R: Create, or interact variables with, factors

i {fixest}

R Documentation

Create, or interact variables with, factors

Description

Treat a variable as a factor, or interacts a variable with a factor. Values to be dropped/kept from the factor can be easily set. Note that to interact fixed-effects, this function should not be used: instead use directly the syntax fe1^fe2.

Usage

i(factor_var, var, ref, keep, ref2, keep2, ...)

Arguments

`factor_var`	A vector (of any type) that will be treated as a factor. You can set references (i.e. exclude values for which to create dummies) with the `ref` argument.
`var`	A variable of the same length as `factor_var`. This variable will be interacted with the factor in `factor_var`. It can be numeric or factor-like. To force a numeric variable to be treated as a factor, you can add the `i.` prefix to a variable name. For instance take a numeric variable `x_num`: `i(x_fact, x_num)` will treat `x_num` as numeric while `i(x_fact, i.x_num)` will treat `x_num` as a factor (it's a shortcut to `as.factor(x_num)`).
`ref`	A vector of values to be taken as references from `factor_var`. Can also be a logical: if `TRUE`, then the first value of `factor_var` will be removed. If `ref` is a character vector, partial matching is applied to values; use "@" as the first character to enable regular expression matching. See examples.
`keep`	A vector of values to be kept from `factor_var` (all others are dropped). By default they should be values from `factor_var` and if `keep` is a character vector partial matching is applied. Use "@" as the first character to enable regular expression matching instead.
`ref2`	A vector of values to be dropped from `var`. By default they should be values from `var` and if `ref2` is a character vector partial matching is applied. Use "@" as the first character to enable regular expression matching instead.
`keep2`	A vector of values to be kept from `var` (all others are dropped). By default they should be values from `var` and if `keep2` is a character vector partial matching is applied. Use "@" as the first character to enable regular expression matching instead.
`...`	Not currently used.

Details

To interact fixed-effects, this function should not be used: instead use directly the syntax fe1^fe2 in the fixed-effects part of the formula. Please see the details and examples in the help page of feols.

Value

It returns a matrix with number of rows the length of factor_var. If there is no interacted variable or it is interacted with a numeric variable, the number of columns is equal to the number of cases contained in factor_var minus the reference(s). If the interacted variable is a factor, the number of columns is the number of combined cases between factor_var and var.

Shorthand in `fixest` estimations

In fixest estimations, instead of using i(factor_var, var, ref), you can instead use the following writing var::factor_var(ref). Note that this way of doing interactions is deprecated and will be removed in the future.

Author(s)

Laurent Berge

Examples


#
# Simple illustration
#

x = rep(letters[1:4], 3)[1:10]
y = rep(1:4, c(1, 2, 3, 4))

# interaction
data.frame(x, y, i(x, y, ref = TRUE))

# without interaction
data.frame(x, i(x, "b"))

# you can interact factors too
z = rep(c("e", "f", "g"), c(5, 3, 2))
data.frame(x, z, i(x, z))

# to force a numeric variable to be treated as a factor: use i.
data.frame(x, y, i(x, i.y))

#
# In fixest estimations
#

data(base_did)
# We interact the variable 'period' with the variable 'treat'
est_did = feols(y ~ x1 + i(period, treat, 5) | id + period, base_did)

# => plot only interactions with iplot
iplot(est_did)

# Using i() for factors
est_bis = feols(y ~ x1 + i(period, keep = 3:6) + i(period, treat, 5) | id, base_did)

# we plot the second set of variables created with i()
# => we need to use keep (otherwise only the first one is represented)
iplot(est_bis, keep = "trea")

# => special treatment in etable
etable(est_bis, dict = c("6" = "six"))

#
# Interact two factors
#

# We use the i. prefix to consider week as a factor
data(airquality)
aq = airquality
aq$week = aq$Day %/% 7 + 1

# Interacting Month and week:
res_2F = feols(Ozone ~ Solar.R + i(Month, i.week), aq)

# Same but dropping the 5th Month and 1st week
res_2F_bis = feols(Ozone ~ Solar.R + i(Month, i.week, ref = 5, ref2 = 1), aq)

etable(res_2F, res_2F_bis)

[Package fixest version 0.9.0 Index]