---------------------------------------------------------------------------------- name: log: /Users/denisardalves/Desktop/EAE 324 2017/DATA SET/Chap5Asyntotics.lo > g log type: text opened on: 1 Jun 2017, 11:36:24 . . * open data . use "/Users/denisardalves/Desktop/EAE 324 2017/DATA SET/CRIME1.DTA", clear . /*Para voces verem que narr86, mesmo com n=2724 está muito longe da dist. normal > , vou usar o comando kdensity em narr86*/ . . kdensity narr86, bwidth(0.2) normal . * Vou selecionar uma amostra das 2756 obs dos dados de CRIME1.DTA. PARA ISSO uso > o COMANDO bsample com n = 500. Quero mostrar que as diferenças entre os testes > de WALD, LR E LM, diminuem a medida que n aumenta.*/ . bsample 500 . * estimar regressão sem restrições . reg narr86 avgsen tottime qemp86 black hispan inc86 Source | SS df MS Number of obs = 500 -------------+---------------------------------- F(6, 493) = 7.78 Model | 28.6077611 6 4.76796018 Prob > F = 0.0000 Residual | 302.310239 493 .613205353 R-squared = 0.0864 -------------+---------------------------------- Adj R-squared = 0.0753 Total | 330.918 499 .663162325 Root MSE = .78307 ------------------------------------------------------------------------------ narr86 | Coef. Std. Err. t P>|t| [95% Conf. Interval] -------------+---------------------------------------------------------------- avgsen | .0106776 .0224314 0.48 0.634 -.0333953 .0547506 tottime | -.0227924 .0169133 -1.35 0.178 -.0560234 .0104385 qemp86 | -.055027 .0316017 -1.74 0.082 -.1171177 .0070637 black | .465073 .1006902 4.62 0.000 .2672383 .6629078 hispan | .1821213 .0875337 2.08 0.038 .0101362 .3541065 inc86 | -.0010432 .0007529 -1.39 0.167 -.0025226 .0004361 _cons | .4455183 .0700229 6.36 0.000 .3079382 .5830984 ------------------------------------------------------------------------------ . . estimates store regsr . ereturn list scalars: e(rank) = 7 e(ll_0) = -606.2848946335176 e(ll) = -583.6806847991243 e(r2_a) = .0753314385537205 e(rss) = 302.3102389002462 e(mss) = 28.60776109975302 e(rmse) = .7830742957975547 e(r2) = .0864496978095876 e(F) = 7.775470585826478 e(df_r) = 493 e(df_m) = 6 e(N) = 500 macros: e(_estimates_name) : "regsr" e(cmdline) : "regress narr86 avgsen tottime qemp86 black hispan in.." e(title) : "Linear regression" e(marginsok) : "XB default" e(vce) : "ols" e(depvar) : "narr86" e(cmd) : "regress" e(properties) : "b V" e(predict) : "regres_p" e(model) : "ols" e(estat_cmd) : "regress_estat" matrices: e(b) : 1 x 7 e(V) : 7 x 7 functions: e(sample) . scalar sqrsr = e(rss) . *estimar regressão com restrições, supondo que raça não afeta criminalidade . reg narr86 avgsen tottime qemp86 inc86 Source | SS df MS Number of obs = 500 -------------+---------------------------------- F(4, 495) = 5.84 Model | 14.9077933 4 3.72694833 Prob > F = 0.0001 Residual | 316.010207 495 .638404458 R-squared = 0.0450 -------------+---------------------------------- Adj R-squared = 0.0373 Total | 330.918 499 .663162325 Root MSE = .799 ------------------------------------------------------------------------------ narr86 | Coef. Std. Err. t P>|t| [95% Conf. Interval] -------------+---------------------------------------------------------------- avgsen | .0107924 .0228157 0.47 0.636 -.0340352 .0556199 tottime | -.0174822 .0171569 -1.02 0.309 -.0511915 .0162271 qemp86 | -.0610515 .0321573 -1.90 0.058 -.1242331 .0021301 inc86 | -.0012051 .0007644 -1.58 0.116 -.0027071 .0002968 _cons | .578014 .0637765 9.06 0.000 .452708 .7033199 ------------------------------------------------------------------------------ . . estimates store regcr . ereturn list scalars: e(rank) = 5 e(ll_0) = -606.2848946335176 e(ll) = -594.7608701611538 e(r2_a) = .0373330417118067 e(rss) = 316.0102066711257 e(mss) = 14.90779332887348 e(rmse) = .7990021639028684 e(r2) = .0450498109165217 e(F) = 5.837910882315367 e(df_r) = 495 e(df_m) = 4 e(N) = 500 macros: e(_estimates_name) : "regcr" e(cmdline) : "regress narr86 avgsen tottime qemp86 inc86" e(title) : "Linear regression" e(marginsok) : "XB default" e(vce) : "ols" e(depvar) : "narr86" e(cmd) : "regress" e(properties) : "b V" e(predict) : "regres_p" e(model) : "ols" e(estat_cmd) : "regress_estat" matrices: e(b) : 1 x 5 e(V) : 5 x 5 functions: e(sample) . scalar sqrcr = e(rss) . * compute LR . scalar LR=e(N)*(log(sqrcr) - log(sqrsr)) . * se usarmos LR2=nln(1+(sqrcr-sqrsr)/sqrsr). temos . . scalar LR2=e(N)*(log(1+(sqrcr-sqrsr)/sqrsr)) . . di LR2 22.160371 . * computo da estatística WALD . scalar WALD=e(N)*(sqrcr-sqrsr)/sqrsr . di WALD 22.658789 . * computo do LM . *1o Passo: . . quietly reg narr86 avgsen tottime qemp86 inc86 . predict rescr, residuals . *2o Passo: . . reg rescr avgsen tottime qemp86 black hispan inc86 Source | SS df MS Number of obs = 500 -------------+---------------------------------- F(6, 493) = 3.72 Model | 13.6999679 6 2.28332798 Prob > F = 0.0012 Residual | 302.310242 493 .613205358 R-squared = 0.0434 -------------+---------------------------------- Adj R-squared = 0.0317 Total | 316.01021 499 .633286993 Root MSE = .78307 ------------------------------------------------------------------------------ rescr | Coef. Std. Err. t P>|t| [95% Conf. Interval] -------------+---------------------------------------------------------------- avgsen | -.0001147 .0224314 -0.01 0.996 -.0441877 .0439582 tottime | -.0053102 .0169133 -0.31 0.754 -.0385412 .0279207 qemp86 | .0060245 .0316017 0.19 0.849 -.0560662 .0681152 black | .465073 .1006902 4.62 0.000 .2672383 .6629078 hispan | .1821213 .0875337 2.08 0.038 .0101362 .3541065 inc86 | .0001619 .0007529 0.22 0.830 -.0013174 .0016412 _cons | -.1324956 .0700229 -1.89 0.059 -.2700758 .0050845 ------------------------------------------------------------------------------ . *3o Passo: . . scalar LM=e(N)*e(r2) . di LM 21.676464 . di LR 22.160371 . di WALD 22.658789 . scalar chic = invchi2tail(2,.05) . di "Chi-square(2) 95th percentile = " chic Chi-square(2) 95th percentile = 5.9914645 . scalar pvalue = chi2tail(2,WALD) . di pvalue .00001201 . log close name: log: /Users/denisardalves/Desktop/EAE 324 2017/DATA SET/Chap5Asyntotics.lo > g log type: text closed on: 1 Jun 2017, 11:36:25 ----------------------------------------------------------------------------------