help reganat-------------------------------------------------------------------------------

Title

reganat-- Graphical inspection of linear multivariate models

Syntax

reganatdepvarvarlist[if] [in] [,options]

optionsDescription ------------------------------------------------------------------------- Optionsdis(varlist)graphs only the variables invarlistand omits the restbilineplots a regression line for the bivariate linear modelbiscatplots a scatterplot for the bivariate linear modelregdisplays the results for the estimation of the multivariate modelnolegendprevents the legend to be displayednocovlistprevents the list of covariates to be displayedscheme(scheme)specifies the graphical scheme -------------------------------------------------------------------------byis not allowed.

Description

reganatis a graphical tool for inspecting the effect of a covariate on a dependent variable in the context of multivariate OLS estimation. The name is an acronym for the expressionregression anatomy, a result in OLS' algebra originally due to Frisch and Waugh (1933) and recently revived by Angrist and Pischke'sMostly Harmless Econometrics(2009). In a bivariate regression modelY = bx1 + g, the graphical inspection of the scatterplot provides useful information on the relation between the independent variablex1and the dependent variableY, but can be highly misleading when the underlyingrealmodel is multivariate of the typeY =X'B + ewhereX'includes alsox1. In general, the OLS multivariate estimator is not equivalent to an OLS estimator obtained using a separate regression on each independent variable since correlation among independent variables must be accounted for. Angrist and Pischke (2009) show that in a multivariable model, the regression parameter for a given regressor isthe bivariate slope coefficient for the correspondingregressor after partialling out all other covariates. Accordingly, this command displays a table of scatterplots, with the dependent variable plotted against the independent variable net of any linear correlation with the other independent variables. This combined graph can be helpful when inspecting the data for outliers, nonlinearities, and other modelling issues.

Options+---------+ ----+ Options +----------------------------------------------------------

dis(varlist)specifies that the sense only the covariates included invarlistwill be plotted in the graph. Nonetheless, the other variables still work as controls.

bilinespecifies that on the graph will also be displayed the fitted regression line for the linear modelY = x1 + g, where x1 is the variable of interest.

biscatspecifies that on the graph will also be displayed the scatterplot for(Y,x1), where x1 is the variable of interest.

regdisplays the numerical results for the estimation of the multivariate modelY = X'B + e.

nolegendprevents the legend to be displayed.

nocovlistprevents the list of covariates to be displayed.

scheme(graphical scheme)can be used to specify the graphical scheme to be applied to the composite graph. By default, the command uses thesjscheme.

ExamplesSetup

. sysuse auto, clearObtain a combined graph of the effect of several regressors

. reganat price length weight headroom mpgObtain a combined graph of the effect of a subset of the regressors, along with scatterplots and fitted line for the univariate models

. reganat price length weight headroom mpg, dis(weight length) biline

ReferencesAngrist, J. and Pischke, J. S. (2009)

Mostly Harmless Econometrics. Princeton University Press.Frisch, R. and Waugh, F. W. (1933) Partial Time Regression as Compared with Individual Trends.

Econometrica1: 387-401.

AcknowledgmentThe author wishes to express his deep gratitude to Joshua Angrist for a number of insights and advices which have greatly enhanced the scope and the reliability of this command.

AuthorValerio Filoso, Department of Economics, University of Naples "Federico II", Italy.