Creamos las mismas variables usadas en la clase anterior

setwd("/Users/mau/Dropbox/Cursos mau/Maestria CEI")

data<-read.csv("BaseIP3.csv",header=TRUE)

table(data$p7_1)
## 
## NS/NC    No    Si 
##    41  8779  2180
table(data$p7_2)
## 
## NS/NC    No    Si 
##    64  8728  2208
table(data$p7_3)
## 
## NS/NC    No    Si 
##    45 10289   666
table(data$p8)
## 
##                        Ambos                           NC 
##                          209                           17 
##                           NS                           No 
##                           24                         8526 
## Si, alguien de esta vivienda       Si, yo he sido victima 
##                          945                         1279
data$lcompras<-ifelse(data$p7_1=="Si",data$ponde,0)
data$lpaseo<-ifelse(data$p7_2=="Si",data$ponde,0)
data$lvive<-ifelse(data$p7_3=="Si",data$ponde,0)
data$delito<-ifelse(data$p8=="Si, alguien de esta vivienda",data$ponde,0)
data$delito<-ifelse(data$p8=="Si, yo he sido victima",data$ponde,data$delito)
data$total<-data$ponde

data2<-aggregate(cbind(lcompras,lpaseo,lvive,delito,total)~edo+muni,data=data,sum)

data2$clave<-(data2$edo*1000)+data2$muni

dataSUN<-read.csv("SUN2012.csv",header=TRUE)

data4<-merge(data2,dataSUN,by.x="clave",by.y="clave1",all.x=T, all.y=F)

data4$tipo<-ifelse(is.na(data4$tipo),0,data4$tipo)

data4$plpaseo<-100*(data4$lpaseo/data4$total)
data4$plcompras<-100*(data4$lcompras/data4$total)
data4$plvive<-100*(data4$lvive/data4$total)
data4$tdelito<-100*(data4$delito/data4$total)

library(ggplot2)

p<-ggplot(data4,aes(tdelito,plcompras))
p+geom_point()+labs(x="Tasa de delitos",y="% ha cambiado lugar de compras", 
                    title="Diagrama de dispersion 1")

cor(data4$plcompras,data4$tdelito)
## [1] 0.265463
p<-ggplot(data4,aes(tdelito,plpaseo))
p+geom_point()+labs(x="Tasa de delitos",y="% ha cambiado lugar de paseo", 
                    title="Diagrama de dispersion 1")

cor(data4$plpaseo,data4$tdelito)
## [1] 0.3287581
p<-ggplot(data4,aes(tdelito,plvive))
p+geom_point()+labs(x="Tasa de delitos",y="% ha cambiado lugar de residencia", 
                    title="Diagrama de dispersion 1")

cor(data4$plvive,data4$tdelito)
## [1] 0.2052642
p<-ggplot(data4,aes(tdelito,plcompras, colour=tipo))
p+geom_point()+labs(x="Tasa de delitos",y="% ha cambiado lugar de compras", 
                    title="Diagrama de dispersion 1")

p<-ggplot(data4,aes(tdelito,plpaseo, colour=tipo))
p+geom_point()+labs(x="Tasa de delitos",y="% ha cambiado lugar de paseo", 
                    title="Diagrama de dispersion 1")

p<-ggplot(data4,aes(tdelito,plvive,colour=tipo))
p+geom_point()+labs(x="Tasa de delitos",y="% ha cambiado lugar de residencia", 
                    title="Diagrama de dispersion 1")

t.test(data4$tipo,data4$plcompras)
## 
##  Welch Two Sample t-test
## 
## data:  data4$tipo and data4$plcompras
## t = -19.181, df = 347.55, p-value < 2.2e-16
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
##  -17.69547 -14.40397
## sample estimates:
##  mean of x  mean of y 
##  0.2614943 16.3112139
t.test(data4$tipo,data4$plpaseo)
## 
##  Welch Two Sample t-test
## 
## data:  data4$tipo and data4$plpaseo
## t = -20.345, df = 347.55, p-value < 2.2e-16
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
##  -18.71932 -15.41900
## sample estimates:
##  mean of x  mean of y 
##  0.2614943 17.3306509
t.test(data4$tipo,data4$plvive)
## 
##  Welch Two Sample t-test
## 
## data:  data4$tipo and data4$plvive
## t = -10.338, df = 348.59, p-value < 2.2e-16
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
##  -6.064571 -4.125781
## sample estimates:
## mean of x mean of y 
## 0.2614943 5.3566702
t.test(data4$tipo,data4$tdelito)
## 
##  Welch Two Sample t-test
## 
## data:  data4$tipo and data4$tdelito
## t = -22.214, df = 347.57, p-value < 2.2e-16
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
##  -19.94582 -16.70116
## sample estimates:
##  mean of x  mean of y 
##  0.2614943 18.5849854
data4$pobmun<-0
data4$pobmun<-ifelse(data4$total>0 & data4$total<50000,1,data4$pobmun)
data4$pobmun<-ifelse(data4$total>=50000 & data4$total<100000,2,data4$pobmun)
data4$pobmun<-ifelse(data4$total>=100000 & data4$total<500000,3,data4$pobmun)
data4$pobmun<-ifelse(data4$total>=500000,4,data4$pobmun)
table(data4$pobmun)
## 
##   1   2   3   4 
##  90  65 132  61

