0
$\begingroup$

In an econometric work, I want to assess the causal effect of n variables on a binary character variable y, while I highly suspect that the relation between one of these regressors, say x (which is numerical), and y, is dependent on the value of x. Thus, I aim at using a non-parametric (local) logistic regression.

In order to avoid the curse of dimensionality, and because a classical logistic regression seems appropriate for the relation between the n-1 other variables and y, it would really help me to get answers to the following question:

Is their a way to do a classical logistic regression involving the n-1 first regressors, then to do the non parametric regression involving simply x, and then to link those two regressions to have the causal effect of each variable ?

In a linear model, I would run the first regression between y and the n-1 regressors, then run a non-parametric regression of the residual on x, but I do not think it is possible to do it in a logistic framework.

Thanks a lot,

PB

Improve this question
$\endgroup$

1 Answer 1

Reset to default
0
$\begingroup$

There are discussions / doubts about whether this approach is useful in general, see e.g. https://besjournals.onlinelibrary.wiley.com/doi/10.1046/j.1365-2656.2002.00618.x

Just from the technical side though (dealing with the logistic residuals), the quantile residuals calculated by the DHARMa package https://cran.r-project.org/web/packages/DHARMa/index.html should allow you to do what you describe.

Improve this answer
$\endgroup$
1
  • $\begingroup$ Thank you very much for your answer ! $\endgroup$ Commented Feb 25, 2021 at 17:49

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.