El diseño muestral es el que se logró armar considerando los diseños de cada país Finalmente se trabaja con un diseño equivalente a 2 etapas, primera donde las psu son las escuelas, el pais es la variable de estratificación y se usan como pesos muestrales los que se consideraron en Uruguay (ver diseño de Uruguay y para POA se creó un nuevo vector de peso usando como variables de ajuste distribucion por sexo y por tipo de escuela)
Preciso que me tires una multiple con aquellas variables que son tiene un p valor <=0,10: 1) País 2) Sociecinómico 4cat 3) Escolmae cat 23 cat 4) Visitacatonde 5) visiquando 6) frecuenciaescovacat 7)Isg 45% 8) tipo de escuela
library(survey)
## Attaching package: 'survey'
##
## The following object(s) are masked from 'package:graphics':
##
## dotchart
library(car)
## Loading required package: MASS Loading required package: nnet
# 28 de octubre 2013
load("~/Dropbox/odontologia/maestria_anunziatta/BancoUnidoBrasUru/BRASURU(28102013).RData")
levels(disenio_urubra$variables$prevoms)
## [1] "1-SinCavi" "2-conCavi"
table(disenio_urubra$variables$cpodoms, disenio_urubra$variables$prevoms)
##
## 1-SinCavi 2-conCavi
## 0 1105 0
## 1 0 495
## 2 0 404
## 3 0 259
## 4 0 217
## 5 0 108
## 6 0 38
## 7 0 22
## 8 0 14
## 9 0 12
## 10 0 3
## 11 0 3
## 14 0 1
## 15 0 1
modelo2multi.bin <- svyglm(prevoms ~ pais + socioecon4cat + escolmaerecat23cat +
visidentcatonde + visiquando + freqescovacat + isg45 + tipoesc, disenio_urubra,
family = quasibinomial())
summary(modelo2multi.bin)
##
## Call:
## svyglm(formula = prevoms ~ pais + socioecon4cat + escolmaerecat23cat +
## visidentcatonde + visiquando + freqescovacat + isg45 + tipoesc,
## disenio_urubra, family = quasibinomial())
##
## Survey design:
## svydesign(id = ~numer_esc, strata = ~pais, weights = ~round(weight.rec,
## 1), data = brasuru, nest = TRUE)
##
## Coefficients:
## Estimate Std. Error t value
## (Intercept) -0.759 0.258 -2.93
## paisUru 0.604 0.130 4.65
## socioecon4cat2-medio-alto 0.442 0.256 1.73
## socioecon4cat3-medio-bajo 0.604 0.261 2.32
## socioecon4cat4-bajo 0.746 0.325 2.30
## escolmaerecat23cat2-high school 0.148 0.212 0.70
## escolmaerecat23cat3-elementary school 0.477 0.261 1.83
## visidentcatonde2-Publico 0.497 0.150 3.32
## visidentcatonde3-Nunca fue al dentista -0.498 0.866 -0.58
## visiquando2- m\xe1s de 2 a\xf1os -0.282 0.148 -1.91
## visiquando3- nunca fue 0.524 0.870 0.60
## freqescovacat2-veces al d\xeda -0.185 0.144 -1.28
## freqescovacat3-veces al d\xeda -0.335 0.170 -1.97
## isg452- 45 a 60 0.134 0.127 1.05
## isg453- >=60 0.517 0.130 3.96
## tipoesc2-Publica 0.120 0.163 0.73
## Pr(>|t|)
## (Intercept) 0.00459 **
## paisUru 1.7e-05 ***
## socioecon4cat2-medio-alto 0.08914 .
## socioecon4cat3-medio-bajo 0.02364 *
## socioecon4cat4-bajo 0.02476 *
## escolmaerecat23cat2-high school 0.48899
## escolmaerecat23cat3-elementary school 0.07205 .
## visidentcatonde2-Publico 0.00149 **
## visidentcatonde3-Nunca fue al dentista 0.56709
## visiquando2- m\xe1s de 2 a\xf1os 0.06108 .
## visiquando3- nunca fue 0.54874
## freqescovacat2-veces al d\xeda 0.20432
## freqescovacat3-veces al d\xeda 0.05285 .
## isg452- 45 a 60 0.29587
## isg453- >=60 0.00018 ***
## tipoesc2-Publica 0.46593
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for quasibinomial family taken to be 0.9922)
##
## Number of Fisher Scoring iterations: 4
anova(modelo2multi.bin)
