You are on page 1of 49

Topic Outline

1) Motivation
2) Representing/Modeling Causal Systems

3) Estimation and Updating


4) Model Search
5) Linear Latent Variable Models
6) Case Study: fMRI

1
Discovering
Pure Measurement Models

Richard Scheines
Carnegie Mellon University

Ricardo Silva*
University College London

Clark Glymour and Peter Spirtes


Carnegie Mellon University
2
Outline

1. Measurement Models & Causal Inference


2. Strategies for Finding a Pure Measurement Model
3. Purify
4. MIMbuild
5. Build Pure Clusters
6. Examples
a) Religious Coping
b) Test Anxiety

3
Goals:
• What Latents are out there?
• Causal Relationships Among Latent Constructs

Relationship
Depression Satisfaction

or
Relationship
Depression Satisfaction

or ?
4
Needed:

Ability to detect
conditional independence
among latent variables

5
Lead and IQ
e2
Parental Resources e3

Lead IQ
Exposure

Lead _||_ IQ | PR

PR ~ N(m=10, s = 3)

Lead = 15 -.5*PR + e2 e2 ~ N(m=0, s = 1.635)

IQ = 90 + 1*PR + e3 e3 ~ N(m=0, s = 15)

6
Psuedorandom sample: N = 2,000

Parental Resources

Lead
Exposure IQ

Regression of IQ on Lead, PR

Independent Coefficient Estimate p-value Screened-off


Variable at .05?

PR 0.98 0.000 No

Lead -0.088 0.378 Yes

7
Measuring the Confounder
e1 e2 e3

X1 X2 X3

Parental
Lead Resources
IQ
Exposure

X1 = g1* Parental Resources + e1

X2 = g2* Parental Resources + e2

X3 = g3* Parental Resources + e3

PR_Scale = (X1 + X2 + X3) / 3

8
Scales don't preserve conditional independence

X1 X2 X3

Parental
Lead Resources
IQ
Exposure

PR_Scale = (X1 + X2 + X3) / 3

Independent Coefficient p-value Screened-off


Variable Estimate at .05?

PR_scale 0.290 0.000 No

Lead -0.423 0.000 No

9
Indicators Don’t Preserve Conditional Independence

X1 X2 X3

Parental
Lead Resources
IQ
Exposure

Regress IQ on: Lead, X1, X2, X3

Independent Coefficient p-value Screened-off


Variable Estimate at .05?

X1 0.22 0.002 No
X2 0.45 0.000 No
X3 0.18 0.013 No
Lead -0.414 0.000 No
10
Structural Equation Models Work

X1 X2 X3

Parental
Resources
Lead
Exposure IQ

Structural Equation Model


• E ( ˆ )  0
• ˆ  .07(p-value = .499)
• Lead and IQ “screened off” by PR
11
Local Independence / Pure Measurement Models

• For every measured item xi:


xi _||_ xj | latent parent of xi

F1 F2 F3

x1 x2 x3 x4 x5 x6 x7 x8 x9 x10 x11 x12

12
Local Independence Desirable
Truth

F1 F2 F3

x1 x2 x3 x4 y1 y2 y3 y4 z1 z2 z3 z4

Specified Model 31

1 2 3

x1 x2 x3 x4 y1 y2 y3 y4 z1 z2 z3 z4

E ( ˆ31 )  0 13
Correct Specification Crucial
Truth

F1 F2 F3

x1 x2 x3 x4 y1 y2 y3 y4 z1 z2 z3 z4

F4

Specified Model 31

1 2 3

x1 x2 x3 x4 y1 y2 y3 y4 z1 z2 z3 z4

E ( ˆ31 )  0 14
Strategies

• Find a Locally Independent Measurement Model

• Correctly specify the MM,


including deviations from Local Independence

15
Correctly Specify Deviations from Local Independence
Truth

F1 F2 F3

x1 x2 x3 x4 y1 y2 y3 y4 z1 z2 z3 z4

F4

Specified Model 31

1 2 3

x1 x2 x3 x4 y1 y2 y3 y4 z1 z2 z3 z4

x4 z4

E ( ˆ31 )  0 16
Correctly Specifying Deviations from Local Independence
is Often Very Hard

