*!version 9.1.0 2022-10-28 capture program drop rdplot program define rdplot, eclass syntax anything [if] [, c(real 0) p(integer 4) nbins(string) covs(string) covs_eval(string) covs_drop(string) binselect(string) scale(string) kernel(string) weights(string) h(string) support(string) masspoints(string) genvars hide ci(real 0) shade graph_options(string) nochecks *] marksample touse tokenize "`anything'" local y `1' local x `2' ******************** Set BW *************************** tokenize `h' local w : word count `h' if `w' == 1 { local h_r = `"`1'"' local h_l = `"`1'"' } if `w' == 2 { local h_l `"`1'"' local h_r `"`2'"' } if `w' >= 3 { di as error "{err}{cmd:h()} accepts at most two inputs" exit 125 } ******************** Set scale *************************** tokenize `scale' local w : word count `scale' if `w' == 1 { local scale_r = `"`1'"' local scale_l = `"`1'"' } if `w' == 2 { local scale_l `"`1'"' local scale_r `"`2'"' } if `w' >= 3 { di as error "{err}{cmd:scale()} accepts at most two inputs" exit 125 } ******************** Set nbins *************************** tokenize `nbins' local w : word count `nbins' if `w' == 1 { local nbins_r = `"`1'"' local nbins_l = `"`1'"' } if `w' == 2 { local nbins_l `"`1'"' local nbins_r `"`2'"' } if `w' >= 3 { di as error "{err}{cmd:nbins()} accepts at most two inputs" exit 125 } ******************** Set support *************************** tokenize `support' local w : word count `support' if `w' == 2 { local support_l = `"`1'"' local support_r = `"`2'"' } if (`w' != 2 & "`support'"!="") { di as error "{err}{cmd:support()} only accepts two inputs" exit 125 } ***************************************** preserve sort `x', stable qui keep if `touse' ************************************************************* **** DROP MISSINGS ****************************************** ************************************************************* qui drop if mi(`y') | mi(`x') if ("`covs'"~="") { qui ds `covs' local covs_list = r(varlist) local ncovs: word count `covs_list' foreach z in `covs_list' { qui drop if mi(`z') } } if ("`weights'"~="") { qui drop if mi(`weights') qui drop if `weights'<=0 } **** CHECK colinearity ****************************************** local covs_drop_coll = 0 if ("`covs_drop'"=="") local covs_drop = "pinv" if ("`covs'"~="") { if ("`covs_drop'"=="invsym") local covs_drop_coll = 1 if ("`covs_drop'"=="pinv") local covs_drop_coll = 2 if ("`covs_drop'"!="off") { qui _rmcoll `covs_list' local nocoll_controls_cat `r(varlist)' local nocoll_controls "" foreach myString of local nocoll_controls_cat { if ~strpos("`myString'", "o."){ if ~strpos("`myString'", "MYRUNVAR"){ local nocoll_controls "`nocoll_controls' `myString'" } } } local covs_new `nocoll_controls' qui ds `covs_new', alpha local covs_list_new = r(varlist) local ncovs_new: word count `covs_list_new' if (`ncovs_new'<`ncovs') { local ncovs = "`ncovs_new'" local covs_list = "`covs_list_new'" di as error "{err}Multicollinearity issue detected in {cmd:covs}. Redundant covariates were removed." } } } **** DEFAULTS *************************************** if ("`masspoints'"=="") local masspoints = "adjust" if ("`covs_eval'"=="") local covs_eval = "mean" ***************************************************** qui su `x' local N = r(N) local x_min = r(min) local x_max = r(max) if ("`support'"!