## Anova table: (Rao-Scott LRT)
## svyglm(formula = prevoms ~ pais, disenio_urubra, family = quasibinomial())
## stats DEff df ddf p
## pais 9.85 3.22 1.00 80 0.0865 .
## socioecon4cat 146.93 2.18 3.00 77 6.1e-09 ***
## escolmaerecat23cat 88.02 2.02 2.00 75 1.7e-06 ***
## visidentcatonde 52.00 1.28 2.00 73 1.1e-06 ***
## visiquando 37.49 1.28 2.00 71 1.7e-05 ***
## freqescovacat 59.55 1.82 2.00 69 5.0e-06 ***
## isg45 16.99 1.05 2.00 67 0.0013 **
## tipoesc 1.07 1.99 1.00 66 0.4652
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
confint(modelo2multi.bin)
## 2.5 % 97.5 %
## (Intercept) -1.26516 -0.251940
## paisUru 0.34942 0.858957
## socioecon4cat2-medio-alto -0.06009 0.943995
## socioecon4cat3-medio-bajo 0.09295 1.114449
## socioecon4cat4-bajo 0.10959 1.381798
## escolmaerecat23cat2-high school -0.26839 0.563851
## escolmaerecat23cat3-elementary school -0.03439 0.987989
## visidentcatonde2-Publico 0.20308 0.790188
## visidentcatonde3-Nunca fue al dentista -2.19437 1.198565
## visiquando2- m\xe1s de 2 a\xf1os -0.57115 0.008055
## visiquando3- nunca fue -1.18060 2.229223
## freqescovacat2-veces al d\xeda -0.46771 0.097807
## freqescovacat3-veces al d\xeda -0.66810 -0.001982
## isg452- 45 a 60 -0.11525 0.383237
## isg453- >=60 0.26140 0.772929
## tipoesc2-Publica -0.20036 0.439942
exp(modelo2multi.bin$coefficients)
## (Intercept)
## 0.4683
## paisUru
## 1.8298
## socioecon4cat2-medio-alto
## 1.5557
## socioecon4cat3-medio-bajo
## 1.8289
## socioecon4cat4-bajo
## 2.1079
## escolmaerecat23cat2-high school
## 1.1592
## escolmaerecat23cat3-elementary school
## 1.6109
## visidentcatonde2-Publico
## 1.6432
## visidentcatonde3-Nunca fue al dentista
## 0.6078
## visiquando2- m\xe1s de 2 a\xf1os
## 0.7546
## visiquando3- nunca fue
## 1.6893
## freqescovacat2-veces al d\xeda
## 0.8311
## freqescovacat3-veces al d\xeda
## 0.7153
## isg452- 45 a 60
## 1.1434
## isg453- >=60
## 1.6773
## tipoesc2-Publica
## 1.1273
modelo3multi.bin <- svyglm(prevoms ~ pais + escolmaerecat23cat + visidentcatonde +
freqescovacat + isg45 + tipoesc, disenio_urubra, family = quasibinomial())
summary(modelo3multi.bin)
##
## Call:
## svyglm(formula = prevoms ~ pais + escolmaerecat23cat + visidentcatonde +
## freqescovacat + isg45 + tipoesc, disenio_urubra, family = quasibinomial())
##
## Survey design:
## svydesign(id = ~numer_esc, strata = ~pais, weights = ~round(weight.rec,
## 1), data = brasuru, nest = TRUE)
##
## Coefficients:
## Estimate Std. Error t value
## (Intercept) -0.611 0.212 -2.88
## paisUru 0.616 0.121 5.08
## escolmaerecat23cat2-high school 0.287 0.196 1.47
## escolmaerecat23cat3-elementary school 0.640 0.237 2.70
## visidentcatonde2-Publico 0.562 0.137 4.10
## visidentcatonde3-Nunca fue al dentista 0.178 0.133 1.33
## freqescovacat2-veces al d\xeda -0.184 0.129 -1.42
## freqescovacat3-veces al d\xeda -0.337 0.167 -2.01
## isg452- 45 a 60 0.150 0.131 1.15
## isg453- >=60 0.545 0.134 4.06
## tipoesc2-Publica 0.257 0.163 1.58
## Pr(>|t|)
## (Intercept) 0.00520 **
## paisUru 2.9e-06 ***
## escolmaerecat23cat2-high school 0.14727
## escolmaerecat23cat3-elementary school 0.00861 **
## visidentcatonde2-Publico 0.00011 ***
## visidentcatonde3-Nunca fue al dentista 0.18627
## freqescovacat2-veces al d\xeda 0.15934
## freqescovacat3-veces al d\xeda 0.04778 *
## isg452- 45 a 60 0.25404
## isg453- >=60 0.00012 ***
## tipoesc2-Publica 0.11886
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for quasibinomial family taken to be 0.9954)
##
## Number of Fisher Scoring iterations: 4
confint(modelo3multi.bin)
## 2.5 % 97.5 %
## (Intercept) -1.02546 -0.195591
## paisUru 0.37862 0.853820
## escolmaerecat23cat2-high school -0.09675 0.669904
## escolmaerecat23cat3-elementary school 0.17587 1.104466
## visidentcatonde2-Publico 0.29298 0.830816
## visidentcatonde3-Nunca fue al dentista -0.08329 0.438796
## freqescovacat2-veces al d\xeda -0.43792 0.069626
## freqescovacat3-veces al d\xeda -0.66429 -0.009062
## isg452- 45 a 60 -0.10581 0.406216
## isg453- >=60 0.28214 0.807585
## tipoesc2-Publica -0.06197 0.575120
exp(modelo3multi.bin$coefficients)
## (Intercept)
## 0.5431
## paisUru
## 1.8519
## escolmaerecat23cat2-high school
## 1.3319
## escolmaerecat23cat3-elementary school
## 1.8968
## visidentcatonde2-Publico
## 1.7540
## visidentcatonde3-Nunca fue al dentista
## 1.1945
## freqescovacat2-veces al d\xeda
## 0.8318
## freqescovacat3-veces al d\xeda
## 0.7141
## isg452- 45 a 60
## 1.1621
## isg453- >=60
## 1.7244
## tipoesc2-Publica
## 1.2925
anova(modelo3multi.bin)
