chapter 13 Design & Analysis of Experiments 7E 2009 Montgomery
9
Either JMP or Minitab can be used to analyze the random effects model
Chapter 13 Design & Analysis of Experiments 7E 2009 Montgomery
10
Minitab Solution (Balanced ANOVA)Factor Type Levels Values
Loom random 4 1 2 3 4
Analysis of Variance for y
Source DF SS MS F P
Loom 3 89.188 29.729 15.68 0.000
Error 12 22.750 1.896
Total 15 111.938
Source Variance Error Expected Mean Square for Each Term
component term (using unrestricted model)
1 Loom 6.958 2 (2) + 4(1)
2 Error 1.896 (2)
Chapter 13 Design & Analysis of Experiments 7E 2009 Montgomery
11
Confidence Intervals on the Variance Components
• Easy to find a 100(1-)% CI on
• Other confidence interval results are given in the book
• Sometimes the procedures are not simple
2 :
22 2
/ 2, 1 ( / 2),
( ) ( )E E
N a N a
N a MS N a MS
Chapter 13 Design & Analysis of Experiments 7E 2009 Montgomery
12
Extension to Factorial Treatment Structure
• Two factors, factorial experiment, both factors random (Section 13.2, pg. 511)
• The model parameters are NID random variables• Random effects model
2 2 2 2
2 2 2 2
1,2,...,
( ) 1,2,...,
1, 2,...,
( ) , ( ) , [( ) ] , ( )
( )
ijk i j ij ijk
i j ij ijk
ijk
i a
y j b
k n
V V V V
V y
Chapter 13 Design & Analysis of Experiments 7E 2009 Montgomery
13
Testing Hypotheses - Random Effects Model
• Once again, the standard ANOVA partition is appropriate
• Relevant hypotheses:
• Form of the test statistics depend on the expected mean squares:
2 2 20 0 0
2 2 21 1 1
: 0 : 0 : 0
: 0 : 0 : 0
H H H
H H H
2 2 20
2 2 20
2 20
2
( )
( )
( )
( )
AA
AB
BB
AB
ABAB
E
E
MSE MS n bn F
MS
MSE MS n an F
MS
MSE MS n F
MS
E MS
Chapter 13 Design & Analysis of Experiments 7E 2009 Montgomery
14
Estimating the Variance Components – Two Factor Random model
• As before, use the ANOVA method; equate expected mean squares to their observed values:
• Potential problems with these estimators
2
2
2
2
ˆ
ˆ
ˆ
ˆ
A AB
B AB
AB E
E
MS MS
bnMS MS
anMS MS
n
MS
Chapter 13 Design & Analysis of Experiments 7E 2009 Montgomery
15
Example 13.2 A Measurement Systems Capability Study
• Gauge capability (or R&R) is of interest
• The gauge is used by an operator to measure a critical dimension on a part
• Repeatability is a measure of the variability due only to the gauge
• Reproducibility is a measure of the variability due to the operator
• See experimental layout, Table 13.3. This is a two-factor factorial (completely randomized) with both factors (operators, parts) random – a random effects model
Chapter 13 Design & Analysis of Experiments 7E 2009 Montgomery
16
Chapter 13 Design & Analysis of Experiments 7E 2009 Montgomery
17
Example 13.2 Minitab Solution – Using Balanced ANOVA
Chapter 13 Design & Analysis of Experiments 7E 2009 Montgomery
18
Example 13.2 Minitab Solution – Balanced ANOVA
• There is a large effect of parts (not unexpected)
• Small operator effect• No Part – Operator interaction• Negative estimate of the Part – Operator
interaction variance component• Fit a reduced model with the Part –
Operator interaction deleted
Chapter 13 Design & Analysis of Experiments 7E 2009 Montgomery
19
Example 13.2 Minitab Solution – Reduced Model
Chapter 13 Design & Analysis of Experiments 7E 2009 Montgomery
20
Example 13.2 Minitab Solution – Reduced Model
• Estimating gauge capability:
• If interaction had been significant?
