All Questions
Tagged with confounding causality
92 questions
2
votes
2
answers
87
views
Definition of selection bias vs confounding bias
I've been learning about causal inference, having read Pearl's Primer and Parts I and II of "What If?".
I was under the impression that the definition of "There is confounding" was
...
- 225
2
votes
1
answer
65
views
Understanding Ignorability and Confounding Variables
I am reading Data Analysis Using Regression and Hierarchical Models and am confused by the concept of ignorability. The description in the book seems to say different things.
Said another way,
we ...
- 133
0
votes
0
answers
23
views
Can I interpret the size of my regression coefficient when I only control for confounders and not non-confounding covariates?
I am very confused at the moment, for my bachelor's thesis I am performing a Panel Individual Fixed Effects analysis and have only controlled for confounders.
I was under the assumption that when I ...
1
vote
1
answer
122
views
When does Dose Response Function estimation work better than simple regression?
I have been recently asked what is the difference between a Dose Response Function (DRF) estimation (as the one proposed in this link and this paper) and a statistical regression method. I therefore ...
- 460
0
votes
0
answers
20
views
IPW recalculating with deterministic treatment
Let's assume I want to calculate the ATE for a certain deterministic treatment, such as surgery (i.e., one either had it or not), and I'm interested in a per-protocol analysis.
Note that those who had ...
- 63
1
vote
1
answer
34
views
Can I simulate an unobserved confounder to test sensitivity?
I have a linear regression model Y ~ b0 + b1*X, with X and Y as continously measured variables (say age -> income).
Assume I know from theory that there is an unobservable variable Z that is ...
- 11
6
votes
2
answers
484
views
Why does controlling for a collider open a path, while controlling for a confounder closes a path, if there are relations to third variable for both?
Collider bias occurs when there is no association between X and Y but when a third variable which is caused by both X and Y is controlled for, this "opens a path" between X and Y and leads ...
- 1,252
7
votes
3
answers
657
views
Should You Specify a Curvilinear/Non-Linear Effect If You Suspect It is Spurious?
Consider the following (simplified) example of a project I am working on:
I assume that $X$ has a linear effect on $Y$. However, after plotting the relationship on a scatter plot, it looks like the ...
- 825
1
vote
0
answers
204
views
Are we allowed to use classifiers to estimate the outcome in the first stage of Double Machine Learning when the outcome is binary?
It is clear to me how to proceed when the outcome is continuous, since the EconML and all other references I checked work with this type of examples (continuous outcome case). We simply apply a ...
- 11
2
votes
0
answers
33
views
Reporting effects of interest with and without a confounder: Is it reasonable?
Suppose that we have the following situation. With a predictor $X$ and a response variable $Y$, plus a confounder $C$, we consider two models: one is
$Y \sim N(\alpha_0+\beta_0 X, ~\sigma_0^2)$
and ...
- 2,759
3
votes
1
answer
173
views
In regression, should we adjust for variables only associated with the independent or dependent variable?
I have recently been reading more about causal inference so am trying to conceptually think about model specification in more detail. From reading (e.g. this paper), we adjust for confounders which, ...
- 133
1
vote
1
answer
52
views
Do confounder variables need to have a causal effect on treatment AND response variables?
Let's look at an example:
We want to know if our new jogging shirt reduces the amount of sweat produced by runners. Ten factory employees in Bangkok, Thailand are recruited to try out the prototype ...
1
vote
1
answer
156
views
In causal inference, can you control for confounders by matching the treatment and control group based on the time series of the outcome variable?
Suppose that Walmart has 100 stores. It has a coupon for cereal, and it wants to know if the coupon increases cereal sales by a significant amount.
Walmart puts the coupon on the cereal shelf in 10 ...
- 359
6
votes
1
answer
62
views
How to adjust for the confounder of a confounder and how to call the confounder of a confounder within treatment effect estimation?
How do we adjust for the confounder of a confounder in order to compute unbiased estimates of the treatment effect of $A$ on $D$? See the causal graph (DAG) below:
What do we call the confounder $C$ (...
- 414
1
vote
0
answers
30
views
Adjusting for mismeasured confounder
Suppose I would like to determine the causal effect of $X$ on $Y$, where the relationship is confounded by $U$, but I measure $W$, which is a mismeasured proxy of $U$, as in this paper. That is,
$$U \...
- 505
4
votes
2
answers
1k
views
Possibly unmeasured confounder and other highly correlated controls
I am developing two models using observational data in which I have a binary outcome, a binary treatment and a series of confounders which I control for. The only difference is that the first model ...
