- Page 1 and 2: Regression-Discontinuity Design Day
- Page 3 and 4: Design Overview Design Examples Vis
- Page 5 and 6: RDD Visual Depiction Comparison
- Page 7 and 8: RDD Visual Depiction Discontinuity,
- Page 9 and 10: Two Rationales for RDD 1. Selection
- Page 11: Required Assumptions for RD design
- Page 15 and 16: Estimating Treatment Effects in Fuz
- Page 17 and 18: “Non”-RD designs • No discont
- Page 19 and 20: 2. Manipulation of the assignment v
- Page 21 and 22: Diagnostic test • Can perform “
- Page 23 and 24: Analytic Threats to RD Design: Para
- Page 25 and 26: The size of the discontinuity at th
- Page 27: Functional Form • Put more techni
- Page 30 and 31: Functional Form: Interactions • S
- Page 32 and 33: If we superimpose the regression li
- Page 34 and 35: Here we see an example where the tr
- Page 36 and 37: Adding Nonlinear Terms to the Model
- Page 38 and 39: Adding Nonlinear and Interaction Te
- Page 40 and 41: Adding Terms, Continued • When in
- Page 42 and 43: Example of Parametric Plot used aga
- Page 44 and 45: Non-parametric Approach lim x¯c E[
- Page 46 and 47: 3. Semi-Parametric Approach Y i = t
- Page 48 and 49: Problem of Incorrect Bandwidth Sele
- Page 50 and 51: RDD Example 1
- Page 52: RD Design assumption 1 • Probabil
- Page 55 and 56: Implementation assumption 2 • No
- Page 57 and 58: Analytic assumption 1 continued •
- Page 59 and 60: Shadish et al. (2011) Vocabulary RD
- Page 61 and 62: RDD Example 2
- Page 63 and 64:
Best case scenario - regression lin
- Page 65 and 66:
Correct specification of functional
- Page 67 and 68:
What to do (2) • Parametric appro
- Page 69 and 70:
What to do (4) 0 50 100 150 locpoly
- Page 71 and 72:
What to do (5) • Move to a tie br
- Page 73 and 74:
Example Cocaine Project • Include
- Page 75 and 76:
What to do (7) • Estimation throu
- Page 77 and 78:
2. Dealing with Overrides to the cu
- Page 79 and 80:
2. Overrides to the cutoff • Mult
- Page 81 and 82:
Specification of Functional Form Pa
- Page 83 and 84:
Specification of Functional Form No
- Page 85 and 86:
Overrides to cutoff (2) Alternative
- Page 87 and 88:
Final estimates Function al form Pa
- Page 89 and 90:
Addressing Potential Issues with us
- Page 91 and 92:
African-American Home language Span
- Page 93 and 94:
What about cutoff Shift in Number o
- Page 95 and 96:
McCrary Test for New Jersey
- Page 97 and 98:
Take Home Message • Like RE, the
- Page 99 and 100:
Inadequate statistical power • Ne
- Page 101 and 102:
Power Considerations Common to Rand
- Page 103 and 104:
What “Uniquely” Affects Power i
- Page 105 and 106:
Generalizing Beyond The Cut-off In
- Page 107 and 108:
Adding an Untreated RD Comparison F
- Page 109 and 110:
Creating A Regression Discontinuity
- Page 111 and 112:
Pretest and Posttest RD • Imagine
- Page 113 and 114:
What Is The Target Parameter? • A
- Page 115 and 116:
Table 1: Bias At the Cutoff Adding
- Page 117 and 118:
Summary of what the Pretest- Supple
- Page 119 and 120:
Additional Topic not covered in Cla
- Page 121 and 122:
Sample Data with Multiple Cutoffs O
- Page 123 and 124:
Typical estimation strategy for RDs
- Page 125 and 126:
Issue 1: To standardize or not stan
- Page 127 and 128:
Generalizing Treatment Effects Acro
- Page 129 and 130:
…Also Common in • Meta-Analysis
- Page 131 and 132:
What is Special about Pooling RD Es
- Page 133 and 134:
Benefits and Tradeoffs of Option 1
- Page 135 and 136:
Benefits and Tradeoffs of Option 2
- Page 137 and 138:
Distribution of Units in an RD with
- Page 139 and 140:
Multivariate RDD with Two Assignmen
- Page 141 and 142:
The Average Treatment Effect along
- Page 143 and 144:
Treatment Weights for τ MRD Weight
- Page 145 and 146:
Estimating Treatment Effects Four P
- Page 147 and 148:
Centering Approach Procedure allows
- Page 149 and 150:
IV Approach (1) Rather than exclude