Previous: Updating from Version 2.0

Up: Clarify: Software for Interpreting

Next: What's new in larify

What ${\mathfrak{C}}$larify Does

${\mathfrak{C}}$larify uses stochastic simulation techniques to help researchers interpret and present their statistical results. It uses whatever statistical model you have chosen and as such changes no statistical assumptions. As a first step, the program draws simulations of the main and ancillary parameters ( $\tilde\gamma$) from their asymptotic sampling distribution, in most cases a multivariate normal with mean equal to the vector of parameter estimates ( $\hat\gamma$) and variance equal to the variance-covariance matrix of estimates $\hat{\text{V}}(\hat\gamma)$.² Thus,

$$\tilde\gamma \sim$$   N$$\left(\hat\gamma,\hat{\text{V}}(\hat\gamma)\right)$$

By default the program draws $M=1000$ sets of simulated parameters, which should be sufficient for most applications.

Next, ${\mathfrak{C}}$larify converts the simulated parameters into substantively interesting quantities, such as predicted values, expected values, or first differences. To achieve this objective, the user need only choose real or hypothetical values for the explanatory variables (the $X$'s) and indicate which quantities should be calculated, conditional on those $X$'s. The program allows researchers to calculate virtually any quantity that would shed light on a particular problem, and provides a number of Stata procedures to do this easily.

${\mathfrak{C}}$larify 2.0 simulates quantities of interest for the most commonly used statistical models, including linear regression, binary logit, binary probit, ordered logit, ordered probit, multinomial logit, Poisson regression, negative binomial regression, weibull regression, seemingly unrelated regression equations, and compositional data.