Questions tagged [causality]
The relationship between cause and effect.
1,961 questions
0
votes
0
answers
13
views
Is an explicit "treatment" variable a necessary condition for instrumental variable analysis?
I'm trying to model the causal impact of our marketing efforts on our ads business, and I'm considering an Instrumental Variable (IV) framework. I'd appreciate a sanity check on my approach and any ...
- 65
0
votes
0
answers
27
views
+50
Conceptual questions around marketing mix modeling (MMM) in the presence of omitted variables and missing not at random (MNAR) data
I need your help.
Imagine a company is currently evaluating a vendor-provided MMM (Marketing Mix Modeling) solution that can be further calibrated (not used for MMM modeling validation) using ...
- 31
0
votes
0
answers
24
views
Sampling counterfactual posterior to mitigate error autocorrelation in event studies
I have question regarding event studies (pre-event data is observed, an event occurs at $t=e$, then following the treatment is assumed to be in-effect.)
There are multiple approaches to event study ...
- 3,948
2
votes
0
answers
51
views
Variance of the average treatment effect using the unit-level variances of the potential outcomes
I'm reading Chapter 19 of Imbens and Rubin (2015), which is on the estimation of variance for estimators of treatment effects. They discuss using the variance of each sample/unit's potential outcomes ...
- 105
10
votes
3
answers
542
views
Why is median treatment effect not estimable?
The recent paper Addanki and Bhandari (2024) claims the median treatment effect (MTE) cannot be estimated for treatment and control groups in an observational study.
The authors illustrate with an ...
- 6,604
5
votes
1
answer
383
views
Which DAG is implied by the (usual) linear regression assumptions?
To avoid the inevitable confusion, let me briefly state the assumptions in their common form:
(Linear in parameters) In the population model, the dependent variable, $y$, is related to the ...
- 1,513
0
votes
0
answers
23
views
outcome-dependent selection bias and IPW
I'm having trouble understanding how inverse probability weighting (IPW) can be used to account for outcome-dependent selection when aiming to estimate the average causal effect of treatment. ...
- 1
17
votes
2
answers
4k
views
Is the logic of the original smoking study valid?
Out of some curiosity and exercise of my understanding of statistics, I've looked up the original study which statistically proved the causal relationship between smoking and lung cancer. Although ...
- 681
1
vote
1
answer
60
views
Estimating Marginal Structural Models with the {WeightIt} Package [closed]
I am used to using the {ipw} R package for estimating marginal structural models. However, I am having problems implementing equivalent models using the {WeightIt} R package.
To demonstrate my point ...
- 161
3
votes
0
answers
47
views
An incorrect causal MAG in a highly cited paper?
In this popular paper the causal DAG below is explored
and the author presents the following MAG as the causal MAG corresponding to the above causal DAG.
But.... there seems to be an inducing path ...
- 4,952
0
votes
0
answers
51
views
Is controlling for the mediator the same as controlling for the confounder?
I understand that it is a convention for statisticians to control for mediators to observe the direct effects of an exposure. However, is it appropriate to do so?
Let's look at an example below:
In ...
- 199
3
votes
1
answer
110
views
Showing that the OLS variance is the same as the variance of difference in means (average treatment effect)
Please bear with me as the preamble might be a bit long. I'm currently reading Imbens and Rubin's Causal Inference book, and unfortunately there's no freely avaiable online copy so below are some ...
- 105
3
votes
1
answer
123
views
Should we include a covariate in a DAG/causal model if it is a necessary cause for an outcome?
I would like to construct a causal model of exposure X on the outcome "fertility". Then there comes a covariate called "sexual intercourse (SI) frequency". This covariate can be a ...
- 199
12
votes
5
answers
2k
views
Is body mass index (BMI) actually the interaction term for weight × height^-2?
The formula for body mass index = weight × height^-2.
Suppose I want to study the effect of the weight of a parent (exposure) on the birth size of their child (outcome). We now understand that the ...
- 199
1
vote
0
answers
39
views
Should one include prognostic variables in outcome regression estimator for ATE?
I am interested in knowing whether adding prognostic variables will improve the asymptotic variance of outcome regression estimator for ATE. I have long heard that I should include prognostic ...
- 11
