| Title: | Simulate Probabilistic Long-Term Effects in Models with Temporal Dependence |
|---|---|
| Description: | Calculates and depicts probabilistic long-term effects in binary models with temporal dependence variables. The package performs two tasks. First, it calculates the change in the probability of the event occurring given a change in a theoretical variable. Second, it calculates the rolling difference in the future probability of the event for two scenarios: one where the event occurred at a given time and one where the event does not occur. The package is consistent with the recent movement to depict meaningful and easy-to-interpret quantities of interest with the requisite measures of uncertainty. It is the first to make it easy for researchers to interpret short- and long-term effects of explanatory variables in binary autoregressive models, which can have important implications for the correct interpretation of these models. |
| Authors: | Christopher Gandrud [aut, cre], Laron K. Williams [aut] |
| Maintainer: | Christopher Gandrud <[email protected]> |
| License: | GPL (>= 3) |
| Version: | 1.0 |
| Built: | 2024-11-20 05:05:45 UTC |
| Source: | https://github.com/christophergandrud/pltesim |
Generate spells for binary variables
btscs(df, event, t_var, cs_unit, pad_ts = FALSE)btscs(df, event, t_var, cs_unit, pad_ts = FALSE)
df |
a data frame |
event |
character string naming the binary variable identifying the
event. Note |
t_var |
character string with the name of the time variable. |
cs_unit |
character string with the name of the cross-sectional unit. |
pad_ts |
logical indicating whether or not to fill in the time-series if panels are unbalanced. |
The original (df) data frame with an additional
spell_time value identifying the number of observed periods in the spell,
i.e. time points since the last period.
This function is a port of Dave Armstrong's btscs function
from:
Dave Armstrong (2015). DAMisc: Dave Armstrong's Miscellaneous Functions. R package version 1.3. https://CRAN.R-project.org/package=DAMisc.
It was ported largely to reduce the dependencies needed for the examples. There are also internal improvements, largely to handle single period spells and to start the spell time counter from 1.
David's package implemented the Stata function from:
Beck, N.. J. Katz and R. Tucker. 1998. "Beyond Ordinary Logit: Taking Time Seriously in Binary-Time-Series-Cross-Section Models". American Journal of Political Science 42(4): 1260-1288.
data('negative') neg_set <- btscs(df = negative, event = 'y', t_var = 'tim', cs_unit = 'group')data('negative') neg_set <- btscs(df = negative, event = 'y', t_var = 'tim', cs_unit = 'group')
Function for finding predicted probability from the systematic linear component of a logistic regression.
logistic_prob_FUN(x)logistic_prob_FUN(x)
x |
a numeric vector of the systematic linear component from a logistic regression model. |
Simulated data set from Williams (2016) to illustrate negative duration dependence
A data set with 1000 observations and 6 variables
Williams, Laron K. 2016. "Long-Term Effects in Models with Temporal Dependence." Political Analysis. 24: 243-62
Simulated data set based on Williams (2016) to illustrate negative duration dependence in examples
A data set with 1000 observations and 4 variables
Create simulations for long-term effects in models with temporal dependence
plte_builder(obj, obj_tvar, cf, cf_duration = "permanent", t_points, FUN = logistic_prob_FUN, ci = 0.95, nsim = 1000)plte_builder(obj, obj_tvar, cf, cf_duration = "permanent", t_points, FUN = logistic_prob_FUN, ci = 0.95, nsim = 1000)
obj |
a fitted model object. |
obj_tvar |
character string specifying the name of the base time variable
in |
cf |
a data frame with the first row containing the counterfactual.
An optional second row could be supplied with values for the baseline
scenario. If not supplied then all values are set to zero for the baseline.
Columns should have names that match variables in |
cf_duration |
a character string or numeric specifying the
counterfactual's duration. If |
t_points |
a numeric vector with a minimum length of 2 and a maximum lentgh of 3. The first and last values should be the time starting and ending points for the simulatinos. The (optional) middle value can specify a point between the first and last time points where a subsequent event occurs. |
FUN |
a function for finding a quantity of interest from the linear
systematic component. See |
ci |
the proportion of the central interval of the simulations to return. Must be in (0, 1] or equivalently (0, 100]. |
nsim |
number of simulations to draw. |
A data frame with the medians and central intervals of the
simulated scenarios. Note that the column scenario_name
encodes scenarios where y = 0 as baseline and y = 1 as
counterfactual.
