evaluateLags: Create lag variables and evaluate models with different...

In JWiley/multilevelTools: Multilevel and Mixed Effects Model Diagnostics and Effect Sizes

Create lag variables and evaluate models with different number of lags

Description

This function creates the desired number of lags and tests consecutive models from a model with no lags (lag 0), lag 0 + lag1, etc. and reports model performance. This helps evaluate how many lags are needed.

Usage

evaluateLags(formula, lagvar, nlags = 0L, idvar, data, ...)

Arguments

formula	A character string giving the lmer() formula to use as a base. The variable to be tested with lags gets added as fixed effects only to this, currently.
lagvar	A character string giving the name of the variable to test lags for.
nlags	An integer (e.g., 0L, 3L) giving the number of lags to test. Defaults to 0L but really should be more. Must be a positive integer.
idvar	A character string giving the name o the ID variable.
data	A data.table dataset ideally or at least a data.frame.
...	Additional arguments passed to lmer, used to control model fitting.

Details

Currently only linear mixed effects models are allowed.

Examples

## these examples are slow to run
data(aces_daily, package = "JWileymisc")

evaluateLags(
 "NegAff ~ Female + Age + BornAUS + (1 | UserID)",
 "STRESS",
 4L,
 "UserID",
 aces_daily)


 ## not run, more complex example with random slope, fails to converge
evaluateLags(
 "NegAff ~ Female + Age + BornAUS + (1 + STRESS | UserID)",
 "STRESS",
 5L,
 "UserID",
 aces_daily)

## use different control to fit model and now converges
strictControl <- lme4::lmerControl(optCtrl = list(
   algorithm = "NLOPT_LN_NELDERMEAD",
   xtol_abs = 1e-10,
   ftol_abs = 1e-10))
evaluateLags(
 "NegAff ~ Female + Age + BornAUS + (1 + STRESS | UserID)",
 "STRESS",
 5L,
 "UserID",
 aces_daily,
control = strictControl)

JWiley/multilevelTools documentation built on April 1, 2024, 9:56 p.m.