You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
@@ -27,13 +27,51 @@ is not perfect, and measurement error in these predictions can cause problems
27
27
in downstream analyses.
28
28
29
29
Bayesian Instrumental Regression for Disparity Estimation (BIRDiE) is a class of
30
-
Bayesian models for accurately estimating conditional distributions by race,
31
-
using BISG probabilities as inputs.
30
+
Bayesian models for accurately estimating conditional distributions by race,
31
+
using BISG probabilities as inputs.
32
32
This package implements BIRDiE as described in [McCartan, Fisher, Goldin, Ho, and Imai (2025)](https://doi.org/10.1080/01621459.2025.2526695).
33
-
It also implements standard BISG and an improved measurement-error BISG model as described
33
+
It also implements standard BISG and an improved measurement-error BISG model as described
34
34
in [Imai, Olivella, and Rosenman (2022)](https://www.science.org/doi/full/10.1126/sciadv.adc9824).
35
35
36
+
<details>
37
+
<summary><b>Do I need BIRDiE?</b></summary>
38
+
<p>
39
+
BIRDiE is a statistical model to let you estimate the average value of a variable in different racial groups.
40
+
To take an example from our research paper, if your data are tax records and you want to estimate the rate at which different racial groups take a certain tax credit, BIRDiE can help you do that.
41
+
</p>
42
+
<p>
43
+
BIRDiE is applied on top of individual-level imputations/predictions of race from methods like BISG.
44
+
If your only goal is to estimate individual race probabilities, then BIRDiE is not helpful—BISG alone suffices.
45
+
The graphic below gives an overview of the problem BIRDiE solves and how it fits together with existing methods.
<summary><b>What is the difference between BIRDiE and BISG?</b></summary>
54
+
<p>
55
+
BISG is a simple model that estimates the probability of each individual belonging to diferent racial groups, based on their last name and/or residence location.
56
+
</p>
57
+
<p>
58
+
BIRDiE is a statistical model that takes race probabilities (like BISG) as <i>inputs</i> to estimate the average value of an outcome variable in different racial groups.
59
+
If your research question involves both race and another (outcome) variable, then you likely need to apply BIRDiE on top of BISG to avoid biases caused by measurement error in BISG predictions.
60
+
</p>
61
+
</details>
62
+
63
+
<details>
64
+
<summary><b>Is BIRDiE better than BISG, fBISG, etc?</b></summary>
65
+
<p>
66
+
There are many methods for imputing or predicting individual race, including BISG, fBISG, and others.
67
+
Mainly, these methods use different data sources or slightly different models.
68
+
</p>
69
+
<p>
70
+
BIRDiE is <i>not</i> a replacement for these methods, but rather a complementary tool that uses the outputs of these methods to properly estimate disparities in other variables.
71
+
When BIRDiE is applied on top of these methods, it generally produces far more accurate estimates than directly thresholding or weighting by the outputs of the prediction methods alone.
72
+
</p>
73
+
</details>
74
+
37
75
38
76
## Installation
39
77
@@ -79,7 +117,7 @@ This is the "no-pooling" model from McCartan et al.
79
117
We'll use Gibbs sampling for inference, which will also let us capture the uncertainty in our estimates.
80
118
81
119
```{r}
82
-
fit = birdie(r_probs, turnout ~ proc_zip(zip), data=pseudo_vf,
120
+
fit = birdie(r_probs, turnout ~ proc_zip(zip), data=pseudo_vf,
83
121
family=cat_dir(), algorithm="gibbs")
84
122
85
123
print(fit)
@@ -101,5 +139,5 @@ tidy(fit)
101
139
plot(fit)
102
140
```
103
141
104
-
A more detailed introduction to the method and software package can be found
142
+
A more detailed introduction to the method and software package can be found
105
143
on the [Get Started](https://corymccartan.com/birdie/articles/birdie.html) page.
0 commit comments