Vamos a seguir una estrategia conocida como “Modelos Anidados” esta aproximacon nos sirve para probar hipotesis alternativas

Consiste en ajustar un modelo solamente con nuestra variable dependiente y la variable independiente de interes,

Despues incorporamos varibles “control”

modelo1<-lm(plcompras~tdelito,data=data4)
summary(modelo1)
## 
## Call:
## lm(formula = plcompras ~ tdelito, data = data4)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -30.531 -11.306  -2.746   7.962  58.743 
## 
## Coefficients:
##             Estimate Std. Error t value Pr(>|t|)    
## (Intercept) 11.30632    1.26772   8.919  < 2e-16 ***
## tdelito      0.26930    0.05258   5.122 5.04e-07 ***
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 15.07 on 346 degrees of freedom
## Multiple R-squared:  0.07047,    Adjusted R-squared:  0.06778 
## F-statistic: 26.23 on 1 and 346 DF,  p-value: 5.04e-07
modelo2<-lm(plcompras~tdelito+plvive,data=data4)
summary(modelo2)
## 
## Call:
## lm(formula = plcompras ~ tdelito + plvive, data = data4)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -32.390  -9.009  -1.811   7.432  52.194 
## 
## Coefficients:
##             Estimate Std. Error t value Pr(>|t|)    
## (Intercept)  9.00920    1.16324   7.745 1.07e-13 ***
## tdelito      0.17785    0.04816   3.693 0.000258 ***
## plvive       0.74610    0.08066   9.250  < 2e-16 ***
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 13.5 on 345 degrees of freedom
## Multiple R-squared:  0.2552, Adjusted R-squared:  0.2509 
## F-statistic:  59.1 on 2 and 345 DF,  p-value: < 2.2e-16

Para comparar modelos usamos una prueba anova

anova(modelo1,modelo2)
## Analysis of Variance Table
## 
## Model 1: plcompras ~ tdelito
## Model 2: plcompras ~ tdelito + plvive
##   Res.Df   RSS Df Sum of Sq      F    Pr(>F)    
## 1    346 78529                                  
## 2    345 62922  1     15606 85.567 < 2.2e-16 ***
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
modelo3<-lm(plcompras~tdelito+plvive+tipo,data=data4)
summary(modelo3)
## 
## Call:
## lm(formula = plcompras ~ tdelito + plvive + tipo, data = data4)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -31.749  -8.956  -1.595   7.386  53.146 
## 
## Coefficients:
##             Estimate Std. Error t value Pr(>|t|)    
## (Intercept)  8.53502    1.18353   7.211 3.55e-12 ***
## tdelito      0.15277    0.04964   3.077  0.00226 ** 
## plvive       0.75926    0.08061   9.419  < 2e-16 ***
## tipo         3.32638    1.69887   1.958  0.05104 .  
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 13.45 on 344 degrees of freedom
## Multiple R-squared:  0.2634, Adjusted R-squared:  0.257 
## F-statistic:    41 on 3 and 344 DF,  p-value: < 2.2e-16
anova(modelo2,modelo3)
## Analysis of Variance Table
## 
## Model 1: plcompras ~ tdelito + plvive
## Model 2: plcompras ~ tdelito + plvive + tipo
##   Res.Df   RSS Df Sum of Sq      F  Pr(>F)  
## 1    345 62922                              
## 2    344 62229  1    693.52 3.8338 0.05104 .
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
modelo4<-lm(plcompras~tdelito+plvive+tipo+factor(pobmun),data=data4)
summary(modelo4)
## 
## Call:
## lm(formula = plcompras ~ tdelito + plvive + tipo + factor(pobmun), 
##     data = data4)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -30.245  -8.429  -2.939   6.723  51.092 
## 
## Coefficients:
##                 Estimate Std. Error t value Pr(>|t|)    
## (Intercept)      5.75932    1.53731   3.746 0.000211 ***
## tdelito          0.12130    0.05066   2.394 0.017200 *  
## plvive           0.75846    0.07990   9.492  < 2e-16 ***
## tipo             3.23907    1.73551   1.866 0.062850 .  
## factor(pobmun)2  5.95890    2.18422   2.728 0.006699 ** 
## factor(pobmun)3  5.42920    1.91890   2.829 0.004941 ** 
## factor(pobmun)4  1.22884    2.33167   0.527 0.598522    
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 13.27 on 341 degrees of freedom
## Multiple R-squared:  0.2896, Adjusted R-squared:  0.2771 
## F-statistic: 23.17 on 6 and 341 DF,  p-value: < 2.2e-16
anova(modelo3,modelo4)
## Analysis of Variance Table
## 
## Model 1: plcompras ~ tdelito + plvive + tipo
## Model 2: plcompras ~ tdelito + plvive + tipo + factor(pobmun)
##   Res.Df   RSS Df Sum of Sq      F   Pr(>F)   
## 1    344 62229                                
## 2    341 60017  3    2211.5 4.1884 0.006255 **
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1