{[ promptMessage ]}

Bookmark it

{[ promptMessage ]}

lect21 - IE 410 Lecture 21 Nested and Hierarchical Designs...

Info iconThis preview shows pages 1–23. Sign up to view the full content.

View Full Document Right Arrow Icon
IE 410 Lecture 21: Nested and Hierarchical Designs and Rules for Expected Mean Squares
Background image of page 1

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full Document Right Arrow Icon
Two Stage Nested Designs Consider the following: Example: A Company deals with 3 suppliers of chemicals. Each Company ships chemicals in Batches.
Background image of page 2
We suspect variation in chemicals is messing up our production and thus do an experiment with two factors:
Background image of page 3

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full Document Right Arrow Icon
Factor A: Supplier (levels = 1, 2, and 3) Factor B: Batch (levels 1, 2, 3, and 4) In this experiment it seems that Factor A is fixed while Factor B is random. For each combination of batch and supplier we test for purity of the chemical three times (in order to have replicates). Thus the data looks like this:
Background image of page 4
Supplier 1 Supplier 2 Supplier 3 ------------------------------------- 1 1 2 1 -1 -2 4 0 -3 0 ------------------------------------- -2 0 -2 2 -3 4 0 -4 2 2 Batches ------------------------------------- -2 -1 1 3 0 0 -1 1 -2 2 ------------------------------------- 1 0 3 4 4 3 2
Background image of page 5

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full Document Right Arrow Icon
0 2 1 -------------------------------------
Background image of page 6
so a=3 b=4 n=3 This design looks like a two factor factorial experiment, BUT IT IS NOT Another way to display the data is like this:
Background image of page 7

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full Document Right Arrow Icon
Supplier 1| Supplier 2| Supplier 3 ----------|-----------|----------- Bch 1 2 3 4| 1 2 3 4| 1 2 3 4 ----------|-----------|----------- 1 -2 -2 1| 1 0 -1 0| 2 -2 1 3 Rp -1 -3 0 4|-2 4 0 3| 4 0 -1 2 0 -4 1 0|-3 2 -2 2| 0 2 2 1 Why isn't this a factorial design?
Background image of page 8
Because there are really 12 batches involved. That is, batch 1 under supplier 1 is not the same as batch 1 under supplier 2. The data could just as well be written this way:
Background image of page 9

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full Document Right Arrow Icon
Supplier 1| Supplier 2| Supplier 3 ----------|-----------|----------- Bch 1 2 3 4| 5 6 7 8| 9 10 11 12 ----------|-----------|----------- 1 -2 -2 1| 1 0 -1 0| 2 -2 1 3 Rp -1 -3 0 4|-2 4 0 3| 4 0 -1 2 0 -4 1 0|-3 2 -2 2| 0 2 2 1 This is a "Nested Design " We say that factor B is "nested within factor A"
Background image of page 10
The example above is a "balanced: nested design” which means that Factor B is at b levels UNDER EACH level of factor A. We shall deal only with balanced nested designs.
Background image of page 11

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full Document Right Arrow Icon
Linear Statistical Model Y ijk = μ + τ i + β j(i) + ε (ij)k where the indexing notation is important to understand: j(i) means "j within i" or "j nested under i"
Background image of page 12
Note also that the replicates index k is nested within combinations of A and B. In factorial designs, replicates are also nested in this way. In fact replicates are, in truth, always nested like this.
Background image of page 13

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full Document Right Arrow Icon
For the fixed effects model we add the following constraints: i τ i = 0 j β j(i) = 0 for all i
Background image of page 14
Decomposition of Sums of Squares SST can be decomposed as follows: SST = i j k[ Y ijk - Y ...] 2 SSA = bn i [ Y i.. - Y ...] 2
Background image of page 15

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full Document Right Arrow Icon
SSB(A) = n i j [ Y ij. - Y i.. ] 2 SSE = i j k [Y ijk - Y ij. ] 2
Background image of page 16
SST = SSA + SSB(A) + SSE df abn-1 a-1 a(b-1) ab(n-1) Hopefully this makes sense. A is at a levels --> a-1 df
Background image of page 17

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full Document Right Arrow Icon
At each level of a we have B at b levels --> b-1 df thus overall a levels of a we have a(b-1) df for factor B
Background image of page 18
Recall the analysis of the two factor factorial design SST = SSA + SSB + SSAB + SSE df abn-1 a-1 b-1 (a-1)(b-1) ab(n-1) We could easily show that if we take the two factor factorial SS then:
Background image of page 19

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full Document Right Arrow Icon
SSB + SSAB = SSB(A) from factorial From nested
Background image of page 20
This is the end of the preview. Sign up to access the rest of the document.

{[ snackBarMessage ]}