Questions tagged [causality]
The relationship between cause and effect.
677
questions
0
votes
1answer
21 views
Is it valid to look at the impact of a feature on residuals?
Background
I'm trying to measure the causal impact of action on outcome (sorry for the vague names, but trying to keep this ...
1
vote
1answer
14 views
hypotheis testing in observational study with propensity score matching to reduce confounding
in observational studies, many people use propensity score matching to reduce confounding (measured co-voriates) between two groups (cohorts). But due to some unobserved confounding co-variates (not ...
0
votes
1answer
25 views
Difference-in-Differences: Matching and Estimating TE for Each Unit and then Using a t-test? [on hold]
I am trying to understand a causal inference method I have recently run across, but I can't find any discussion of this sort of technique in the literature. I don't have a background in econometrics, ...
1
vote
0answers
29 views
Shouldn't there be a rung four on the the Ladder of Causation? [on hold]
Referencing this question and its smart answer:
I'm still confused and the confusion has been in part fueled by "recent" Pearl/Hernán tweets (tweet, tweet and tweet).
Pearl's Ladder of causation ...
0
votes
0answers
14 views
Interpreting the Granger Causality Test
I am trying to go through an online time series analysis tutorial and they use the Granger Causality Test to see if there is some benefit to jointly model two time ...
1
vote
0answers
15 views
Propensity score matching in r using panel data [closed]
I want to conduct PSM using firm panel data in r.
matchit(treat ~ leverage + cash + roa + mtb + asset, data=data)
This gave me a result of only very similar one ...
0
votes
1answer
51 views
Under heterogeneous treatment effects, will the usual unconfoundedness assumption need modification for observational studies?
Suppose that a set of covariates, $X_i$ follows a distribution that is conditional on another variable, $A_i$, for $i \in \{1, \ldots N\}$ individuals. For example, $X_i$ can be income, and $A_i$ can ...
0
votes
1answer
22 views
For the ATT (Average Treatment Effect on the Treated), why is it usually talked about in the population perspective when it is defined w.r.t. samples?
The $ATT$ or the Average Treatment Effect on the Treated, is defined as:
$$
ATT = E[Y(1) - Y(0) | T=1]
$$
for potential outcomes $Y(1), Y(0)$ and treatment indicator $T \in \{0,1\}$. It is my ...
2
votes
1answer
40 views
In an observational study where matching is properly conducted, should we expect the $ATE$ to be equal to the $ATT$?
In a randomized study, I know that the $ATE=ATT=ATC$, where $ATE$ is the average treatment effect, $ATT$ is the average treatment effect on the treated, and $ATC$ is the average treatment effect on ...
0
votes
0answers
56 views
Predicted individual treatment effect with continous treatments
I'm trying to apply Rubin's counterfactual model in an observational setting using machine learning predictions to simulate the unseen treatment-outcome pairs, according to https://www.ncbi.nlm.nih....
0
votes
1answer
68 views
Matching subjects with themselves when evaluating short term outcomes
I am considering a simple causal inference scenario; Let's say we want to examine the effect of paracetamol (treatment) on curing headache (outcome).
When performing matching, is it okay to match ...
2
votes
1answer
38 views
What is the virtue of loading absolutely-summability in the definition of causality of ARMA model?
An ARMA series $y_t$ is causal function of $\nu_t$ if there exists constants $\psi_j$ such that
$\sum_{j=0}^{\infty} |\psi_j|<\infty$ and $y_t=\sum_{j=0}^{\infty} \psi_j\nu_{t-j}<\infty$ for ...
2
votes
4answers
314 views
Are there any differences in causality between linear and logistic regression?
I'm guessing this is a pretty basic question, but I am having a hard time wrapping my head around it.
So my understanding with linear regression, is that it shows how much a change in X, will cause a ...
1
vote
0answers
17 views
Any research on learning Bayesian network structure with a limit on the parent set size?
Learning a maximum-scored Bayesian network structure with bounded treewidth is rather popular in recent years, as stated in the paper A survey on Bayesian network structure learning from data in 2019. ...
3
votes
1answer
79 views
Should I use a machine learning model to calculate propensity score?
In my study, running a simple linear model to calculate de propensity score for each example seemed to not be able to model my treatment choosing process correctly. My question is, does it make sense ...
1
vote
1answer
70 views
Formula for E(MSE) from Recursive Partitioning for Heterogeneous Causal Effects (Athey and Imben 2015)
I'm reading Recursive Partitioning for Heterogeneous Causal Effects (Athey and Imben 2015) and I'm confused about the formula for $\hat E[\mu(x;\Pi)]$ on page 8. The formula offered in the paper is
...
