<div dir="ltr"><div>Prezados(as)</div><div>Ajustei o seguinte modelo de GEE para dados binários, longitudinais:</div><div>load(<a href="https://dl.dropboxusercontent.com/u/73337918/d6.rda">https://dl.dropboxusercontent.com/u/73337918/d6.rda</a>)</div><div>attach(d6)</div><div><br></div><div>d6$REF_P <- as.factor(d6$REF_P)</div><div><br></div><div>d6$TCM_PRE <- as.factor(d6$TCM_PRE)</div><div><br></div><div>d6$TCM_REL <- as.factor(d6$TCM_REL)</div><div><br></div><div>d6$MAN_CON <- as.factor(d6$MAN_CON)</div><div><br></div><div>d6$MAN_PER <- as.factor(d6$MAN_PER)</div><div><br></div><div>d6$EST_CIV <- as.factor(d6$EST_CIV)</div><div><br></div><div>d6$GR_ESC <- as.factor(d6$GR_ESC)</div><div><br></div><div>d6$GR_OCUP <- as.factor(d6$GR_OCUP)</div><div><br></div><div>d6$GR_ESCP <- as.factor(d6$GR_ESCP)</div><div><br></div><div>d6$INT_PGOV <- as.factor(d6$INT_PGOV)</div><div><br></div><div>d6$INT_PPRE <- as.factor(d6$INT_PPRE)</div><div><br></div><div>d6$INT_PART <- as.factor(d6$INT_PART)</div><div><br></div><div>d6$GR_DCAP <- as.factor(d6$GR_DCAP)</div><div><br></div><div>d6$RESC_ANT <- as.factor(d6$RESC_ANT)</div><div><br></div><div>#</div><div><br></div><div>d6$GR_ESC <- relevel(d6$GR_ESC, ref="LE_FI")</div><div><br></div><div>d6$REG_MES <- relevel(d6$REG_MES, ref="Ms_Mt")</div><div><br></div><div>library(geepack)</div><div><br></div><div>mod6c <- geeglm(RES_CONT ~ GR_ESC+GR_OCUP+SEXO+MAN_PER+REF_P+ Â  Â  Â  Â  Â  Â  Â  Â  Â TCM_PRE+TCM_REL+INT_PGOV+INT_PPRE+REG_MES+GR_DCAP+IDHM_PNUD+RESC_ANT,id = MUNIC, Â data = d6, family = binomial(link=logit), corstr = "ar1")</div><div><br></div><div>summary(mod6c)</div><div><br></div><div> </div><div><br></div><div>Coefficients:</div><div>  Â  Â  Â  Â  Â  Â  Estimate Â Std.err Â  Â Wald Pr(>|W|) Â  Â </div><div>(Intercept) Â  Â 1.54944 Â 0.58054 Â  7.123 Â 0.00761 ** </div><div>GR_ESCFC_MI Â  Â 0.05587 Â 0.15362 Â  0.132 Â 0.71610 Â  Â </div><div>GR_ESCMC_SI Â  Â 0.25302 Â 0.12323 Â  4.215 Â 0.04006 * Â </div><div>GR_ESCSC_PG Â  Â 0.47851 Â 0.15347 Â  9.722 Â 0.00182 ** </div><div>GR_OCUP2 Â  Â  Â -0.01318 Â 0.19240 Â  0.005 Â 0.94540 Â  Â </div><div>GR_OCUP3 Â  Â  Â -0.49723 Â 0.20344 Â  5.974 Â 0.01452 * Â </div><div>GR_OCUP4 Â  Â  Â  0.20693 Â 0.18915 Â  1.197 Â 0.27397 Â  Â </div><div>GR_OCUP5 Â  Â  Â  0.05465 Â 0.13705 Â  0.159 Â 0.69005 Â  Â </div><div>GR_OCUP6 Â  Â  Â -0.16555 Â 0.20084 Â  0.679 Â 0.40979 Â  Â </div><div>GR_OCUP7 Â  Â  Â  0.25329 Â 0.22578 Â  1.259 Â 0.26192 Â  Â </div><div>GR_OCUP8 Â  Â  Â -0.56514 Â 0.20535 Â  7.574 Â 0.00592 ** </div><div>GR_OCUP9 Â  Â  Â  0.11389 Â 0.23092 Â  0.243 Â 0.62187 Â  Â </div><div>GR_OCUP10 Â  Â  -0.32457 Â 0.16519 Â  3.860 Â 0.04944 * Â </div><div>SEXOM Â  Â  Â  Â  -0.38449 Â 0.15291 Â  6.322 Â 0.01192 * Â </div><div>MAN_PER2 Â  Â  Â -0.02760 Â 0.08483 Â  0.106 Â 0.74493 Â  Â </div><div>MAN_PER3 Â  Â  Â  0.89274 Â 0.22867 Â 15.241 9.46e-05 ***</div><div>REF_P2 Â  Â  Â  Â -0.90280 Â 0.16081 Â 31.516 1.98e-08 ***</div><div>TCM_PRE2 Â  Â  Â  0.32531 Â 0.17539 Â  3.440 Â 0.06362 . Â </div><div>TCM_PRE3 Â  Â  Â  1.10453 Â 0.38724 Â  8.136 Â 0.00434 ** </div><div>TCM_PRE4 Â  Â  Â  0.07393 Â 0.16962 Â  0.190 Â 0.66292 Â  Â </div><div>TCM_PRE5 Â  Â  Â  0.45656 Â 0.17748 Â  6.618 Â 0.01010 * Â </div><div>TCM_PRE6 Â  Â  Â -1.07012 Â 0.43169 Â  6.145 Â 0.01318 * Â </div><div>TCM_REL2 Â  Â  Â -0.88695 Â 0.22036 Â 16.202 5.69e-05 ***</div><div>TCM_REL3 Â  Â  Â -0.51005 Â 0.23513 Â  4.706 Â 0.03006 * Â </div><div>TCM_REL4 