Questions tagged [dag]
DAG stands for Directed Acyclic Graph. DAGs are commonly used to help people think about causal patterns amongst variables.
161
questions
1
vote
0
answers
63
views
Is there any algorithmic-pseudocode/Python-implementation for the DAG-related API(s) in the R package dagitty? [closed]
The statistical software DAGgity offers a graphical web-based UI as well as an implementation in R that allows for finding conditional independences corresponding to d-separation, minimal adjustment ...
1
vote
1
answer
61
views
Should we adjust for this variable?
Four variables $X$, $Y$, $A$, and $B$ are assumed to have relationships as in the following diagram:
Here, $X$ is the predictor and $Y$ is the outcome variable.
Suppose that the research interest is ...
- 2,640
4
votes
2
answers
59
views
Variable selection with a theoretical DAG vs algorithmically discovered DAG
I'm analysing data from an electronic health record and determining what variables to include in a model to close back doors and omit bias.
I've read that it is important to have a subject specific ...
- 521
2
votes
1
answer
23
views
May unobserved variable confound or create open backdoor paths, why didn't controlling for the collider O make bad?
Is the U, the unobserved creating an open backdoor path or confounding?
Why condition on the collider Occupation good here?
user384464
0
votes
0
answers
16
views
If you condition on O, you stop the flow of information from which arrow into or out of O? [duplicate]
Why does conditioning on O open up this second channel below?
- 1
1
vote
0
answers
14
views
Can my cycle-cut algorithm sample any DAG?
I am doing some computational simulations to validate a procedure involving structural equation models and causal inference. I wanted to sample from the possible space of DAGs on $n$ vertices. I can ...
- 6,988
3
votes
1
answer
38
views
Questions on estimating causal influence from Directed Acyclic Graphs
I am reading McElreath's Statistical Rethinking book and I'm wondering if anyone can help clarify my doubts on confronting confounding with DAGs. I'll specify 2 examples:
Taken from link. The DAG ...
- 461
3
votes
1
answer
47
views
Selection of Competing Exposures in a DAG: How many to use?
I am trying to get my head around when and how to include Competing Exposures in DAGs. In my searching I keep finding statements that are very similar to the quote from the dagitty tutorials - "...
- 43
0
votes
0
answers
11
views
Markov and Faithfullness Condition in disconnected DAG/Bayesian network
I'm very confused about the Markov- and the faithfulness condition in disconnected DAGs, as I've never seen such examples.
Assume for example I had a DAG where X -> Y, Z. Thus, Z is disconnected ...
- 11
6
votes
1
answer
279
views
Does information criteria (AIC, BIC and DIC...) imply "causality"?
I am interested in finding out the graphical causal structure. Causal Discovery algorithms (e.g., DAG learning) are used to identify potential causal graphs. In score-based causal discovery methods, ...
- 61
1
vote
0
answers
12
views
In a causal diagram where the change score $\Delta Y=Y_1-Y_0$, does $Y_1$ cause $\Delta Y$ or does $\Delta Y$ cause $Y_1$?
In Shahar & Shahar (2010), the authors argue that in a typical pre/post observational study on two longitudinal outcomes $Y_0$ and $Y_1$:
the change score $\Delta Y=Y_1-Y_0$ causes the post-...
- 5,442
4
votes
1
answer
87
views
Covariate adjustment for "mediated" reverse causality
I have made the following DAG in dagitty.net:
DAGitty says "The total effect cannot be estimated by covariate adjustment".
I don't understand why I can't close the backdoor path by ...
- 2,040
1
vote
1
answer
26
views
Selection Bias in Conflict Studies
A common critique I have heard levied against conflict studies (research examining the causes, consequences, and solutions to violence such as civil war, terrorism, etc.) is the problem of selection ...
- 519
1
vote
1
answer
27
views
How to simplify the following conditional probability distributions using the given DAG?
Using the above DAG I need to simplify the following conditional probabilities:
$$i) \quad p(x_4|x_1,x_2)$$
For this one I guess I can just remove the conditioning on $x_1$ (using the DAG) and the ...
- 113
4
votes
2
answers
79
views
According to DAG theory, why controlling for this variable doesn't close the backdoor path opened by controlling for the collider?
I have made the following model in DAGitty:
Where $X_2$ is controlled for.
DAGitty says:
The total effect cannot be estimated due to adjustment for an intermediate or a descendant of an intermediate....
- 2,040
1
vote
2
answers
90
views
DAG - why is the path open?
I have this DAG
As I understand it, the paths D <- Ed -> St -> P -> Su and D <- A -> P -> Su are both closed because the contain the collider P. If I condition on P, both these ...
0
votes
1
answer
45
views
Biased sample same like conditioning on a collider?
