piecewiseSEM (version 1.2.1)

sem.predict: Returns model predictions for SEM

Description

Returns predictions for responses in the list of structured equations.

Usage

sem.predict(object, newdata, sefit = FALSE, ...)

Arguments

object

a single model or list of regressions representing the structural equation model.

newdata

a data.frame of predictors used to generate the model predictions.

sefit

whether standard errors of predictions should be returned. Default is FALSE.

...

additional arguments passed to predict.

Value

Returns a data.frame containing the new data and the predicted responses based on fixed effects only for the innermost level of grouping (i.e., population residuals) corresponding to re.form = NA for lme4 models and level = Q for nlme models.

Details

Mixed model predictions includes only fixed effects and *not* random effects (default is level = 0 or re.form = 0). This can be changed by passing additional arguments from predict.lme and predict.merMod.

If sefit = TRUE for mixed models, then standard errors on predictions are estimated using fixed effects *only*. See explanation here: http://glmm.wikidot.com/faq.

See Also

predict, predict.lme, predict.merMod

Examples

# NOT RUN {
# Load example data
data(shipley2009)

# Load model packages
library(lme4)
library(nlme)

# Create list of models 
shipley2009.modlist = list(
  
  lme(DD ~ lat, random = ~1|site/tree, na.action = na.omit, 
      data = shipley2009),
  
  lme(Date ~ DD, random = ~1|site/tree, na.action = na.omit, 
      data = shipley2009),
  
  lme(Growth ~ Date, random = ~1|site/tree, na.action = na.omit, 
      data = shipley2009),
  
  glmer(Live ~ Growth+(1|site)+(1|tree), 
        family=binomial(link = "logit"), data = shipley2009) 
  
)

# Create new data for predictions
shipley2009.new = data.frame(
  DD = seq(min(shipley2009$DD, na.rm = TRUE), 
           max(shipley2009$DD, na.rm = TRUE), 
           by = 0.01)
)

# Generate predictions
shipley2009.new.pred = sem.predict(shipley2009.modlist, shipley2009.new)
head(shipley2009.new.pred)

# Plot predicted fit
plot(shipley2009$Date ~ shipley2009$DD, col = "grey60")
lines(shipley2009.new.pred$Date.fit ~ shipley2009.new.pred$DD, lwd = 2, col = "red")

# Generate predictions with standard errors (based on fixed effects only)
shipley2009.new.pred = sem.predict(shipley2009.modlist, shipley2009.new, sefit = TRUE)

# Add 95 percent confidence intervals
lines(shipley2009.new.pred$DD, 
      shipley2009.new.pred$Date.fit + 2 * shipley2009.new.pred$Date.se.fit, 
      lwd = 1.5, lty = 2, col = "red")

lines(shipley2009.new.pred$DD, 
      shipley2009.new.pred$Date.fit - 2 * shipley2009.new.pred$Date.se.fit,
      lwd = 1.5, lty = 2, col = "red")
# }