Truth

F1 F2 F3

x1 x2 x3 x4 y1 y2 y3 y4 z1 z2 z3 z4

F5 F4 F6

17
Finding Pure Measurement Models -
Much Easier

Truth

F1 F2 F3

x1 x2 x3 x4 y1 y2 y3 y4 z1 z2 z3 z4

F5 F4 F6

Truth

F1 F2 F3

x1 x2 x3 y1 y2 y3 y4 z3 z4

F5 F6

18
Tetrad Constraints

• Fact: given a graph with this structure

L W =  1 L + 1
1 2 4 X = 2 L +  2
3
Y = 3L + 3
W X Y Z Z = 4L + 4
• it follows that tetrad
constraints
CovWXCovYZ = (122L) (342L) =
= (132L) (242L) = CovWYCovXZ

WXYZ = WYXZ = WZXY


Early Progenitors

Charles Spearman (1904)

Statistical Constraints  Measurement Model Structure

g
rm1 * rr1 = rm2 * rr2
m1 m2 r1 r2
Impurities/Deviations from Local Independence
defeat tetrad constraints selectively

Truth Truth

F1 F1

x1 x2 x3 x4 x1 x2 x3 x4

F5

rx1,x2 * rx3,x4 = rx1,x3 * rx2,x4 rx1,x2 * rx3,x4 = rx1,x3 * rx2,x4

rx1,x2 * rx3,x4 = rx1,x4 * rx2,x3 rx1,x2 * rx3,x4 = rx1,x4 * rx2,x3

rx1,x3 * rx2,x4 = rx1,x4 * rx2,x3 rx1,x3 * rx2,x4 = rx1,x4 * rx2,x3


21
Purify

True Model F1 F2 F3

x1 x2 x3 x4 y1 y2 y3 y4 z1 z2 z3 z4

Initially Specified F3
F1 F2
Measurement Model

x1 x2 x3 x4 y1 y2 y3 y4 z1 z2 z3 z4

22
Purify
Iteratively remove item whose removal most improves
measurement model fit (tetrads or c2)
– stop when confirmatory fit is acceptable

F1 F2 F3

Remove x4

x1 x2 x3 x4 y1 y2 y3 y4 z1 z2 z3 z4

F1 F2 F3
Remove z2

x1 x2 x3 x4 y1 y2 y3 y4 z1 z2 z3 z4

F
Purify

Detectibly Pure Subset of Items


Detectibly Pure Measurement Model

F1 F2 F3

x1 x2 x3 y1 y2 y3 y4 z1 z3 z4

24
Purify

25
How a pure measurement model is useful

F1 F2 F3

x1 x2 x3 y1 y2 y3 y4 z1 z3 z4

1. Consistently estimate covariances/correlations among latents


- test conditional independence with estimated latent correlations

2. Test for conditional independence among latents directly


2. Test conditional independence relations
among latents directly

Question: L1 _||_ L2 | {Q1, Q2, ..., Qn}

b21

b21 = 0  L1 _||_ L2 | {Q1, Q2, ..., Qn}


MIMbuild

Input:
- Purified Measurement Model
- Covariance matrix over set of pure items

MIMbuild
PC algorithm with independence tests
performed directly on latent variables

Output: Equivalence class of structural models


over the latent variables

28
Purify & MIMbuild

29
Goal 2: What Latents are out there?

• How should they be measured?

30
Latents and the clustering of items they measure
imply tetrad constraints diffentially

Model 1 Model 2

F1 F2 F1 F2

x1 x2 x3 x4 x5 x5 x1 x2 x3 x4 x5 x5

Model 3 Model 4

F3 F1 F2 F3
F1 F2

x6 x1 x2 x3 x4 x5 x6
x1 x2 x3 x4 x5

31
Build Pure Clusters (BPC)
Input:
- Covariance matrix over set of original items

BPC
1) Cluster (complicated boolean combinations of tetrads)
2) Purify

Output: Equivalence class of measurement models over a


pure subset of original Items

32
Build Pure Clusters

33
Build Pure Clusters

Qualitative Assumptions
1. Two types of nodes: measured (M) and latent (L)
2. M L (measured don’t cause latents)
3. Each m  M measures (is a direct effect of) at least one l  L
4. No cycles involving M

Quantitative Assumptions:
1. Each m  M is a linear function of its parents plus noise
2. P(L) has second moments, positive variances, and no deterministic
relations

34
Build Pure Clusters
Output - provably reliable (pointwise consistent):
Equivalence class of measurement models over a pure subset of M
For example:

L1 L2 L3

True Model
m1 m2 m3 m4 m5 m6 m10 m11 m7 m8 m9

L1 L2 L3
Output

m1 m2 m3 m4 m5 m6 m7 m8 m9

35
Build Pure Clusters
Measurement models in the L1 L2 L3 L4
equivalence class are at most
refinements, but never
coarsenings or permuted m1 m2 m3 m4 m5 m6 m7 m8 m9
clusterings.