="") { if (`support_l'<`x_min') { local x_min = `support_l' } if (`support_r'>`x_max') { local x_max = `support_r' } } local range_l = abs(`c'-`x_min') local range_r = abs(`x_max'-`c') qui su `y' if `x'<`c' local var_l = r(sd) local n_l = r(N) qui su `y' if `x'>=`c' local var_r = r(sd) local n_r = r(N) local n = `n_r' + `n_l' if ("`h_l'"=="" & "`h_r'"=="") { local h_l = `range_l' local h_r = `range_r' } if "`kernel'"=="" local kernel = "uni" qui count if `x'<`c' & `x'>=`c'-`h_l' local n_h_l = r(N) qui count if `x'>=`c' & `x'<=`c'+`h_r' local n_h_r = r(N) **************************** ERRORS if ("`scale_l'"=="" & "`scale_r'"=="") { local scale_r = 1 local scale_l = 1 } if ("`nbins_l'"=="" & "`nbins_r'"=="") { local nbins_r = 0 local nbins_l = 0 } if ("`binselect'"=="") { local binselect = "esmv" } if ("`nochecks'"=="") { if (`c'<=`x_min' | `c'>=`x_max'){ di as error "{err}{cmd:c()} should be set within the range of `x'" exit 125 } if ("`p'"<"0" | "`nbins_l'"<"0" | "`nbins_r'"<"0"){ di as error "{err}{cmd:p()} and {cmd:nbins()} should be a positive integers" exit 411 } if (`n'<20){ di as error "{err}Not enough observations to perform bin calculations" exit 2001 } } ******************************* ****** Start MATA ************* ******************************* mata{ n_l = `n_l' n_r = `n_r' p = `p' n = `n' c = `c' x_min = `x_min' x_max = `x_max' h_l = strtoreal("`h_l'"); h_r = strtoreal("`h_r'") nbins_l = strtoreal("`nbins_l'"); nbins_r = strtoreal("`nbins_r'") scale_l = strtoreal("`scale_l'"); scale_r = strtoreal("`scale_r'") y = st_data(.,("`y'"), 0); x = st_data(.,("`x'"), 0) ind_l = selectindex(x:=c) x_l = x[ind_l]; x_r = x[ind_r] y_l = y[ind_l]; y_r = y[ind_r] *** Mass points check ******************************************** masspoints_found = 0 if ("`masspoints'"=="check" | "`masspoints'"=="adjust") { X_uniq_l = sort(uniqrows(x_l),-1) X_uniq_r = uniqrows(x_r) M_l = length(X_uniq_l) M_r = length(X_uniq_r) M = M_l + M_r st_numscalar("M_l", M_l); st_numscalar("M_r", M_r) mass_l = 1-M_l/n_l mass_r = 1-M_r/n_r if (mass_l>=0.2 | mass_r>=0.2){ masspoints_found = 1 display("{err}Mass points detected in the running variable.") if ("`masspoints'"=="adjust") { if ("`binselect'"=="es") st_local("binselect","espr") if ("`binselect'"=="esmv") st_local("binselect","esmvpr") if ("`binselect'"=="qs") st_local("binselect","qspr") if ("`binselect'"=="qsmv") st_local("binselect","qsmvpr") } if ("`masspoints'"=="check") display("{err}Try using option {cmd:masspoints(adjust)}.") } } ****************************************************************************************** } mata{ *if ("`hide'"=="" | "`genvars'"!="" ){ ************************************************************ ************ Polynomial curve (order = p) ****************** ************************************************************ rp_l = J(n_l,(p+1),.); rp_r = J(n_r,(p+1),.) for (j=1; j<=(p+1); j++) { rp_l[.,j] = (x_l:-c):^(j-1) rp_r[.,j] = (x_r:-c):^(j-1) } wh_l = rdrobust_kweight(x_l, c, h_l+1e-8, "`kernel'") wh_r = rdrobust_kweight(x_r, c, h_r+1e-8, "`kernel'") if ("`weights'"~="") { fw = st_data(.,("`weights'"), 0) fw_l = fw[ind_l]; fw_r = fw[ind_r] wh_l = fw_l:*wh_l; wh_r = fw_r:*wh_r } invG_p_l = cholinv(cross(rp_l, wh_l, rp_l)) invG_p_r = cholinv(cross(rp_r, wh_r, rp_r)) if ("`covs'"=="") { gamma_p1_l = invG_p_l*cross(rp_l, wh_l, y_l) gamma_p1_r = invG_p_r*cross(rp_r, wh_r, y_r) } else { z = st_data(.,tokens("`covs'"), 0); dZ = cols(z) z_l = z[ind_l,]; z_r = z[ind_r,] d_l = y_l,z_l; d_r = y_r,z_r U_p_l = quadcross(rp_l:*wh_l,d_l); U_p_r = quadcross(rp_r:*wh_r,d_r) beta_p_l = invG_p_l*quadcross(rp_l:*wh_l,d_l) beta_p_r = invG_p_r*quadcross(rp_r:*wh_r,d_r) ZWD_p_l = quadcross(z_l,wh_l,d_l) ZWD_p_r = quadcross(z_r,wh_r,d_r) colsZ = (2)::(2+dZ-1) UiGU_p_l = quadcross(U_p_l[,colsZ],invG_p_l*U_p_l) UiGU_p_r = quadcross(U_p_r[,colsZ],invG_p_r*U_p_r) ZWZ_p_l = ZWD_p_l[,colsZ] - UiGU_p_l[,colsZ] ZWZ_p_r = ZWD_p_r[,colsZ] - UiGU_p_r[,colsZ] ZWY_p_l = ZWD_p_l[,1] - UiGU_p_l[,1] ZWY_p_r = ZWD_p_r[,1] - UiGU_p_r[,1] ZWZ_p = ZWZ_p_r + ZWZ_p_l ZWY_p = ZWY_p_r + ZWY_p_l if ("`covs_drop_coll'"=="0") gamma_p = cholinv(ZWZ_p)*ZWY_p if ("`covs_drop_coll'"=="1") gamma_p = invsym(ZWZ_p)*ZWY_p if ("`covs_drop_coll'"=="2") gamma_p = pinv(ZWZ_p)*ZWY_p s_Y = (1 \ -gamma_p[,1]) gamma_p1_l = (s_Y'*beta_p_l')' gamma_p1_r = (s_Y'*beta_p_r')' st_matrix("gamma_p", gamma_p) } st_matrix("gamma_p1_l", gamma_p1_l) st_matrix("gamma_p1_r", gamma_p1_r) *********** Preparte data for polynomial curve plot ***** nplot = 500 x_plot_l = rangen(c-h_l, c, nplot) x_plot_r = rangen(c, c+h_r, nplot) rplot_l = J(nplot,(p+1),.); rplot_r = J(nplot,(p+1),.) for (j=1; j<=(p+1); j++) { rplot_l[.,j] = (x_plot_l:-c):^(j-1) rplot_r[.,j] = (x_plot_r:-c):^(j-1) } gammaZ = 0 if ("`covs_eval'"=="mean" & "`covs'"!="") gammaZ = mean(z)*gamma_p y_plot_l = rplot_l*gamma_p1_l :+ gammaZ y_plot_r = rplot_r*gamma_p1_r :+ gammaZ *} ******************************************************* **** Optimal Bins (using polynomial order k) ********** ******************************************************* k = 4 rk_l = J(n_l,(k+1),.); rk_r = J(n_r,(k+1),.) for (j=1; j<=(k+1); j++) { rk_l[.,j] = x_l:^(j-1) rk_r[.,j] = x_r:^(j-1) } invG_k_l = cholinv(cross(rk_l,rk_l)) invG_k_r = cholinv(cross(rk_r,rk_r)) if (det(invG_k_l)==. | det(invG_k_r)==.) { k = 3 rk_l = J(n_l,(k+1),.) rk_r = J(n_r,(k+1),.) for (j=1; j<=(k+1); j++) { rk_l[.,j] = x_l:^(j-1) rk_r[.,j] = x_r:^(j-1) } invG_k_l = cholinv(cross(rk_l,rk_l)) invG_k_r = cholinv(cross(rk_r,rk_r)) } if (det(invG_k_l)==. | det(invG_k_r)==.) { k = 2 rk_l = J(n_l,(k+1),.) rk_r = J(n_r,(k+1),.) for (j=1; j<=(k+1); j++) { rk_l[.,j] = x_l:^(j-1) rk_r[.,j] = x_r:^(j-1) } invG_k_l = cholinv(cross(rk_l,rk_l)) invG_k_r = cholinv(cross(rk_r,rk_r)) } gamma_k1_l = invG_k_l*cross(rk_l,y_l) gamma_k1_r = invG_k_r*cross(rk_r,y_r) gamma_k2_l = invG_k_l*cross(rk_l,y_l:^2) gamma_k2_r = invG_k_r*cross(rk_r,y_r:^2) *** Bias w/sample mu0_k1_l = rk_l*gamma_k1_l mu0_k1_r = rk_r*gamma_k1_r mu0_k2_l = rk_l*gamma_k2_l mu0_k2_r = rk_r*gamma_k2_r drk_l = J(n_l,k,.) drk_r = J(n_r,k,.) for (j=1; j<=k; j++) { drk_l[.,j] = j*x_l:^(j-1) drk_r[.,j] = j*x_r:^(j-1) } dxi_l=(x_l[2::n_l]-x_l[1::(n_l-1)]) dxi_r=(x_r[2::n_r]-x_r[1::(n_r-1)]) dyi_l=(y_l[2::n_l]-y_l[1::(n_l-1)]) dyi_r=(y_r[2::n_r]-y_r[1::(n_r-1)]) x_bar_i_l = (x_l[2::n_l]+x_l[1::(n_l-1)])/2 x_bar_i_r = (x_r[2::n_r]+x_r[1::(n_r-1)])/2 drk_i_l = J(n_l-1,k,.); rk_i_l = J(n_l-1,(k+1),.) drk_i_r = J(n_r-1,k,.); rk_i_r = J(n_r-1,(k+1),.) for (j=1; j<=(k+1); j++) { rk_i_l[.,j] = x_bar_i_l:^(j-1) rk_i_r[.,j] = x_bar_i_r:^(j-1) } for (j=1; j<=k; j++) { drk_i_l[.,j] = j*x_bar_i_l:^(j-1) drk_i_r[.,j] = j*x_bar_i_r:^(j-1) } mu1_i_hat_l = drk_i_l*(gamma_k1_l[2::(k+1)]) mu1_i_hat_r = drk_i_r*(gamma_k1_r[2::(k+1)]) mu0_i_hat_l = rk_i_l*gamma_k1_l mu0_i_hat_r = rk_i_r*gamma_k1_r mu2_i_hat_l = rk_i_l*gamma_k2_l mu2_i_hat_r = rk_i_r*gamma_k2_r mu0_hat_l = rk_l*gamma_k1_l mu0_hat_r = rk_r*gamma_k1_r mu2_hat_l = rk_l*gamma_k2_l mu2_hat_r = rk_r*gamma_k2_r mu1_hat_l = drk_l*(gamma_k1_l[2::(k+1)]) mu1_hat_r = drk_r*(gamma_k1_r[2::(k+1)]) mu1_i_hat_l = drk_i_l*(gamma_k1_l[2::(k+1)]) mu1_i_hat_r = drk_i_r*(gamma_k1_r[2::(k+1)]) var_y_l = variance(y_l) var_y_r = variance(y_r) sigma2_hat_l_bar = mu2_i_hat_l - mu0_i_hat_l:^2 sigma2_hat_r_bar = mu2_i_hat_r - mu0_i_hat_r:^2 ind_s2_l = selectindex(sigma2_hat_l_bar:<0) ind_s2_r = selectindex(sigma2_hat_r_bar:<0) sigma2_hat_l_bar[ind_s2_l] = 0*ind_s2_l :+ var_y_l sigma2_hat_r_bar[ind_s2_r] = 0*ind_s2_r :+ var_y_r sigma2_hat_l = mu2_hat_l - mu0_hat_l:^2 sigma2_hat_r = mu2_hat_r - mu0_hat_r:^2 ind_s2_l = selectindex(sigma2_hat_l:<0) ind_s2_r = selectindex(sigma2_hat_r:<0) sigma2_hat_l[ind_s2_l] = 0*ind_s2_l :+ var_y_l sigma2_hat_r[ind_s2_r] = 0*ind_s2_r :+ var_y_r B_es_hat_dw = (((c-x_min)^2/(12*n))*sum(mu1_hat_l:^2),((x_max-c)^2/(12*n))*sum(mu1_hat_r:^2)) V_es_hat_dw = ((0.5/(c-x_min))*sum(dxi_l:*dyi_l:^2),(0.5/(x_max-c))*sum(dxi_r:*dyi_r:^2)) V_es_chk_dw = ((1/(c-x_min))*sum(dxi_l:*sigma2_hat_l_bar),(1/(x_max-c))*sum(dxi_r:*sigma2_hat_r_bar)) J_es_hat_dw = ceil((((2*B_es_hat_dw):/V_es_hat_dw)*n):^(1/3)) J_es_chk_dw = ceil((((2*B_es_hat_dw):/V_es_chk_dw)*n):^(1/3)) B_qs_hat_dw = ((n_l^2/(24*n))*sum(dxi_l:^2:*mu1_i_hat_l:^2), (n_r^2/(24*n))*sum(dxi_r:^2:*mu1_i_hat_r:^2)) V_qs_hat_dw = ((1/(2*n_l))*sum(dyi_l:^2),(1/(2*n_r))*sum(dyi_r:^2)) V_qs_chk_dw = ((1/n_l)*sum(sigma2_hat_l), (1/n_r)*sum(sigma2_hat_r)) J_qs_hat_dw = ceil((((2*B_qs_hat_dw):/V_qs_hat_dw)*n):^(1/3)) J_qs_chk_dw = ceil((((2*B_qs_hat_dw):/V_qs_chk_dw)*n):^(1/3)) J_es_hat_mv = (ceil((var_y_l/V_es_hat_dw[1])*(n/log(n)^2)), ceil((var_y_r/V_es_hat_dw[2])*(n/log(n)^2))) J_es_chk_mv = (ceil((var_y_l/V_es_chk_dw[1])*(n/log(n)^2)), ceil((var_y_r/V_es_chk_dw[2])*(n/log(n)^2))) J_qs_hat_mv = (ceil((var_y_l/V_qs_hat_dw[1])*(n/log(n)^2)), ceil((var_y_r/V_qs_hat_dw[2])*(n/log(n)^2))) J_qs_chk_mv = (ceil((var_y_l/V_qs_chk_dw[1])*(n/log(n)^2)), ceil((var_y_r/V_qs_chk_dw[2])*(n/log(n)^2))) if ("`binselect'"=="es" ) { J_star_l_orig = J_es_hat_dw[1] J_star_r_orig = J_es_hat_dw[2] } if ("`binselect'"=="esmv" | "`binselect'"=="") { J_star_l_orig = J_es_hat_mv[1] J_star_r_orig = J_es_hat_mv[2] } if ("`binselect'"=="espr" ) { J_star_l_orig = J_es_chk_dw[1] J_star_r_orig = J_es_chk_dw[2] } if ("`binselect'"=="esmvpr" ) { J_star_l_orig = J_es_chk_mv[1] J_star_r_orig = J_es_chk_mv[2] } if ("`binselect'"=="qs" ) { J_star_l_orig = J_qs_hat_dw[1] J_star_r_orig = J_qs_hat_dw[2] } if ("`binselect'"=="qsmv" ) { J_star_l_orig = J_qs_hat_mv[1] J_star_r_orig = J_qs_hat_mv[2] } if ("`binselect'"=="qspr" ) { J_star_l_orig = J_qs_chk_dw[1] J_star_r_orig = J_qs_chk_dw[2] } if ("`binselect'"=="qsmvpr" ) { J_star_l_orig = J_qs_chk_mv[1] J_star_r_orig = J_qs_chk_mv[2] } if (nbins_l!=0 & nbins_r!