0
votes
1answer
22 views
Arbitrary functions Synthetic Control Time series
I am aware that in synthetic control you create the synthetic control time series by effectively creating a weighted average of the matched/donor controls (plus extra covariates).
My question is, is ...
0
votes
0answers
30 views
Interpretation of adjusted $R^2$ in causal inference [duplicate]
If one is only interested in the causal effect of a feature on the outcome $Y \sim F + C$ (here $F$ is the treatment and $C$ the control variables), what is the interpretation of adjusted $R^2$? ...
0
votes
0answers
4 views
Help on whole model specification and guide on procedures
Could you guide me what I should do in my research in terms of econometrics.
I want to emphasize the causality in energy savings by buildings and various factors (like climate etc.). I don't have a ...
0
votes
1answer
21 views
Causal Inference in Mortality Rates
I was wondering how does one study the average treatment affect in scenarios suchs as mortality rates.
For example: suppose we want to study the effect that a certain medicine has on the mortality ...
1
vote
0answers
23 views
Causal Inference in a employee churn context (difference-in-differences / Propensity score matching)
For my master thesis I'm trying to determine the causes of an employee leaving a company. Currently I'm trying to study the effect that giving a raise has on employee leaving a company or not. So my ...
1
vote
0answers
29 views
Are boosted machine learning methods robust against low probable feature combinations when predicting?
I would like to use machine learning methods in the potential outcome framework, that is, simulating outcome for all observations under different values of a specific predictor, while keeping all ...
1
vote
0answers
29 views
Ok to use PSM to create treatment groups and then plug into CausalImpact? [closed]
Is it ok to use propensity score matching to create treatment and control groups and then plug these two time series into CausalImpact to estimate your treatment effect? I might want to do this, for ...
0
votes
0answers
10 views
Two-step residualization in fuzzy RD
I'm adopting a two-step residualization approach in fuzzy RD, should my first stage also include a polynomial of running variable (age)?
0
votes
0answers
36 views
Learning Causal Graph from data
I am quite new to the theory of causal graphs, but from what I understand they are DAG, like Bayesian Networks. Since we have structure learning methods for Bayesian Networks like score based ...
1
vote
0answers
28 views
Is causal relationship between two variables a theory?
I was asked if Y causes Z a theory, and I noticed that I have a gap in my knowledge on that. I know that theory could be causal, descriptive or predictive in its explanation. However, a theory is an ...
2
votes
1answer
29 views
Causal inference: is it possible to estimate the actual users who did X because of Y?
Suppose we have a dataset with many (~100) features and a binary outcome. I am interested in not only assessing whether any given feature is causally related to the outcome, but in actually being ...
2
votes
1answer
148 views
Causality: Structural Causal Model and DAG
I know that in general a structural causal model (SCM) can be written in terms of structural equations. And in a more qualitative but formal manner, we can rewrite a structural model in terms of DAG.
...
1
vote
1answer
17 views
Directly take the difference of the metric between 2 groups after the treatment or doing a difference-in-difference?
Assume we are doing a randomized experiment. The dependent variable is $Y$. Usually, we randomly put some of the subjects in the treatment group ($n$ subjects) and the rest to the control group ($m$ ...
2
votes
1answer
47 views
Expectation of potential outcomes formula
In Mostly Harmless Econometrics, the author uses the following identity to derive an estimator for the causal effect:
$$E \left[ \frac{Y_i D_i} {p(X_i)} \right] = E \left[Y_{1i} \right]$$
where:
$...
0
votes
0answers
17 views
Proving that average treatment effect for RDD with multiple cutoffs is a weighted average of the causal effect at each cutoff
Let $Xi$ denote the original running variable and $Ci$ be the cutoff that unit $i$ faces (e.g., the nearest cutoff).
For simplicity, I consider a case where the binary treatment is assigned if $Xi$ ...
3
votes
4answers
75 views
How to form groups before randomizing the treatment assignment?
Consider the following example. You want to assess the effect of a new class program on educational outcomes relative to the old program. You have recruited $N$ subjects for a randomized control trial ...
7
votes
2answers
279 views
Argument on Interactions in The Book of Why
There is a paragraph on interactions in The Book of Why (Pearl & Mackenzie, 2018), Chapter 9 (I cannot share the page number because I have the book in epub format), where the authors argue that:
...