L1 L3
L1 L2 L3

m1 m2 m3 m 4 m 5 m6 m7 m 8 m 9
m 1 m 2 m3 m4 m5 m6 m7 m8 m9

L1 L2 L3
Output
m1 m2 m3 m4 m 5 m6 m7 m 8 m 9

36
Build Pure Clusters
Algorithm Sketch:
1. Use particular rank (tetrad) constraints on the measured correlations
to find pairs of items mj, mk that do NOT share a single latent parent
2. Add a latent for each subset S of M such that no pair in S was found
NOT to share a latent parent in step 1.
3. Purify
4. Remove latents with no children

37
Build Pure Clusters + MIMbuild

38
Case Studies

Stress, Depression, and Religion (Lee, 2004)

Test Anxiety (Bartholomew, 2002)

39
Case Study: Stress, Depression, and Religion

Masters Students (N = 127) 61 - item survey (Likert Scale)


• Stress: St1 - St21
• Depression: D1 - D20
• Religious Coping: C1 - C20

St1 Dep1
12 Specified Model 12
St2 Dep2
1.2 + Depression 12
Stress .
. .
+
-
St21 Dep20
12 Coping 12

p = 0.00
C1 C2 . . C20 40
Case Study: Stress, Depression, and Religion

Build Pure Clusters


St3
12
St4 Dep9
12 12
St16 Stress Depression Dep13
12 12
St18 Dep19
12 Coping 12
St20
12
C9 C12 C14 C15

41
Case Study: Stress, Depression, and Religion

Assume Stress temporally prior:


MIMbuild to find Latent Structure:
St3
12
St4 Dep9
12 + 12
St16 Stress Depression Dep13
12 12
St18 + Dep19
12 Coping 12
St20
12
C9 C12 C14 C15 p = 0.28

42
Case Study : Test Anxiety
Bartholomew and Knott (1999), Latent variable models and factor analysis
12th Grade Males in British Columbia (N = 335)
20 - item survey (Likert Scale items): X1 - X20:

X3
X2 Exploratory Factor Analysis:
X4
X8
X5
X9
X6
Emotionality Worry
X10
X7
X15
X14
X16
X17
X18
X20 43
Case Study : Test Anxiety

Build Pure Clusters:

X2 X3
Cares About
X8 Achieving X5

X9
X7
X10 Emotionalty

X11 X6
Self-
X16 Defeating
X14

X18
44
Case Study : Test Anxiety

Exploratory Factor Analysis: Build Pure Clusters:

X3
X2 X3
X2
X4 Worries About
X8 X8 Achieving X5
X5
X9 X9
X7
X6
Emotionality Worry Emotionalty
X10 X10
X7
X15 X11 X6
X14 Self-
X16 X16 Defeating
X14
X17
X18
X18
X20

p-value = 0.00 p-value = 0.47

45
Case Study : Test Anxiety

MIMbuild Scales: No Independencies or


Conditional Independencies
X2 X3
Worries About Worries About
X8 Achieving X5 Achieving-Scale

X9
X7
Emotionalty-
X10 Emotionalty
Scale

X11 X6
Self- Self-
X16 Defeating Defeating
X14

X18

p = .43 Uninformative

46
Limitations

• In simulation studies, requires large sample sizes to be


really reliable (~ 400-500).
• 2 pure indicators must exist for a latent to be discovered
and included
• Moderately computationally intensive (O(n6)).
• No error probabilities.

47
Open Questions/Projects

• IRT models?
• Bi-factor model extensions?
• Appropriate incorporation of background knowledge

48
References
• Tetrad: www.phil.cmu.edu/projects/tetrad_download
• Spirtes, P., Glymour, C., Scheines, R. (2000). Causation, Prediction,
and Search, 2nd Edition, MIT Press.
• Pearl, J. (2000). Causation: Models of Reasoning and Inference,
Cambridge University Press.
• Silva, R., Glymour, C., Scheines, R. and Spirtes, P. (2006) “Learning the
Structure of Latent Linear Structure Models,” Journal of Machine Learning
Research, 7, 191-246.
• Learning Measurement Models for Unobserved Variables, (2003). Silva, R.,
Scheines, R., Glymour, C., and Spirtes. P., in Proceedings of the Nineteenth
Conference on Uncertainty in Artificial Intelligence , U. Kjaerulff and C.
Meek, eds., Morgan Kauffman

49

You might also like