=0) { J_star_l_orig = nbins_l J_star_r_orig = nbins_r } if (`var_l'==0) { J_star_l = 1 J_star_l_orig = 1 display("{err}Warning: not enough variability in the outcome variable below the threshold") } if (`var_r'==0) { J_star_r = 1 J_star_r_orig = 1 display("{err}Warning: not enough variability in the outcome variable above the threshold") } J_star_l = round(`scale_l'*J_star_l_orig) J_star_r = round(`scale_r'*J_star_r_orig) st_numscalar("nbins_l", nbins_l) st_numscalar("nbins_r", nbins_r) st_numscalar("J_star_l", J_star_l) st_numscalar("J_star_r", J_star_r) st_numscalar("J_star_l_orig", J_star_l_orig) st_numscalar("J_star_r_orig", J_star_r_orig) st_matrix("J_es_hat_dw", J_es_hat_dw) st_matrix("J_qs_hat_dw", J_qs_hat_dw) st_matrix("J_es_chk_dw", J_es_chk_dw) st_matrix("J_qs_chk_dw", J_qs_chk_dw) st_matrix("J_es_hat_mv", J_es_hat_mv) st_matrix("J_qs_hat_mv", J_qs_hat_mv) st_matrix("J_es_chk_mv", J_es_chk_mv) st_matrix("J_qs_chk_mv", J_qs_chk_mv) } ******************************************************** **** Generate id and rdplot vars *********************** ******************************************************** local J_star_l = J_star_l local J_star_r = J_star_r if ("`binselect'"=="qs" | "`binselect'"=="qspr" | "`binselect'"=="qsmv" | "`binselect'"=="qsmvpr") { pctile binsL = `x' if `x'<`c', nq(`J_star_l') pctile binsR = `x' if `x'>=`c', nq(`J_star_r') } mata { x_min = `x_min' x_max = `x_max' if ("`binselect'"=="es" | "`binselect'"=="espr" | "`binselect'"=="esmv" | "`binselect'"=="esmvpr" | "`binselect'"=="") { binsL = rangen(x_min-1e-8,c , `J_star_l'+1) binsR = rangen(c ,x_max+1e-8, `J_star_r'+1) bins = binsL[1..length(binsL)-1]\binsR } if ("`binselect'"=="qs" | "`binselect'"=="qspr" | "`binselect'"=="qsmv" | "`binselect'"=="qsmvpr") { bins = (x_min-1e-8 \ st_data(.,"binsL",0) \ c \ st_data(.,"binsR",0) \ x_max+1e-8 ) binsL = (x_min-1e-8 \ st_data(.,"binsL",0) \ c ) binsR = (c \ st_data(.,"binsR",0) \ x_max+1e-8 ) } bin_x_l = rdrobust_groupid(x_l, binsL) bin_x_r = rdrobust_groupid(x_r, binsR) bin_x = bin_x_l:-(J_star_l+1) \ bin_x_r } ************************************************************************* **** covs_eval ********************************************************** ************************************************************************* if ("`covs_eval'"=="mean" & "`covs'"!="") { qui getmata bin_x , replace force tempvar yhatZ bin_x2 qui gen `bin_x2' = bin_x + `J_star_l' qui reg `y' `covs_list' i.`bin_x2' qui predict `yhatZ' } mata { if ("`covs_eval'"=="mean" & "`covs'"!="") { yhatZ = st_data(.,("`yhatZ'"), 0) y_l = yhatZ[ind_l]; y_r = yhatZ[ind_r] } d_l = x_l, y_l d_r = x_r, y_r rdbin_collapse_l = rdrobust_collapse(d_l, bin_x_l) rdbin_collapse_r = rdrobust_collapse(d_r, bin_x_r) rdplot_N_l = rdbin_collapse_l[,1] rdplot_N_r = rdbin_collapse_r[,1] rdplot_mean_x_l = rdbin_collapse_l[,2] rdplot_mean_x_r = rdbin_collapse_r[,2] rdplot_mean_y_l = rdbin_collapse_l[,3] rdplot_mean_y_r = rdbin_collapse_r[,3] rdplot_sd_y_l = sqrt(rdbin_collapse_l[,4]) rdplot_sd_y_r = sqrt(rdbin_collapse_r[,4]) rdplot_na_l = uniqrows(bin_x_l) rdplot_na_r = uniqrows(bin_x_r) rdplot_min_bin_l = binsL[1::J_star_l] rdplot_min_bin_r = binsR[1::J_star_r] rdplot_max_bin_l = binsL[2::(J_star_l+1)] rdplot_max_bin_r = binsR[2::(J_star_r+1)] rdplot_mean_bin_l = rowsum( (rdplot_min_bin_l , rdplot_max_bin_l))/2 rdplot_mean_bin_r = rowsum( (rdplot_min_bin_r , rdplot_max_bin_r))/2 rdplot_id = rdplot_na_l:-J_star_l:-1 \ rdplot_na_r rdplot_mean_x = rdplot_mean_x_l \ rdplot_mean_x_r rdplot_mean_y = rdplot_mean_y_l \ rdplot_mean_y_r