# Chapter 13 | Exercise 1

Using the data set `let`

from `Rling`

, fit an `mlogit`

model with *allow* as the reference level.

Which predictors are significant?

Compare the coefficients with the ones from the model `m.let`

. Are they similar?

```
> library(Rling); library(mlogit)
> data(let)
> let1 <- mlogit.data(let, shape = "wide", choice = "Verb")
> m.let.ex <- mlogit(Verb ~ 1 | Year + Reg + Permitter + Imper, data = let1, reflevel = 1)
> summary(m.let.ex)
```

Significant predictors:

```
- permit:Year, b = –0.12
- let:PermitterInanim, b = –2.6
- let:PermitterUndef, b = –1.27
- let:ImperYes, b = 3.08
```

the absolute values of the coefficients with *let* are the same, but the sign is opposite.