# Posts Tagged ‘ probit ’

## Optim, you’re doing it wrong?

May 28, 2012
By

Call me uncouth, but I like my TV loud, my beer cold and my optimization functions as simple as possible. Therefore, what I write in this blog post is very much from a layman’s perspective, and I am happy to be corrected on any fundamental errors. I have recently become interested in writing my own

## Probit/Logit Marginal Effects in R

April 23, 2012
By

The common approach to estimating a binary dependent variable regression model is to use either the logit or probit model. Both are forms of generalized linear models (GLMs), which can be seen as modified linear regressions that allow the dependent variable to originate from non-normal distributions. The coefficients in a linear regression model are marginal

## Stata-like Marginal Effects for Logit and Probit Models in R [2]

May 18, 2011
By

My thanks to those who emailed comments and suggestions for my ‘mfx’ function, I’m happy that I could fill a void for some people. I also received a request/suggestion from Tony Cookson, along with a helpful fix for a bug in the code, to include an option that would allow the user to specify values

## Stata-like Marginal Effects for Logit and Probit Models in R

May 17, 2011
By
$Stata-like Marginal Effects for Logit and Probit Models in R$

Although this blog’s primary focus is time series, one feature I missed from Stata was the simple marginal effects command, ‘mfx compute’, for cross-sectional work, and I could not find an adequate replacement in R. To bridge this gap, I’ve written a (rather messy) R function to produce marginal effects readout for logit and probit

## Multivariate probit regression using (direct) maximum likelihood estimators

May 11, 2011
By

Consider a random pair of binary responses, i.e. with taking values 1 or 2. Assume that probability can be function of some covariates . The Gaussian vector latent structure A standard model is based a latent Gaussian structure, i.e. there exi...