rdplot_mean_bin = rdplot_mean_bin_l[rdplot_na_l] \ rdplot_mean_bin_r[rdplot_na_r] rdplot_N = rdplot_N_l \ rdplot_N_r rdplot_min_bin = rdplot_min_bin_l[rdplot_na_l] \ rdplot_min_bin_r[rdplot_na_r] rdplot_max_bin = rdplot_max_bin_l[rdplot_na_l] \ rdplot_max_bin_r[rdplot_na_r] rdplot_se_y = rdplot_sd_y_l:/sqrt(rdplot_N_l) \ rdplot_sd_y_r:/sqrt(rdplot_N_r) rdplot_length_l = rdplot_max_bin_l - rdplot_min_bin_l rdplot_length_r = rdplot_max_bin_r - rdplot_min_bin_r bin_avg_l = mean(rdplot_length_l) bin_avg_r = mean(rdplot_length_r) bin_med_l = rdrobust_median(rdplot_length_l) bin_med_r = rdrobust_median(rdplot_length_r) quant = -invt(rdplot_N, abs((1-(`ci'/100))/2)) rdplot_ci_l = rdplot_mean_y - quant:*rdplot_se_y rdplot_ci_r = rdplot_mean_y + quant:*rdplot_se_y st_numscalar("bin_avg_l", bin_avg_l) st_numscalar("bin_avg_r", bin_avg_r) st_numscalar("bin_med_l", bin_med_l) st_numscalar("bin_med_r", bin_med_r) } if ("`binselect'"=="es"){ local binselect_type="evenly spaced number of bins using spacings estimators." scalar J_star_l_IMSE = J_es_hat_dw[1,1] scalar J_star_r_IMSE = J_es_hat_dw[1,2] scalar J_star_l_MV = J_es_hat_mv[1,1] scalar J_star_r_MV = J_es_hat_mv[1,2] } if ("`binselect'"=="espr"){ local binselect_type="evenly spaced number of bins using polynomial regression." scalar J_star_l_IMSE = J_es_chk_dw[1,1] scalar J_star_r_IMSE = J_es_chk_dw[1,2] scalar J_star_l_MV = J_es_chk_mv[1,1] scalar J_star_r_MV = J_es_chk_mv[1,2] } if ("`binselect'"=="esmv" | "`binselect'"==""){ local binselect_type="evenly spaced mimicking variance number of bins using spacings estimators." scalar J_star_l_IMSE = J_es_hat_dw[1,1] scalar J_star_r_IMSE = J_es_hat_dw[1,2] scalar J_star_l_MV = J_es_hat_mv[1,1] scalar J_star_r_MV = J_es_hat_mv[1,2] } if ("`binselect'"=="esmvpr"){ local binselect_type="evenly spaced mimicking variance number of bins using polynomial regression." scalar J_star_l_IMSE = J_es_chk_dw[1,1] scalar J_star_r_IMSE = J_es_chk_dw[1,2] scalar J_star_l_MV = J_es_chk_mv[1,1] scalar J_star_r_MV = J_es_chk_mv[1,2] } if ("`binselect'"=="qs"){ local binselect_type="quantile spaced number of bins using spacings estimators." scalar J_star_l_IMSE = J_qs_hat_dw[1,1] scalar J_star_r_IMSE = J_qs_hat_dw[1,2] scalar J_star_l_MV = J_qs_hat_mv[1,1] scalar J_star_r_MV = J_qs_hat_mv[1,2] } if ("`binselect'"=="qspr"){ local binselect_type="quantile spaced number of bins using polynomial regression." scalar J_star_l_IMSE = J_qs_chk_dw[1,1] scalar J_star_r_IMSE = J_qs_chk_dw[1,2] scalar J_star_l_MV = J_qs_chk_mv[1,1] scalar J_star_r_MV = J_qs_chk_mv[1,2] } if ("`binselect'"=="qsmv"){ local binselect_type="quantile spaced mimicking variance quantile spaced using spacings estimators." scalar J_star_l_IMSE = J_qs_hat_dw[1,1] scalar J_star_r_IMSE = J_qs_hat_dw[1,2] scalar J_star_l_MV = J_qs_hat_mv[1,1] scalar J_star_r_MV = J_qs_hat_mv[1,2] } if ("`binselect'"=="qsmvpr"){ local binselect_type="quantile spaced mimicking variance number of bins using polynomial regression." scalar J_star_l_IMSE = J_qs_chk_dw[1,1] scalar J_star_r_IMSE = J_qs_chk_dw[1,2] scalar J_star_l_MV = J_qs_chk_mv[1,1] scalar J_star_r_MV = J_qs_chk_mv[1,2] } if (nbins_l!=0 | nbins_r!