#' Two One-Sided Tests for Fixed Effects in Multilevel Models
#'
#' @param model Multilevel model object fitted with lmerTest::lmer.
#' @param fe Character vector. Name of the fixed effect variable.
#' @param bound_l Lower bound based on SESOI (smallest effect size of interest).
#' @param bound_u Upper bound based on SESOI (smallest effect size of interest).
#'
#' @return Vector of test statistics against lower and upper bound.
#' @export
#'
#' @examples
#' set.seed(234)
#' n1 <- 3 # groups
#' n2 <- 3 # observations per groups
#'
#' dat <- data.frame(
#' group = rep(c(LETTERS[1:n1]), each = n2),
#' x1 = sample(1:5, n1 * n2, replace = TRUE),
#' x2 = sample(1:5, n1 * n2, replace = TRUE),
#' x3 = sample(1:5, n1 * n2, replace = TRUE)
#' )
#'
#' tost_ml(
#' model = lmerTest::lmer(x1 ~ x2 + x3 + (1 | group),
#' data = dat, REML = FALSE
#' ),
#' fe = "x2", bound_l = -0.30, bound_u = 0.30
#' )
tost_ml <- function(model,
fe,
bound_l,
bound_u) {
# contrast vector
conts <- as.numeric(names(lme4::fixef(model)) == fe)
# two-sided test against lower bound
lower <- lmerTest::contest1D(
model = model, L = conts,
confint = TRUE,
rhs = bound_l
) # get t value for test against lower bound
# two-sided test against upper bound
upper <- lmerTest::contest1D(
model = model, L = conts,
confint = TRUE,
rhs = bound_u
) # get t value for test against upper bound
# obtain one-sided p-values
p.lower <- stats::pt(lower$`t value`, lower$df, lower.tail = FALSE)
p.upper <- stats::pt(upper$`t value`, upper$df, lower.tail = TRUE)
output <- cbind(
Estimate = lower$Estimate,
"Std. Error" = lower$"Std. Error",
df = lower$df,
t_lower = lower$"t value",
t_upper = upper$"t value",
p_lower = p.lower,
p_upper = p.upper
)
return(output)
}
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.