Title: Statistics and Art: Sampling, Response Error, Mixed Models, Missing Data, and Inference
1Statistics and Art Sampling, Response Error,
Mixed Models, Missing Data, and Inference
2(No Transcript)
3- Outline
- Example Dose-response Models in Toxicology-
Threshold vs Hormetic Models - What is truth? Predict what?
- Subsets- sampling
- Prediction
- Results on Predictor of Realized Subject True
Value - Illustration and Dilemma
- Extension to two-stage problems
- Missing data framework
- Conclusions
41. Example Dose-response Models - Threshold vs
Hormetic Models
- Yeast data- 2189 chemicals, 13 yeast strains, 5
doses x 2 replications- Focus on doses below BMD
5i chemical J dose k replication
6(No Transcript)
7(No Transcript)
82. What is truth? Predict what?
Population, subjects, true response Subject
Labels True Response
Population Parameters Mean Variance
Subject Deviation
9Non-Stochastic Model Index for
response Response error Assume
Response Error ModelFor each subject
103. Subsets, Sampling
- Select n of N subjects (a subset, sample)
- Let all subsets be equally likely
Sample Mean Note difference with
11Sample as a Sequence(part of Permutation)
- Represent Positions in a Permutation
- Assume all Permutations Equally Likely
Define Sample positions Sample Mean
12Population
s2 Ed
s3 Wenjun
s1 Julio
13Position in Permutation
i1
i2
i3
s2
s3
s1
14i1
i2
Position in Permutation
i1
i2
i3
i3
s2
s1
s3
s1
s3
s2
15Position in Permutation
i1
i2
i3
s3
s1
s2
16Position in Permutation
i1
i2
i3
s3
s2
s1
17 Position in Permutation
i1
i2
i3
Sample
Remainder
s1
s2
s3
18 Position in Permutation
i1
i2
i3
Sample
Remainder
s2
s1
s3
19- Population size (N) is most likely gt 3
- We only see n subjects in the sample
- For example Suppose n3, and N7
- We may see
20 Position in Permutation
i1
i2
i3
Sample
Remainder
i4
i
s3
s4
s5
21 Position in Permutation
i1
i2
i3
Sample
Remainder
s2
s4
s7
i
22Traditional Sampling Approach
1
2
N
Horvitz-Thompson Estimator
Missing Data
23With Response Error Model
Sample Mean
Sample is a Sequence
To represent positions
24 Position in Permutation
i1
i2
i3
Sample
s1
s2
s3
25First Position in Permutation
Suppose s1,,3N
Then
26Positions in Sample Sequences
Sample
Remainder
27Basic Random Variables
Sample
Remainder
Population
28Response Error Model
Response Error Model
Finite Population Mixed Model
29Mixed Model
Mixed Model
30Properties of Basic Random Variables (N3)
Sum
Expected Value
Sum
Average
Expected Value
Average
31Sample Random Variables (n2)
Sum
Expected Value
Sum
Expected Value
32Prediction of Mean in a Simple Case No Response
Error (N3, n2)
Sample
Remainder
Note Criteria Linear Function of
sample Unbiased Smallest Mean Squared Error
33Prediction of Mean No Response Error (N3, n2)
Target
Sample Data
Realized
Best Linear Unbiased Predictor
34Prediction of a Subjects Mean in Position i with
No Resp. Error (N3, n2)
Target
Sample Data
Realized
Best Linear Unbiased Predictor
35Prediction of a Subjects Mean in Position i with
Response Error
Target
Sample Data
Realized
Best Linear Unbiased Predictor
36Prediction of Realized Random Effect Other
Examples
SRS Subject Resp. Error
SRS Position Resp. Error
Cluster Sampling Balanced
Cluster Sampling Un-Balanced
Similar form, more complicated
37(No Transcript)
38(No Transcript)
39Delimma
- Pooled Response Error Variance should be used for
K (Using theoretical Results) - Empirical example illustrates smaller MSE results
with K depending on realized Subject -- but no
theory! - What should we do?.... Is there a gap in the
framework?
40Basic Sample Random Variables
Sum
Sum
41Basic Random Variables
42More Work is needed!
Thanks