=0 ) local binselect_type= "RD plot with manually set number of bins." scalar scale_l = J_star_l / J_star_l_IMSE scalar scale_r = J_star_r / J_star_r_IMSE qui getmata x_plot_l x_plot_r y_plot_l y_plot_r rdplot_id rdplot_mean_bin rdplot_mean_x rdplot_mean_y rdplot_N rdplot_min_bin rdplot_max_bin rdplot_se_y rdplot_ci_l rdplot_ci_r, replace force ereturn clear ereturn scalar N_l = `n_l' ereturn scalar N_r = `n_r' ereturn scalar c = `c' ereturn scalar J_star_l = J_star_l ereturn scalar J_star_r = J_star_r ereturn matrix coef_l = gamma_p1_l ereturn matrix coef_r = gamma_p1_r if ("`covs'"!="") { ereturn matrix coef_covs = gamma_p } ereturn local binselect = "`binselect'" ****** polynomial equation for plots ****************** mat coef_l = e(coef_l) mat coef_r = e(coef_r) local eq_l = "y = coef_l[1, 1]*(x-$c)^0 " local eq_r = "y = coef_r[1, 1]*(x-$c)^0 " forvalues i = 1(1)`p' { local tt_l = "+ coef_l[`i'+1, 1]*(x-$c)^`i'" local tt_r = "+ coef_r[`i'+1, 1]*(x-$c)^`i'" local eq_l = " `eq_l' `tt_l'" local eq_r = " `eq_r' `tt_r'" } ereturn local eq_l = "`eq_l'" ereturn local eq_r = "`eq_r'" ****************************************************** if ("`kernel'"=="epanechnikov" | "`kernel'"=="epa") local kernel_type = "Epanechnikov" else if ("`kernel'"=="uniform" | "`kernel'"=="uni") local kernel_type = "Uniform" else local kernel_type = "Triangular" disp "" disp in smcl in yellow "RD Plot with " "`binselect_type'" disp "" disp in smcl in gr "{ralign 21: Cutoff c = `c'}" _col(22) " {c |} " _col(23) in gr "Left of " in yellow "c" _col(36) in gr "Right of " in yellow "c" _col(54) in gr "Number of obs = " in yellow %10.0f `n' disp in smcl in gr "{hline 22}{c +}{hline 22}" _col(54) in gr "Kernel = " in yellow "{ralign 10:`kernel_type'}" disp in smcl in gr "{ralign 21:Number of obs}" _col(22) " {c |} " _col(23) as result %9.0f `n_l' _col(37) %9.0f `n_r' disp in smcl in gr "{ralign 21:Eff. Number of obs}" _col(22) " {c |} " _col(23) as result %9.0f `n_h_l' _col(37) %9.0f `n_h_r' disp in smcl in gr "{ralign 21:Order poly. fit (p)}" _col(22) " {c |} " _col(23) as result %9.0f `p' _col(37) %9.0f `p' disp in smcl in gr "{ralign 21:BW poly. fit (h)}" _col(22) " {c |} " _col(23) as result %9.3f `h_l' _col(37) %9.3f `h_r' disp in smcl in gr "{ralign 21:Number of bins scale}" _col(22) " {c |} " _col(23) as result %9.3f `scale_l' _col(37) %9.3f `scale_r' disp "" disp "Outcome: `y'. Running variable: `x'." disp in smcl in gr "{hline 22}{c TT}{hline 22}" disp in smcl in gr _col(22) " {c |} " _col(23) in gr "Left of " in yellow "c" _col(36) in gr "Right of " in yellow "c" disp in smcl in gr "{hline 22}{c +}{hline 22}" disp in smcl in gr "{ralign 21:Bins selected}" _col(22) " {c |} " _col(23) as result %9.0f e(J_star_l) _col(37) %9.0f e(J_star_r) disp in smcl in gr "{ralign 21:Average bin length}" _col(22) " {c |} " _col(23) as result %9.3f scalar(bin_avg_l) _col(37) %9.3f scalar(bin_avg_r) disp in smcl in gr "{ralign 21:Median bin length}" _col(22) " {c |} " _col(23) as result %9.3f scalar(bin_med_l) _col(37) %9.3f scalar(bin_med_r) disp in smcl in gr "{hline 22}{c +}{hline 22}" disp in smcl in gr "{ralign 21:IMSE-optimal bins}" _col(22) " {c |} " _col(23) as result %9.0f J_star_l_IMSE _col(37) %9.0f J_star_r_IMSE disp in smcl in gr "{ralign 21:Mimicking Var. bins}" _col(22) " {c |} " _col(23) as result %9.0f J_star_l_MV _col(37) %9.0f J_star_r_MV disp in smcl in gr "{hline 22}{c +}{hline 22}" disp in smcl in gr "{lalign 1:Rel. to IMSE-optimal:}" _col(22) " {c |} " disp in smcl in gr "{ralign 21:Implied scale}" _col(22) " {c |} " _col(23) as result %9.3f scale_l _col(37) %9.3f scale_r disp in smcl in gr "{ralign 21:WIMSE var. weight}" _col(22) " {c |} " _col(23) as result %9.3f 1/(1+scale_l^3) _col(37) %9.3f 1/(1+scale_r^3) disp in smcl in gr "{ralign 21:WIMSE bias weight}" _col(22) " {c |} " _col(23) as result %9.3f scale_l^3/(1+scale_l^3) _col(37) %9.3f scale_r^3/(1+scale_r^3) disp in smcl in gr "{hline 22}{c BT}{hline 22}" disp "" if ("`covs'"!