Causality Help!?!?
I have a regression model with 1 Y variable and 2 X variables (R sq = .66). The Y variable, and one of the X variables are likely interdependent (based on deep domain knowledge of 20+ years).
When I swap the Y variable and the X variable (making the Y independent and X dependent )…. the resulting new regression model shows the new X variable (formerly Y) with a value of .49. This is pretty strong, but significantly weaker than the former X variable (now Y variable) which previously had a p value of .10.
What conclusions can I make? Can I say that the original X variable has ~twice as much effect on the trend (.90 / .51) as the new (swapped) X variable? Even better, can I say that ~2/3rds [.90/(.90+.51)] of trend is a function of the original X variable (this is actually what I believe based on domain knowledge)?
(Note - I know only enough to be dangerous about regression)
Answer
- The questioner was satisfied with and accepted the answer, or
- The answer was evaluated as being 100% correct by the judge.
- answered
- 661 views
- $10.00
Related Questions
- Classification problem using one or more Simple Linear Perceptron
- Stats project
- Intro to stats symmetry vs skew and missing data values questions
- ANCOVA: R Squared and Partial Eta Squared
- What statistical test should I use to compare 2 multiple linear regression models
- Prove that $\lim_{n\rightarrow \infty} \int_{[0,1]^n}\frac{|x|}{\sqrt{n}}=\frac{1}{\sqrt{3}}$
- Multivariate Student-t Posterior Predictive - Detailed Derivation
- Hypothesis Testing, Probabilities