I am studying chapter 5 "The many variables & the spurious waffles" of the book Rethinking and trying to answer the following question:
How is biased sample like conditioning on a ...
- 407
2
votes
2
answers
130
views
Can controlling for a variable block the backdoor path opened by controlling for a collider?
I have made the following model in DAGitty:
Where X2 is controlled for.
DAGitty says:
The total effect cannot be estimated due to adjustment for an intermediate or a descendant of an intermediate.
...
- 2,040
0
votes
0
answers
32
views
Why implied Conditional Independencies of mediator and confounder are the same?
I am trying to understand why the impliedConditionalIndependencies function of the rethinking package returns the same value for ...
- 407
3
votes
2
answers
62
views
Causal modeling in the presence of a latent variable
Suppose that four variables of $X$, $Y$, $L$, and $C$ have the following relationships in the form of directed acyclic graph.
$X$, $Y$, and $C$ are observable variables while $L$ is a latent (...
- 2,640
1
vote
1
answer
60
views
Adjusting for variables outside of minimal adjustment set for total causal effect in a DAG
I have a fairly elaborate Directed Acyclic Graph (DAG) for the analysis that I am running, but I am presenting a simplified example here to clarify a few things.
Here is a DAG from dagitty.net:
...
- 55
3
votes
1
answer
130
views
A graph that helps understand DGP of a mixture of Gaussians (DAG, Kruschke's DBDA style graph, etc.)?
Hastie, et al in “Elements of Statistical Learning” describe a particular DGP(Data Generating Process) or causal model on page 13.
The training data in each class (2 classes total) came from a ...
- 4,166
4
votes
1
answer
73
views
Presenting DAGs in Journal-Quality Research
One of the benefits of DAGs is that they openly state the causal assumptions a researcher is making, allowing for greater transparency. This is nice in theory. However, in practice, the DAGs I have ...
- 519
5
votes
1
answer
73
views
DAGs with Ambiguous Temporal Ordering Between Nodes
I am working on a project where I am attempting to estimate the causal impact of civil war peace agreements (treatment) on levels of violence (outcome). While developing the DAG, I came across an ...
- 519
1
vote
0
answers
22
views
Modeling the causal impact of insurance benefits on customer satisfaction
Consider the following scenario:
A health insurance company offers a few dozen different health insurance plans.
Within each plan there are many benefits which can take on different values, such as ...
- 657
0
votes
0
answers
28
views
Understanding DAGs
I am learning about DAGs. On Daggity, I entered in the following DAG. Daggity indicates that I can estimate the direct effect of the test on the the outcome by adjusting for B. But my understanding is ...
- 1
3
votes
0
answers
76
views
How can I compute the set of interventional probability distributions compatible with a DAG?
Let $P$ be a probability distribution on a set $V$ of variables and for any $X\subseteq V$ and any possible realization $x$ of $X$ let $P_x$ be a distribution on $V\setminus X$. Let $P^*$ the ...
0
votes
0
answers
25
views
In covariate-specific effect - $P(Y=y|do(X=x),Z=z)$ - is $Z=z$ pre- or post-treatment measure?
Causal Inference In Statistics by Pearl, section 3.5, page 70 clearly mentions that -
This effect, written $P(Y=y|do(X=x),Z=z)$, measures the distribution
of $Y$ in a subset of the population for ...
0
votes
3
answers
216
views
When we say extract a causal DAG from a multivariate time series, what does it actually mean?
I am from CS background and as part of my PhD, I am doing a project where I need to used causal inference to construct a causal DAG (directed acyclic graph) from a multivariate time series data. As I ...
- 11
2
votes
1
answer
50
views
What is a "directed path" in context of causal graphs?
I am going through Causal Inference In Statistics by Pearl and I have come across the definition of path and directed path (section 1.4, page 25).
Path: A path between two nodes $X$ and $Y$ is a ...
1
vote
1
answer
52
views
Pearl's Causal Inference In Statistics, equation 3.11 - Calculation of group specific causal effects
In the book Causal Inference In Statistics by Pearl, page 63, while referring to the below DAG, it says -
Thus to compute the $w$-specific causal effect, written
$P(y|do(x),w)$, we adjust for $T$, ...
1
vote
1
answer
29
views
What does the causal path of controlled direct effect look like on the graph?
For the given graph above, the controlled direct effect will be
$E[Y|\operatorname{do}(X),\operatorname{do}(M)]$. This would break all the incoming edges to node X and M, so $X\rightarrow M$ is ...
- 11
2
votes
1
answer
71
views
Representation of unconfoundedness of Rubin causal models on Pearl causal models
Disclaimer: I'm new to the potential outcome framework so this question may not make lot of sense, in case please let me understand where I am failing.
I was reading this question about the ...