0
votes
0answers
21 views
Finding causation in data with Conditional entropy
Conditional entropy is defined here. I was wondering about an algorithm to find causality between random variables. In particular, I want to calculate $H(X|Y)$ and $H(Y|X)$ and make a guess as to ...
2
votes
1answer
124 views
How to perform a what-if study using observational data?
A team fit a Random Forest model to a dataset $S=\{\mathbf{x}_i,y_i\}_{i=1}^N$, where $\mathbf{x}$ is a vector of continuous and categorical variables, and $y$ is a binary response. The model has a ...
0
votes
0answers
21 views
experimental design to test the effects of price discount
I have a panel dataset with hourly sales：(0:00~23:00) by products, after 16:00, the platform will give a discount to several products if they monitor a high inventory for these products.
I want to ...
0
votes
1answer
28 views
How to generate a model for the causal effects for a Panel dataset
I have a dataset such as hourly sales data：(0:00~23:00) by products, after 18:00, the platform will give a discount to several products to speed up the sales. If I want to know the causal effects of ...
10
votes
2answers
184 views
Does direction of causality between instrument and variable matter?
The standard scheme of instrumental variable in terms of causality (->) is:
Z -> X -> Y
Where Z is an instrument, X ...
0
votes
0answers
44 views
The impact of other continuous variable on the DID estimate
I have searched around but didn't have any luck. I am trying to estimate and then compare the DID estimates of the effect of a Treatment on Life-expectation across different Income levels, to know if ...
0
votes
0answers
68 views
Why do stabilized IPW weights give the same estimates and SEs as unstabilized weights?
In Cole & Hernán (2008), the authors mention that using stabilized weights can decrease the variance of the effect estimate. Regular inverse probability weights use the probability of being in the ...
0
votes
0answers
15 views
Causality in variance test, Cheung and Ng (1996)
I have been reading about the Cheung and Ng (1996) causality in variance test.
Would anyone know of any code or instructions on how I can code it up.
I have heard it is pretty simple, but I havent ...
0
votes
1answer
25 views
Are there any examples of multinomial or logistic regression as an outcome model using propensity score weighting?
I apologize if this is an inappropriate question, but does anyone have more recent texts on implementing some type of covariate balance weighting scheme (entropy balancing, IPTW, etc) where the ...
1
vote
1answer
96 views
Third Newton's law as a DAG [closed]
A we all know, Newton's third law is:
For every action, there is an equal and opposite reaction.
So, If A is pushed with force K by B, then it pushes back B with force k.
How can I represent this with ...
1
vote
1answer
49 views
Can causation be inferred when all possible covariates are included in a multiple regression?
Say we were interested in SAT scores for high school students as our dependent variable in a multiple regression. Now, assume we are God and can include literally all relevant covariates in the model (...
0
votes
0answers
22 views
When to use an interaction with dummy variables, and when to estimate separate regressions? [duplicate]
I am interested in exploring heterogeneous treatment effects by category. As a simple example, imagine that I'm trying to predict the impact of job training on income in New York, and I want to ...
0
votes
0answers
38 views
Exploring Causal Relationship in R
I need to explore if there are any causal relationships in my dataset, but the data has 30 points per year (for 30 different countries) and only 10 years worth of data. Accordingly, I can't use ...
-1
votes
1answer
36 views
sum rule in conditional probability
P and S are the common cause of c. If P(C=true| P,S ) is given , can I introduce S to P(C|P) as P(C= true|P= true)= P(C=true| P =true , S= true)* P(P=true ,S=true )+ P(C= true | P=true ,S =false )*P(P=...
0
votes
1answer
55 views
How to eliminate variable given conditional probabilities
$P$ and $S$ are the common cause of $c$. If $P(C=true| P,S )$ is given as the table below, and $P(S=true) =0.3$, $P(P=true) =0.9$ how can I eliminate $S$ and calculate $P(C=true | P=true )$ and $P(C= ...
0
votes
0answers
55 views
Dealing with Endogeneity in a Logit Regression when the Endogenous Regressors are Discrete
I would like to estimate a logit model in the presence of endogeneity. The dependent variable is binary (actually, it is non-binary with multiple ordinal categories, but from what I've read dealing ...
0
votes
0answers
34 views
Are feature importances from tree based models directly actionable for business?
If my response variable say is "has_repurchased" [0 or 1] and I have all customer level features. Can I rank the features in order of importance from the random forest model and report them as whats ...
