I am replicating a logit model example from Econometrics book from Gujarati and Porter (Spanish edition). I have no problems with the model estimation, but I can't replicate marginal effects. In book, regression results are the following:
In the following lines I show my regression results:
> dat <- data.frame(
+ debito = c(0, 1, 0, 0, 0, 0, 0, 1, 0, 0, 1, 1, 0, 1, 0, 1, 0, 0, 0, 1,
+ 1, 0, 0, 0, 1, 0, 1, 1, 0, 1, 1, 1, 0, 1, 1, 0, 0, 0, 0, 1,
+ 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 0, 0, 1, 1, 1, 0, 1),
+ saldo = c(1756, 748, 1501, 1831, 1622, 1886, 740, 1593, 1169, 2125,
+ 1554, 1474, 1913, 1218, 1006, 2215, 137, 167, 343, 2557,
+ 2276, 1494, 2144, 1995, 1053, 1526, 1120, 1838, 1746, 1616,
+ 1958, 634, 580, 1320, 1675, 789, 1735, 1784, 1326, 2051, 1044,
+ 1885, 1790, 765, 1645, 32, 1266, 890, 2204, 2409, 1338, 2076,
+ 1708, 2138, 2375, 1455, 1487, 1125, 1989, 2156),
+ cajero = c(13, 9, 10, 10, 14, 17, 6, 10, 6, 18, 12, 12, 6, 10, 12, 20,
+ 7, 5, 7, 20, 15, 11, 17, 10, 8, 8, 8, 7, 11, 10, 6, 2, 4, 4,
+ 6, 8, 12, 11, 16, 14, 7, 10, 11, 4, 6, 2, 11, 7, 14, 16, 14,
+ 12, 13, 18, 12, 9, 8, 6, 12, 14),
+ interes = c(1, 0, 0, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
+ 0, 1, 0, 0, 0, 1, 0, 1, 0, 1, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0,
+ 0, 1, 1, 0, 1, 0, 0, 1, 0, 0, 0, 0, 0, 1, 0, 1, 0, 0, 1, 0)
+ )
> mod <- glm(debito ~ saldo + cajero + interes,data = dat,
+ family = "binomial")
> summary(mod)
Call:
glm(formula = debito ~ saldo + cajero + interes, family = "binomial",
data = dat)
Deviance Residuals:
Min 1Q Median 3Q Max
-1.6619 -0.9712 -0.6629 1.1440 1.7076
Coefficients:
Estimate Std. Error z value Pr(>|z|)
(Intercept) -0.5749004 0.7857866 -0.732 0.4644
saldo 0.0012481 0.0006973 1.790 0.0735 .
cajero -0.1202251 0.0939842 -1.279 0.2008
interes -1.3520856 0.6809872 -1.985 0.0471 *
---
Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
(Dispersion parameter for binomial family taken to be 1)
Null deviance: 82.108 on 59 degrees of freedom
Residual deviance: 75.500 on 56 degrees of freedom
AIC: 83.5
Number of Fisher Scoring iterations: 4
As you can see, I obtained the same results. My problem is when I try to replicate marginal effects, which in book are the following:
I try to replicate thus results with the following method:
> coef(mod) * mean(dlogis(predict(mod, type = "link")))
(Intercept) saldo cajero interes
-0.1262098468 0.0002740024 -0.0263934252 -0.2968279711
As you can see, I failed. I know results in book are obtained from Stata, so I don't know the details of calculation or if there is posible to replicate the calculation in R. There exist a way?
Use the package "margins" for marginal effects.
install.packages("margins")
library("margins")
mod <- glm(debito ~ saldo + cajero + interes, data = dat, family = "binomial")
(m <- margins(mod))
Here are the outputs:
Average marginal effects
glm(formula = debito ~ saldo + cajero + interes, family = "binomial", data = dat)
saldo cajero interes
0.000274 -0.02639 -0.2968
Are your column names correct?
The technical post webpages of this site follow the CC BY-SA 4.0 protocol. If you need to reprint, please indicate the site URL or the original address.Any question please contact:yoyou2525@163.com.