- 317
0
votes
1
answer
106
views
Adjusting for confounding with a negative control outcome
When studying the effect of an exposure $T$ on an outcome $Y$, we can control for measured confounders $X$, but the treatment effect can still be biased by residual confounding due to unmeasured ...
- 505
3
votes
1
answer
152
views
Why does unmeasured mediator–outcome confounding possibly remain even in randomized control trial
I have heard the following but I cannot be totally convinced. The DAG is shown below.
$Q:$ Why does unmeasured mediator–outcome confounding possibly remain even in randomized control trial? Should ...
- 1,465
3
votes
0
answers
49
views
Does selecting confounder variables for a model with multiple correlation tests risk biasing results (similar to forward selection)?
My team is conducting a counterfactual difference-in-differences (DiD) healthcare analysis to estimate the benefits of home nursing visits compared with a control group.
We've "pre-selected" ...
- 6,286
1
vote
1
answer
185
views
Does an unobserved mediator causes endogeneity?
Suppose I'm modeling the probability to apply for a bank loan as a function of gender.
I have then the following DAG:
Wikipedia lists 3 causes of endogeneity
There is measurement error. Suppose I ...
- 2,433
5
votes
1
answer
200
views
Causal inference, stratification to mitigate confounders in continuous variables?
Handling confounders in continuous variables
In Statistical Rethinking, the author shows that in different situations, a confounder (fork, pipe, collider, descendent) will induce spurious correlations....
- 3,520
7
votes
1
answer
2k
views
how to perform double ML with binary data (either in the treatment or in the outcome)?
I have grown interested in double Machine Learning (ML) for causal inference because it answers an intuitive question: if the relationship between a variable $X$ (the treatment) and a variable $Y$ (...
- 475
5
votes
1
answer
106
views
VanderWeele bias calculation
In his book "Explanation in Causal Inference Methods for Mediation and Interaction", Tyler VanderWeele gives the following formula for bias (also Cinelli & Hazlet).
Let $\beta'$ be the ...
- 911
2
votes
1
answer
261
views
Causal discovery for pairwise independent joint dependent variables
Consider the standard example for variables that are pairwise independent but joint dependent.
$$
(x,y,z)=
\begin{cases}
(0,0,0) & \text{probability 1/4} \\
(1,1,0) & \text{probability 1/4} \\
...
2
votes
1
answer
155
views
What will happen if we condition on the parent of a confounder?
Say we have a confounder $U$ between $X$ and $Y$, i.e. graphically, $U\rightarrow X$ and $U \rightarrow Y$, $X$ and $Y$ are not directly connected.
According to the d-seperation rule, I know ...
- 593
0
votes
0
answers
24
views
Statistical methods for overcoming residual confounding with administrative health data
I have extensive administrative health data (general patient demographics + medical information, place of residence, many study years, few outcome measures) that have limited variables for effectively ...
- 2,277
0
votes
1
answer
503
views
IPW/MSM: Adjusting for confounding under time-varying treatment and confounding
Problem
I am playing around with what I thought would be a simple, straightforward simulation of recovering a lack of treatment effect under confounding in a longitudinal setting. I am trying to ...
- 15
1
vote
1
answer
821
views
Why do we call an assignment mechanism unconfounded assignment mechanism?
Consider some experiment with $n$ units along outcomes $Y=(Y_1,\dots, Y_n)$, covariates $X=(X_1,\dots, X_n)$ and treatment vector $W=(W_1,\dots, W_n)$ where $Y_i=(Y_i^1,Y_i^0)$ for treatment and ...
- 1,465
1
vote
0
answers
214
views
Does the regression model have to be the same as DAG?
I have built a DAG based on previous literature where a variable was used as a mediator in a paper and as a confounder in another paper. So, I used a bidirectional arrow between the exposure and the ...
- 11
2
votes
2
answers
97
views
How to derive the size of a simple bias in a mediation setting?
Consider the following DAG which shows the direct and indirect effect of $U$ on $Y$.
The total effect of $U \rightarrow Y$ is simply $(2\times4) + 3 = 11$.
I am looking for the derivation of the ...
- 911
2
votes
1
answer
122
views
Can I use a continuous variable as a confounder?
I am validating causality for a model that looks like below. Risk and age are continuous variables, intervention is binary. Age is numeric with intervals of a year, it is also a confounder. I was ...
- 579
2
votes
0
answers
467
views
What can I do to get better overlap in propensity score distributions?
I would like to verify the positivity assumption to identify causal effects from observational data. My exposure prevalence is about 6%. When I included several potential confounders in my exposure ...
- 21
2
votes
1
answer
241
views
Matching is not recovering the true effect in simulated data
I am trying to recover the true (simulated) effect of a treatment Z on an outcome Y, which is set to ATE = 5 (the csv file for the data is located here: https://www.dropbox.com/s/92obn9hsu3tqy92/...
