As I guess we all know, $\beta$ is the standardized version of $b$, which is the estimated slope parameter. However, how do I calculate it? I assume I have to divide $b$ by some kind of standard deviation. Any useful reference? I know there are other formulas, but I want to know how to get it from $b$.

  • 2
    $\begingroup$ I've removed the tag "beta distribution", which doesn't apply at all. $\endgroup$ – Nick Cox Nov 17 '15 at 19:51
  • 4
    $\begingroup$ Just a note to say that "we all know" varies by discipline; for example, $\beta$ is sometimes used to refer to what you term $b$, whether standardized or not. $\endgroup$ – Alexis Nov 17 '15 at 19:51

Actually, it is much simpler. To get $\beta$ instead of $b$, all you have to do is standardize your $x$ and your $y$ are run a regression of the standardized $y$ on the standardized $x$. Consider a simple bivariate model: $$y=x\beta+u$$

Here, $\beta$ measures the marginal effect of x on the conditional mean of y . Now, let us standardize both $x$ and $y$ as :$$\tilde{y}=\frac{y-\mu_{y}}{\sigma_{y}}$$ where $\mu_{y}$ denotes the mean of y and $\sigma_{y}$ denotes the standard deviation of $y$. Do a similar transformation for $x$. Then run the regression:$$\tilde{y}=\tilde{x}b+u$$ and you will estimate the slope paramter in standard deviation units.

If you want to use the OLS estimate directly, just multiply $\hat{\beta}$ with $\frac{\sigma_{x}}{\sigma_{y}}$.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.