2 2 2ˆ ˆ ˆ
0.88 0.01
0.89
gauge
Chapter 13 Design & Analysis of Experiments 7E 2009 Montgomery
21
The Two-Factor Mixed Model• Two factors, factorial experiment, factor A fixed,
factor B random (Section 13.3, pg. 515)
• The model parameters are NID random variables, the interaction effect is normal, but not independent
• This is called the restricted model
2 2 2
1 1
1,2,...,
( ) 1, 2,...,
1, 2,...,
( ) , [( ) ] [( 1) / ] , ( )
0, ( ) 0
ijk i j ij ijk
j ij ijk
a a
i iji i
i a
y j b
k n
V V a a V
and j ijk
Chapter 13 Design & Analysis of Experiments 7E 2009 Montgomery
22
Testing Hypotheses - Mixed Model
• Once again, the standard ANOVA partition is appropriate
• Relevant hypotheses:
• Test statistics depend on the expected mean squares:
2 20 0 0
2 21 1 1
: 0 : 0 : 0
: 0 : 0 : 0
i
i
H H H
H H H
2
2 2 10
2 20
2 20
2
( ) 1
( )
( )
( )
a
ii A
AAB
BB
E
ABAB
E
E
bnMS
E MS n Fa MS
MSE MS an F
MS
MSE MS n F
MS
E MS
Chapter 13 Design & Analysis of Experiments 7E 2009 Montgomery
23
Estimating the Variance Components – Two Factor Mixed model
• Use the ANOVA method; equate expected mean squares to their observed values:
• Estimate the fixed effects (treatment means) as usual
2
2
2
ˆ
ˆ
ˆ
B E
AB E
E
MS MS
anMS MS
n
MS
Chapter 13 Design & Analysis of Experiments 7E 2009 Montgomery
24
Example 13.3 The Measurement Systems Capability
Study Revisited
• Same experimental setting as in example 13.2
• Parts are a random factor, but Operators are fixed
• Assume the restricted form of the mixed model
• Minitab can analyze the mixed model
Chapter 13 Design & Analysis of Experiments 7E 2009 Montgomery
25
Example 13.3Minitab Solution – Balanced ANOVA
Chapter 13 Design & Analysis of Experiments 7E 2009 Montgomery
26
Example 13.3 Minitab Solution – Balanced ANOVA
• There is a large effect of parts (not unexpected)• Small operator effect• No Part – Operator interaction• Negative estimate of the Part – Operator
interaction variance component• Fit a reduced model with the Part – Operator
interaction deleted• This leads to the same solution that we found
previously for the two-factor random model
Chapter 13 Design & Analysis of Experiments 7E 2009 Montgomery
27
The Unrestricted Mixed Model
• Two factors, factorial experiment, factor A fixed, factor B random (pg. 518)
• The random model parameters are now all assumed to be NID
2 2 2
1
1,2,...,
( ) 1,2,...,
1, 2,...,
( ) , [( ) ] , ( )
0
ijk i j ij ijk
j ij ijk
a
ii
i a
y j b
k n
V V V
Chapter 13 Design & Analysis of Experiments 7E 2009 Montgomery
28
Testing Hypotheses – Unrestricted Mixed Model
• The standard ANOVA partition is appropriate
• Relevant hypotheses:
• Expected mean squares determine the test statistics:
2 20 0 0
2 21 1 1
: 0 : 0 : 0
: 0 : 0 : 0
i
i
H H H
H H H
2
2 2 10
2 2 20
2 20
2
( ) 1
( )
( )
( )
a
ii A
AAB
BB
AB
ABAB
E
E
bnMS
E MS n Fa MS
MSE MS n an F
MS
MSE MS n F
MS
E MS
Chapter 13 Design & Analysis of Experiments 7E 2009 Montgomery
29
Estimating the Variance Components – Unrestricted Mixed Model
• Use the ANOVA method; equate expected mean squares to their observed values:
• The only change compared to the restricted mixed model is in the estimate of the random effect variance component
2
2
2
ˆ
ˆ
ˆ
B AB
AB E
E
MS MS
anMS MS
n
MS
Chapter 13 Design & Analysis of Experiments 7E 2009 Montgomery
30
Example 13.4 Minitab Solution – Unrestricted Model
Chapter 13 Design & Analysis of Experiments 7E 2009 Montgomery
31
Finding Expected Mean Squares
• Obviously important in determining the form of the test statistic
• In fixed models, it’s easy:
• Can always use the “brute force” approach – just apply the expectation operator
• Straightforward but tedious• Rules on page 523-524 work for any balanced model• Rules are consistent with the restricted mixed model – can be
modified to incorporate the unrestricted model assumptions
2( ) (fixed factor)E MS f
Chapter 13 Design & Analysis of Experiments 7E 2009 Montgomery
32
Approximate F Tests
• Sometimes we find that there are no exact tests for certain effects (page 525)
• Leads to an approximate F test (“pseudo” F test)• Test procedure is due to Satterthwaite (1946), and uses linear
combinations of the original mean squares to form the F-ratio• The linear combinations of the original mean squares are
sometimes called “synthetic” mean squares• Adjustments are required to the degrees of freedom• Refer to Example 13.7• Minitab will analyze these experiments, although their
“synthetic” mean squares are not always the best choice
Chapter 13 Design & Analysis of Experiments 7E 2009 Montgomery
33
Notice that there are no exact tests for main effects
Chapter 13 Design & Analysis of Experiments 7E 2009 Montgomery
34
Chapter 13 Design & Analysis of Experiments 7E 2009 Montgomery
35
Chapter 13 Design & Analysis of Experiments 7E 2009 Montgomery
36
Chapter 13 Design & Analysis of Experiments 7E 2009 Montgomery
37
Chapter 13 Design & Analysis of Experiments 7E 2009 Montgomery
38
We chose a different linear combination
Chapter 13 Design & Analysis of Experiments 7E 2009 Montgomery
39
Chapter 13 Design & Analysis of Experiments 7E 2009 Montgomery
40
JMP uses a maximum likelihood approach to analyzing models with random factors
Chapter 13 Design & Analysis of Experiments 7E 2009 Montgomery
41