help retroclassical
-------------------------------------------------------------------------------

Title

{bf: retroclassical -- estimates neoclassical education transitions model witho > ut selection}

Syntax retroclassical Y1 "TC1" "TV1Names" "Exclusions" Y1Independent Y2 "TC1" "TV2Names" Y2Independent [cluster] ; Description

retroclassical estimates a bivariate probit model with coefficients re-scaled t > o facilitate cross-equation testing for coefficient equality. This model sets th > e cross-equation error covariance equal to zero. It is termed retroclassical for > it is midway between a variant of the neoclassical education transitions framework pr > oposed in Lucas, Fucella, and Berends (2011) and the classical education transitions approach. Note that the first transition appears in the second equation in the > output. Yet, in specifying the model the temporally earlier transition is alwa > ys regarded as the first transition. The estimates are the re-scaled coefficients > and the appropriate standard errors. Any subsequent statistical tests will use thi > s second set of results. A related command, neoclassical, estimates a neoclassical education transitions model while estimating a possibly non-zero cross-equation error variance. neoclassical provides one part of one way to implement the neoclassical education transitions framework, calibrating the coefficients across transitions, while accounting for selection.

Y1 is the outcome variable for the first transition--this is the selection equation for the second transition

TC1 contains the list of time constant variables; these should be common across the first and second equations. The set of variables should be in quotes

TV1Names contains the names of the variables that have specific values for the first transition (and possibly different values for the second transition). The set of variables should be in quotes

Exclusions are those variables in the selection equation that are not in the equation for the second transition. An explicit category is made for these variables to reinforce the necessity of having such variables to identify the selection process. The set of variables should be in quotations, even if it is only 1 variable

Y1Independent is the total number of independent variables in the first equation

Y2 is the outcome variable for the second transition--this is the last of the two transitions

TC1 contains the list of variables that are common across the first and second equations. The set of variables should be in quotes and should have the exact same order as when the variables appeared earlier in the command

TV2Names contains the variables that have specific values for the second transition (and possibly different values for the first transition). These variables should be in quotes

Y2Independent is the total number of independent variables in the second transition equation

deff MUST be supplied. It is the value of the design effect. If no design effect is needed, input 1.0.

cluster is a variable you may use to obtain Huber-White standard errors

Also see

Lucas, Samuel R., Phillip N. Fucella, and Mark Berends. 2011. "Neo-Classical Education Transitions: A Corrected Tale for Three