Williams, Laron K. 2016. "Long-Term Effects in Models with Temporal Dependence". Political Analysis: 24(2): 243-262.
data('negative') # BTSCS set the data neg_set <- btscs(df = negative, event = 'y', t_var = 'tim', cs_unit = 'group', pad_ts = FALSE) # Create temporal dependence variable neg_set$t <- neg_set$spell + 1 m1 <- glm(y ~ x + t + I(t^2) + I(t^3), family = binomial(link = 'logit'), data = neg_set) # Create fitted counterfactual counterfactual <- data.frame(x = 0.5) # Permanent counterfactual, one event sim1 <- plte_builder(obj = m1, obj_tvar = 't', cf = counterfactual, t_points = c(13, 25)) # Multiple events sim2 <- plte_builder(obj = m1, obj_tvar = 't', cf = counterfactual, t_points = c(13, 18, 25)) # One-time counterfactual sim3 <- plte_builder(obj = m1, obj_tvar = 't', cf = counterfactual, t_points = c(13, 25), cf_duration = 'one-time') # Temporary (4 period counterfactual) sim4 <- plte_builder(obj = m1, obj_tvar = 't', cf = counterfactual, t_points = c(13, 25), cf_duration = 4) # Custom baseline scenario # Note: the second row is the custom baseline counterfactual_baseline <- data.frame(x = c(1, 0.5)) sim5 <- plte_builder(obj = m1, obj_tvar = 't', cf_duration = 4, cf = counterfactual_baseline, t_points = c(13, 25)) # Time splines library(splines) m2 <- glm(y ~ x + bs(t, degree = 3), family = binomial(link = 'logit'), data = neg_set) sim6 <- plte_builder(obj = m2, obj_tvar = 't', cf_duration = 4, cf = counterfactual, t_points = c(13, 25))data('negative') # BTSCS set the data neg_set <- btscs(df = negative, event = 'y', t_var = 'tim', cs_unit = 'group', pad_ts = FALSE) # Create temporal dependence variable neg_set$t <- neg_set$spell + 1 m1 <- glm(y ~ x + t + I(t^2) + I(t^3), family = binomial(link = 'logit'), data = neg_set) # Create fitted counterfactual counterfactual <- data.frame(x = 0.5) # Permanent counterfactual, one event sim1 <- plte_builder(obj = m1, obj_tvar = 't', cf = counterfactual, t_points = c(13, 25)) # Multiple events sim2 <- plte_builder(obj = m1, obj_tvar = 't', cf = counterfactual, t_points = c(13, 18, 25)) # One-time counterfactual sim3 <- plte_builder(obj = m1, obj_tvar = 't', cf = counterfactual, t_points = c(13, 25), cf_duration = 'one-time') # Temporary (4 period counterfactual) sim4 <- plte_builder(obj = m1, obj_tvar = 't', cf = counterfactual, t_points = c(13, 25), cf_duration = 4) # Custom baseline scenario # Note: the second row is the custom baseline counterfactual_baseline <- data.frame(x = c(1, 0.5)) sim5 <- plte_builder(obj = m1, obj_tvar = 't', cf_duration = 4, cf = counterfactual_baseline, t_points = c(13, 25)) # Time splines library(splines) m2 <- glm(y ~ x + bs(t, degree = 3), family = binomial(link = 'logit'), data = neg_set) sim6 <- plte_builder(obj = m2, obj_tvar = 't', cf_duration = 4, cf = counterfactual, t_points = c(13, 25))
Plot objects created by plte_builder
plte_plot(obj, t_labels = TRUE)plte_plot(obj, t_labels = TRUE)
obj |
a |
t_labels |
logical whether or not to include time labels for each point. |
A gg ggplot2 object that can be modified using the
+ in combination with other ggplot2 functions.
data('negative') # BTSCS set the data neg_set <- btscs(df = negative, event = 'y', t_var = 'tim', cs_unit = 'group', pad_ts = FALSE) # Create temporal dependence variables neg_set$t <- neg_set$spell + 1 m1 <- glm(y ~ x + t + I(t^2) + I(t^3), family = binomial(link = 'logit'), data = neg_set) counterfactual <- data.frame(x = 0.5) sim1 <- plte_builder(obj = m1, obj_tvar = 't', cf = counterfactual, t_points = c(13, 25), cf_duration = 4, ci = 99) # With time point labels plte_plot(sim1) # Without time point labels plte_plot(sim1, t_labels = FALSE)data('negative') # BTSCS set the data neg_set <- btscs(df = negative, event = 'y', t_var = 'tim', cs_unit = 'group', pad_ts = FALSE) # Create temporal dependence variables neg_set$t <- neg_set$spell + 1 m1 <- glm(y ~ x + t + I(t^2) + I(t^3), family = binomial(link = 'logit'), data = neg_set) counterfactual <- data.frame(x = 0.5) sim1 <- plte_builder(obj = m1, obj_tvar = 't', cf = counterfactual, t_points = c(13, 25), cf_duration = 4, ci = 99) # With time point labels plte_plot(sim1) # Without time point labels plte_plot(sim1, t_labels = FALSE)