> mdata <-read.table("vote.csv", header=T, sep=",")
> y  <-mdata[,1]
> x1 <-mdata[,2]
> x2 <-mdata[,3]
> x3 <-mdata[,4]
> x4 <-mdata[,5]
> mydata <-data.frame(y,x1,x2,x3,x4)
> logit <- glm(y ~ x1 + x2 + x3 + x4, family=binomial(link="logit"), data=mydata) 
> summary(logit)

Call:
glm(formula = y ~ x1 + x2 + x3 + x4, family = binomial(link = "logit"), 
    data = mydata)

Deviance Residuals: 
    Min       1Q   Median       3Q      Max  
-1.7744  -1.0751   0.5313   1.0443   1.6281  

Coefficients:
            Estimate Std. Error z value Pr(>|z|)   
(Intercept)  -9.0785     6.8056  -1.334  0.18221   
x1           -0.6009     0.7231  -0.831  0.40601   
x2            2.9663     1.4785   2.006  0.04482 * 
x3            2.2680     0.7607   2.982  0.00287 **
x4           -1.8901     0.9936  -1.902  0.05713 . 
---
Signif. codes:  0 �e***�f 0.001 �e**�f 0.01 �e*�f 0.05 �e.�f 0.1 �e �f 1

(Dispersion parameter for binomial family taken to be 1)

    Null deviance: 126.07  on 94  degrees of freedom
Residual deviance: 109.64  on 90  degrees of freedom
AIC: 119.64

Number of Fisher Scoring iterations: 5

> library(mfx)  
> logitor(y ~ x1 + x2 + x3 + x4, data=mydata)
Call:
logitor(formula = y ~ x1 + x2 + x3 + x4, data = mydata)

Odds Ratio:
   OddsRatio Std. Err.       z    P>|z|   
x1   0.54832   0.39652 -0.8309 0.406008   
x2  19.42065  28.71261  2.0064 0.044817 * 
x3   9.65999   7.34799  2.9816 0.002867 **
x4   0.15105   0.15009 -1.9023 0.057133 . 
---
Signif. codes:  0 �e***�f 0.001 �e**�f 0.01 �e*�f 0.05 �e.�f 0.1 �e �f 1

># Classification
> y_pred_prob <- predict(logit, data=mydata, type="response")
> y_pred <-I(y_pred_prob>0.5)
> sum(y*y_pred)/sum(y)
[1] 0.8474576
> sum((1-y)*(1-y_pred))/sum(1-y)
[1] 0.3888889
> sum(y*y_pred+(1-y)*(1-y_pred))/95
[1] 0.6736842


> # marginal effects
> logitmfx(y ~ x1 + x2 + x3 + x4, data=mydata)
Call:
logitmfx(formula = y ~ x1 + x2 + x3 + x4, data = mydata)

Marginal Effects:
       dF/dx Std. Err.       z     P>|z|    
x1 -0.141848  0.177310 -0.8000  0.423710    
x2  0.378944  0.075144  5.0429 4.586e-07 ***
x3  0.504408  0.162596  3.1022  0.001921 ** 
x4 -0.420366  0.216248 -1.9439  0.051906 .  
---
Signif. codes:  0 �e***�f 0.001 �e**�f 0.01 �e*�f 0.05 �e.�f 0.1 �e �f 1

dF/dx is for discrete change for the following variables:

[1] "x1" "x2"
> 

> probit <- glm(y ~ x1 + x2 + x3 + x4, family=binomial(link="probit"), data=mydata)
> summary(probit)

Call:
glm(formula = y ~ x1 + x2 + x3 + x4, family = binomial(link = "probit"), 
    data = mydata)

Deviance Residuals: 
    Min       1Q   Median       3Q      Max  
-1.7734  -1.0785   0.5198   1.0457   1.6191  

Coefficients:
            Estimate Std. Error z value Pr(>|z|)   
(Intercept)  -5.3625     4.0691  -1.318  0.18756   
x1           -0.3755     0.4401  -0.853  0.39344   
x2            1.7760     0.7984   2.224  0.02612 * 
x3            1.3692     0.4433   3.089  0.00201 **
x4           -1.1579     0.5833  -1.985  0.04712 * 
---
Signif. codes:  0 �e***�f 0.001 �e**�f 0.01 �e*�f 0.05 �e.�f 0.1 �e �f 1

(Dispersion parameter for binomial family taken to be 1)

    Null deviance: 126.07  on 94  degrees of freedom
Residual deviance: 109.38  on 90  degrees of freedom
AIC: 119.38

Number of Fisher Scoring iterations: 7

> y_pred_prob <- predict(probit, data=mydata, type="response")
> y_pred <-I(y_pred_prob>0.5)
> sum(y*y_pred)/sum(y)
[1] 0.8474576
> sum((1-y)*(1-y_pred))/sum(1-y)
[1] 0.3888889
> sum(y*y_pred+(1-y)*(1-y_pred))/95
[1] 0.6736842

> # marginal effects
> probitmfx(y ~ x1 + x2 + x3 + x4, data=mydata)
Call:
probitmfx(formula = y ~ x1 + x2 + x3 + x4, data = mydata)

Marginal Effects:
       dF/dx Std. Err.       z    P>|z|    
x1 -0.143777  0.173202 -0.8301 0.406478    
x2  0.393362  0.071022  5.5386 3.05e-08 ***
x3  0.501062  0.158507  3.1611 0.001572 ** 
x4 -0.423731  0.210724 -2.0108 0.044343 *  
---
Signif. codes:  0 �e***�f 0.001 �e**�f 0.01 �e*�f 0.05 �e.�f 0.1 �e �f 1

dF/dx is for discrete change for the following variables:

[1] "x1" "x2"
>