## Anova table: (Rao-Scott LRT)
## svyglm(formula = prevoms ~ pais, disenio_urubra, family = quasibinomial())
## stats DEff df ddf p
## pais 9.85 3.22 1.00 80 0.08646 .
## escolmaerecat23cat 175.44 1.69 2.00 78 1.4e-12 ***
## visidentcatonde 68.50 1.38 2.00 76 6.7e-08 ***
## freqescovacat 63.42 1.80 2.00 74 4.4e-06 ***
## isg45 20.33 1.11 2.00 72 0.00071 ***
## tipoesc 5.45 2.19 1.00 71 0.12190
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
anova(modelo3multi.bin, modelo2multi.bin)
## Working (Rao-Scott+F) LRT for socioecon4cat visiquando
## in svyglm(formula = prevoms ~ pais + socioecon4cat + escolmaerecat23cat +
## visidentcatonde + visiquando + freqescovacat + isg45 + tipoesc,
## disenio_urubra, family = quasibinomial())
## Working 2logLR = 38.46 p= 0.00014
## (scale factors: 2 1.4 0.61 0.54 0.44 ); denominator df= 66
modelo4multi.bin <- svyglm(prevoms ~ pais + escolmaerecat23cat + freqescovacat +
isg45, disenio_urubra, family = quasibinomial())
summary(modelo4multi.bin)
##
## Call:
## svyglm(formula = prevoms ~ pais + escolmaerecat23cat + freqescovacat +
## isg45, disenio_urubra, family = quasibinomial())
##
## Survey design:
## svydesign(id = ~numer_esc, strata = ~pais, weights = ~round(weight.rec,
## 1), data = brasuru, nest = TRUE)
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) -0.418 0.213 -1.96 0.0536
## paisUru 0.666 0.120 5.56 4.1e-07
## escolmaerecat23cat2-high school 0.483 0.158 3.06 0.0031
## escolmaerecat23cat3-elementary school 0.977 0.178 5.50 5.1e-07
## freqescovacat2-veces al d\xeda -0.272 0.136 -2.00 0.0496
## freqescovacat3-veces al d\xeda -0.409 0.156 -2.62 0.0106
## isg452- 45 a 60 0.207 0.132 1.57 0.1214
## isg453- >=60 0.619 0.146 4.24 6.3e-05
##
## (Intercept) .
## paisUru ***
## escolmaerecat23cat2-high school **
## escolmaerecat23cat3-elementary school ***
## freqescovacat2-veces al d\xeda *
## freqescovacat3-veces al d\xeda *
## isg452- 45 a 60
## isg453- >=60 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for quasibinomial family taken to be 1.002)
##
## Number of Fisher Scoring iterations: 4
confint(modelo4multi.bin)
## 2.5 % 97.5 %
## (Intercept) -0.83630 -0.0002948
## paisUru 0.43092 0.9005478
## escolmaerecat23cat2-high school 0.17356 0.7928739
## escolmaerecat23cat3-elementary school 0.62892 1.3249564
## freqescovacat2-veces al d\xeda -0.53926 -0.0049417
## freqescovacat3-veces al d\xeda -0.71529 -0.1035642
## isg452- 45 a 60 -0.05187 0.4654884
## isg453- >=60 0.33336 0.9052886
exp(modelo4multi.bin$coefficients)
## (Intercept)
## 0.6582
## paisUru
## 1.9459
## escolmaerecat23cat2-high school
## 1.6213
## escolmaerecat23cat3-elementary school
## 2.6563
## freqescovacat2-veces al d\xeda
## 0.7618
## freqescovacat3-veces al d\xeda
## 0.6640
## isg452- 45 a 60
## 1.2297
## isg453- >=60
## 1.8577
anova(modelo4multi.bin)
## Anova table: (Rao-Scott LRT)
## svyglm(formula = prevoms ~ pais, disenio_urubra, family = quasibinomial())
## stats DEff df ddf p
## pais 9.85 3.22 1.00 80 0.08646 .
## escolmaerecat23cat 175.44 1.69 2.00 78 1.4e-12 ***
## freqescovacat 66.45 1.72 2.00 76 7.7e-07 ***
## isg45 24.57 1.21 2.00 74 0.00043 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
anova(modelo4multi.bin, modelo3multi.bin)
## Working (Rao-Scott+F) LRT for visidentcatonde tipoesc
## in svyglm(formula = prevoms ~ pais + escolmaerecat23cat + visidentcatonde +
## freqescovacat + isg45 + tipoesc, disenio_urubra, family = quasibinomial())
## Working 2logLR = 40.78 p= 1.1e-05
## (scale factors: 1.7 0.8 0.5 ); denominator df= 71