- 134
2
votes
1
answer
118
views
RCCP (Reichenbach's Common Cause Principle) and confounder definition
RCCP states that if X and Y are statistically dependent, then there exists Z causally influencing
both.
I've heard a variation of RCCP that states that if X and Y are statistically dependent, one of ...
- 45
2
votes
1
answer
2k
views
What are "post-treatment confounders" and "pre-treatment confounders"?
I am reading the vignette for the R mediation package.
At page 4 it makes the sequential ignorability assumption, and says that
Equation 6 requires that the mediator is also ignorable given the ...
- 2,433
13
votes
1
answer
2k
views
How to interpret Pearl's do notation?
I'm going through the Dragonnet paper (slides available here), and the authors use Pearl's do notation to make this claim:
How can I interpret the do notation? Is the author claiming that the average ...
- 233
3
votes
0
answers
88
views
All-subsets regression and parameter shift to estimate or identify omitted variable biases?
I have multiple ($12$) predictors ($X$) for an outcome (spending) where it's likely/possible that:
Some predictors are correlated
Some predictors could (partially) mediate the effect of others
There ...
- 435
1
vote
0
answers
272
views
How to choose covariate adjustment method for effect of treatment estimation? [closed]
When trying to estimate average treatment effect (ATE) how does one choose between the different methods available to adjust for confounders?
For example, how to choose between propensity score block ...
- 221
3
votes
1
answer
3k
views
DAG: interpretation difference of TOTAL and DIRECT effect in terms of adjusting
Could one explain in simple words how examining the total and direct effect differ in terms of adjusting? How to interpret the findings of these two approaches?
DAG
Minimal sufficient adjustment sets ...
- 2,277
5
votes
1
answer
785
views
Term for variable which is both confounder and mediator
Is there a term for a variable that is both a mediator and a confounder at the same time? By this a mean a variable that both influences the exposure and the outcome but there is also an influence ...
- 228
13
votes
2
answers
1k
views
Do-Calculus for Causal Diagram 7.5 from "The Book of Why" (napkin problem)
In "The Book of Why" the below causal diagram is described as the "simplest model" where estimation of the causal effect goes beyond front and back-door adjustment and thus ...
- 584
7
votes
2
answers
1k
views
An intuitive explanation of the instrumental variable
This is something that I had dealt with in my MSc Economics many years ago, passed the exams with flying colours, yet when I thought about it in more depth today, I was somewhat puzzled. This could ...
- 1,226
4
votes
1
answer
850
views
Simulating potential outcomes with a binary outcome
I want to create some simple simulations of potential outcomes to explore issues of confounding. I start with a binary confounder X and a binary treatment A. When my outcome is continuous, I can ...
- 83
1
vote
1
answer
18
views
Can you determine if a correlation exists independent from outliers without conducting an experiment?
Let's say A correlates with B. But, A is correlated with C, D, E, and F which also correlate with B.
Could you determine if A's correlation with B is solely due to the fact that C, D, E, and F ...
7
votes
1
answer
608
views
Causal Inference After Feature Selection
I am interested in this forum's thoughts concerning the use of LASSO for feature selection in a high dimensional dataset and subsequent OLS regression to adjust for confounding on the most frequently ...
- 81
10
votes
1
answer
4k
views
Difference Omitted Variable Bias and Confounding?
Is there a difference between omitted variable bias and confounding bias in linear models?
To my knowledge, when investigating the causal effect of $X$ on $Y$, a confounder is a variable $Z$ that is ...
- 317
3
votes
0
answers
53
views
Failing to fully control for a variable
Lets assume we want to perform a 'reduced-form' causal analysis to evaluate the impact of a program on the dependent variable of interest. (However the question is more universal).
Lets further assume,...
- 1,854
7
votes
1
answer
526
views
DAG: no back-door paths but background information shows a need for adjusting
I am interested in the effect of town of residence on income.
Though the DAG has many arrows, it's interpretation is actually very simple:
I have 6 covariates (Cov1-6), all causing mediation ...
- 2,277
5
votes
2
answers
1k
views
Using random effects to adjust for cluster-level confounding?
There is a usage of random intercepts to adjust for unobserved cluster-level confounding, as for example argued here:
Are random effects confounding variables?
How do random effects adjust for ...
- 1,149
8
votes
1
answer
946
views
How do we handle a confounder which is collinear with the exposure?
X - treatment variable
Y - outcome variable
Z - confounder
DAG:
Model:
y ~ x + z
Question
If x and z strongly correlate with each other, then multicollinearity ...
- 2,277