="") disp "Covariate-adjusted estimates. Additional covariates included: `ncovs'" if (`covs_drop_coll'==1) di as error "{err}Variables dropped due to multicollinearity." if ("`hide'"==""){ if (`"`graph_options'"'=="" ) local graph_options = `"title("Regression function fit", color(gs0)) "' if (`ci'==0) { twoway (scatter rdplot_mean_y rdplot_mean_bin, sort msize(small) mcolor(gs10)) /// (line y_plot_l x_plot_l, lcolor(black) sort lwidth(medthin) lpattern(solid) ) /// (line y_plot_r x_plot_r, lcolor(black) sort lwidth(medthin) lpattern(solid) ), /// xline(`c', lcolor(black) lwidth(medthin)) xscale(r(`x_min' `x_max')) legend(cols(2) order(1 "Sample average within bin" 2 "Polynomial fit of order `p'" )) `graph_options' } else { if ("`shade'"==""){ twoway (rcap rdplot_ci_l rdplot_ci_r rdplot_mean_bin, color(gs11)) /// (scatter rdplot_mean_y rdplot_mean_bin, sort msize(small) mcolor(gs10)) /// (line y_plot_l x_plot_l, lcolor(black) sort lwidth(medthin) lpattern(solid)) /// (line y_plot_r x_plot_r, lcolor(black) sort lwidth(medthin) lpattern(solid)), /// xline(`c', lcolor(black) lwidth(medthin)) xscale(r(`x_min' `x_max')) legend(cols(2) order(2 "Sample average within bin" 3 "Polynomial fit of order `p'" )) `graph_options' } else { twoway (rarea rdplot_ci_l rdplot_ci_r rdplot_mean_bin if rdplot_id<0, sort color(gs11)) /// (rarea rdplot_ci_l rdplot_ci_r rdplot_mean_bin if rdplot_id>0, sort color(gs11)) /// (scatter rdplot_mean_y rdplot_mean_bin, sort msize(small) mcolor(gs10)) /// (line y_plot_l x_plot_l, lcolor(black) sort lwidth(medthin) lpattern(solid)) /// (line y_plot_r x_plot_r, lcolor(black) sort lwidth(medthin) lpattern(solid)) , /// xline(`c', lcolor(black) lwidth(medthin)) xscale(r(`x_min' `x_max')) legend(cols(2) order(2 "Sample average within bin" 3 "Polynomial fit of order `p'" )) `graph_options' } } } restore **************************** ** PART 2: genvars=TRUE **************************** if ("`genvars'"!="") { qui for any id N min_bin max_bin mean_bin mean_x mean_y se_y ci_l ci_r hat_y: qui gen rdplot_X = . } mata { if ("`genvars'"!="") { rdplot = rdplot_id, rdplot_N, rdplot_min_bin, rdplot_max_bin, rdplot_mean_bin, rdplot_mean_x, rdplot_mean_y, rdplot_se_y, rdplot_ci_l, rdplot_ci_r st_view(ZZ=.,., "`x' rdplot_id rdplot_N rdplot_min_bin rdplot_max_bin rdplot_mean_bin rdplot_mean_x rdplot_mean_y rdplot_se_y rdplot_ci_l rdplot_ci_r rdplot_hat_y", "`touse'") for (i=1; i<=rows(ZZ); i++) { if (ZZ[i,1]!=.) { bin_i = 2; while(ZZ[i,1] >= bins[bin_i] & bin_i < length(bins)) bin_i++ rdplot_i = bin_i - `J_star_l' - 2 if (rdplot_i >= 0) rdplot_i = rdplot_i + 1 ZZ[i,2..11] = select(rdplot, rdplot[.,1]:==rdplot_i) ZZ[i,12] = 0; for (j=0; j<=p; j++) { if (ZZ[i,2] <0) ZZ[i,12] = ZZ[i,12] + ((ZZ[i,1]-c)^j)*gamma_p1_l[j+1] else ZZ[i,12] = ZZ[i,12] + ((ZZ[i,1]-c)^j)*gamma_p1_r[j+1] } } } } } mata mata clear end