Â  Â  Â -0.20505 Â 0.21699 Â  0.893 Â 0.34466 Â  Â </div><div>TCM_REL5 Â  Â  Â  0.05124 Â 0.26072 Â  0.039 Â 0.84418 Â  Â </div><div>TCM_REL6 Â  Â  Â -0.53652 Â 0.27731 Â  3.743 Â 0.05303 . Â </div><div>TCM_REL7 Â  Â  Â -0.49148 Â 0.21464 Â  5.243 Â 0.02204 * Â </div><div>TCM_REL8 Â  Â  Â  0.21678 Â 0.21973 Â  0.973 Â 0.32385 Â  Â </div><div>TCM_REL9 Â  Â  Â  0.32581 Â 0.31959 Â  1.039 Â 0.30799 Â  Â </div><div>TCM_REL10 Â  Â  Â 0.59204 Â 0.25881 Â  5.233 Â 0.02216 * Â </div><div>TCM_REL11 Â  Â  -0.06144 Â 0.22802 Â  0.073 Â 0.78759 Â  Â </div><div>TCM_REL12 Â  Â  -0.51748 Â 0.28703 Â  3.250 Â 0.07140 . Â </div><div>INT_PGOV1 Â  Â  -0.86442 Â 0.37774 Â  5.237 Â 0.02211 * Â </div><div>INT_PGOV2 Â  Â  -0.62852 Â 0.39466 Â  2.536 Â 0.11126 Â  Â </div><div>INT_PPRE1 Â  Â  Â 1.07552 Â 0.37019 Â  8.441 Â 0.00367 ** </div><div>INT_PPRE2 Â  Â  Â 0.94990 Â 0.43197 Â  4.836 Â 0.02788 * Â </div><div>REG_MESMs_CN Â -0.29830 Â 0.19545 Â  2.329 Â 0.12696 Â  Â </div><div>REG_MESMs_CS Â -0.25052 Â 0.22930 Â  1.194 Â 0.27460 Â  Â </div><div>REG_MESMs_EO Â  0.01757 Â 0.31662 Â  0.003 Â 0.95575 Â  Â </div><div>REG_MESMs_Nd Â -0.41248 Â 0.20274 Â  4.139 Â 0.04190 * Â </div><div>REG_MESMs_S Â  -1.09423 Â 0.22003 Â 24.732 6.59e-07 ***</div><div>REG_MESMs_VSF -0.68058 Â 0.26449 Â  6.621 Â 0.01008 * Â </div><div>GR_DCAP2 Â  Â  Â  0.05559 Â 0.15406 Â  0.130 Â 0.71821 Â  Â </div><div>GR_DCAP3 Â  Â  Â -0.02412 Â 0.18018 Â  0.018 Â 0.89351 Â  Â </div><div>GR_DCAP4 Â  Â  Â  0.21081 Â 0.21921 Â  0.925 Â 0.33622 Â  Â </div><div>GR_DCAP5 Â  Â  Â  0.62488 Â 0.22953 Â  7.412 Â 0.00648 ** </div><div>GR_DCAP6 Â  Â  Â  0.60297 Â 0.74007 Â  0.664 Â 0.41522 Â  Â </div><div>IDHM_PNUD Â  Â  -2.48469 Â 0.77384 Â 10.310 Â 0.00132 ** </div><div>RESC_ANT1 Â  Â  Â 1.60297 Â 0.08113 390.413 Â < 2e-16 ***</div><div> </div><div><br></div><div>Não consegui entender os resultados seguintes:</div><div><br></div><div> </div><div><br></div><div>INT_PGOV1 Â  Â  -0.86442 Â 0.37774 Â  5.237 Â 0.02211 * </div><div>INT_PGOV2 Â  Â  -0.62852 Â 0.39466 Â  2.536 Â 0.11126</div><div> </div><div><br></div><div>..............................................</div><div><br></div><div>> tapply(fitted(mod6c), INT_PGOV, mean)</div><div>  Â  Â 0 Â  Â  Â 1 Â  Â  Â 2 </div><div>0.6868 0.7403 0.7717 </div><div>> tapply(predict(mod6c), INT_PGOV, mean)</div><div>  Â  Â 0 Â  Â  Â 1 Â  Â  Â 2 </div><div>0.9619 1.2828 1.4899 </div><div> </div><div><br></div><div>A variável categórica INT_PGOV tem três níveis: 0, 1, e 2. O R coloca o 0 como baseline. Se o preditor linear e o valor ajustado da resposta são maiores nos níveis 1 e 2 de INT_PGOV, as estimativas dos parâmetros correspondentes a (INT_PGOV1:-0.86442) e (INT_PGOV2: -0.62852) não deviam ser positivas ????.</div><div><br></div><div> </div><div><br></div><div>Alguém tem alguma ideia? Desde já agradeço qualquer comentário.</div><div><br></div><div>Att.</div><div><br></div>-- <br><div class="gmail_signature"><div dir="ltr"><div><div dir="ltr">Gilênio Borges Fernandes<div>Professor Associado IV (Aposentado)</div><div>Professor Adjunto A (Substituto)<br>Universidade Federal da Bahia<br>Instituto de Matemática<br>Departamento de Estatística<br>Av. Adhemar de Barros, s/n â€“ Ondina.<br>40.170-110 - Salvador - BA, Brasil<br>Tel.: (071)3283-6340/6341/6337 Â Fax:  (071)3283-6336<br>URL: <a href="http://lattes.cnpq.br/6764860618464860" target="_blank">http://lattes.cnpq.br/6764860618464860</a><br></div></div></div></div></div>
</div>