- 306
1
vote
1
answer
386
views
Intuition of conditional independence in DAGs
In the DAG above, we have $A$ conditionally independent of $E$ when $C$ and $B$ are observed (that is $A\perp E|B,C$), but not when only $C$ is observed (that is $A\not\perp E|C$). I don't have a ...
- 13
0
votes
0
answers
9
views
dependence for variables that are not d-separated
I need to show that for a linear SEM having X->Y<-Z means that X and Z are dependent conditional on Y. For a linear SEM with errors that have finite variances this is doable, but for a model ...
- 1
6
votes
2
answers
540
views
Can we just "pre-test" the backdoor criterion?
I am trying to use DAGs to think more carefully about my regression models. I have a question about the "backdoor criterion", as usually seen in the DAG below (we are interested in the ...
- 470
4
votes
1
answer
463
views
Advantage of applying DAG over SEM?
I recently learned about DAG-Directed Acyclic Graphs while reading econometrics-related papers. So far, I am used to Structural Equation Modeling for Psychology and Education studies, and DAG looks ...
- 51
0
votes
0
answers
21
views
Regression on dataset whose features form a DAG
I have a dataset with the feature set X that comes in a hierarchical fashion. In other words, X forms a DAG that ends at the dependent variable y. The question is, is there a regression model that can ...
6
votes
1
answer
4k
views
What is G-computation and G-estimation in causal inference
I am looking for a beginner-friendly explanation of what Robbins' G-computation and G-estimation are for estimating causal effects. What problem(s) do they solve. Ideally I would like an example of ...
- 161
2
votes
1
answer
195
views
Causal inference - effect modifier
I have a somewhat complicated DAG that looks like this:
where Y is the outcome variable. To give some context, X and Y could be two diseases with X being the precursor to disease Y. A is age, and B ...
- 127
1
vote
0
answers
38
views
Conditional equivalent class in DAGs (Bayesian network, causal graph)
Suppose we are dealing with a prediction problem, the target variable is $Y$, the predictive covariates are chosen from $\mathbf{X}=\{X_1,X_2,...,X_n\}$. The causal graph (DAG) on $Y,\mathbf{X}$ is ...
- 361
7
votes
1
answer
320
views
Pearl, Causal Inference in Statistics Q3.5.1 (Backdoor criterion)
This is a question about backdoor criterion (as per J. Pearl) on finding causal effects. It is linked to a specific exercise in a specific book, but I hope it will be sufficiently generic and self-...
- 426
3
votes
2
answers
209
views
How to evaluate validity about Causal Discovery?
When I was trying to do structural learning for causal inference, I perplex for too many causal discovery algorithms(PC, FCI, GES, NOTEARS and more...)
There are many structural learning algorithm for ...
- 63
2
votes
0
answers
86
views
Chapter 4.3.1 table 4.2 and Equation (4.13), (4.14) from Pearl et al. "Causal Inference in Statistics: A Primer"
$\begin{equation}\begin{split} X & = u_{1} \\ Z & = a X + u_{2} \\ Y & = b Z \end{split}\end{equation}$
$u_{1}$
$u_{2}$
$X(u)$
$Z(u)$
$Y(u)$
$Y_{0}(u)$
$Y_{1}(u)$
$Z_{0}(u)$
$Z_{1}(u)$
0
...
- 63
2
votes
0
answers
119
views
Directed Acyclic Graph including a categorical variable with 20 levels
Is it possible within causal inference using DAGs to sensibly include a categorical variable with 20 levels? I have seen that regression trees can be used in this situation but not in combination with ...
- 333
2
votes
0
answers
308
views
What should I study after finishing 'Causal Inference in Statistics: A Primer'?
I have almost finished studying 'Causal Inference in Statistics: A Primer', but I still feel that I need to learn more.
I considered 'Causality' (Pearl, 2009), but there seem to be several good ...
1
vote
0
answers
30
views
How can I efficiently test whether a subsets of nodes in a DAG are "lined up" more often than expected by random chance?
I have a directed acyclic graph with N nodes, each of which is assigned to one of K groups, with K < N. My hypothesis is that nodes in the same group tend to "line up" along a linear path....
- 343
2
votes
0
answers
115
views
Question about using potential outcomes in DAGs in real world example
I am trying to understand how DAGs and potential outcomes look together. I came across these excellent posts (here and here, but I am trying to understand how this looks in a real world example. ...
- 29
3
votes
1
answer
371
views
Meaning of multiple exposures in a DAG
I am working with DAGs as a way to do some causal modeling. I am using dagitty - both the website and the R package. I feel like I have a good grasp of most things related to confounding, adjustment ...
- 111
0
votes
0
answers
143
views
How is backdoor criterion used in practice?
Is the backdoor criterion applicable only for "learning" in a causal model (i.e. for estimating the causal effects between variables) or must it also be used when running that model, as in ...
- 11