Skip to contents

ANOVA Table for Piecewise Linear Models

Source: R/anova.R
anova.sc_plm.Rd

Model comparison for piecewise regression models fitted with plm(), hplm(), or mplm() using likelihood ratio tests.

Usage

# S3 method for class 'sc_plm'
anova(object, ...)

# S3 method for class 'sc_hplm'
anova(object, ...)

# S3 method for class 'sc_mplm'
anova(object, ...)

Arguments

object

An object containing the results returned by plm(), hplm(), or mplm().

...

additional objects for model comparison.

Value

An object of class anova containing the results of the model comparison.

Details

The function performs likelihood ratio tests to compare nested piecewise regression models. It extracts the underlying model from the sc_plm, sc_hplm, or sc_mplm object and passes them to the generic anova() function for model comparison.

Examples

## For glm models with family = "gaussian"
mod1 <- plm(exampleAB$Johanna, level = FALSE, slope = FALSE)
mod2 <- plm(exampleAB$Johanna)
anova(mod1, mod2)
#> Analysis of Deviance Table
#> 
#> Model 1: values ~ 1 + mt
#> Model 2: values ~ 1 + mt + phaseB + interB
#>   Resid. Df Resid. Dev Df Deviance      F Pr(>F)
#> 1        18     449.64                          
#> 2        16     403.46  2   46.183 0.9157 0.4202
## For glm models with family = "poisson"
mod0 <- plm(example_A24, formula = injuries ~ 1, family = "poisson")
mod1 <- plm(example_A24, trend = FALSE, family = "poisson")
anova(mod0, mod1, mod2)
#> Warning: models with response ‘"values"’ removed because response differs from model 1
#> Analysis of Deviance Table
#> 
#> Model 1: injuries ~ 1
#> Model 2: injuries ~ 1 + phaseB + interB
#>   Resid. Df Resid. Dev Df Deviance  Pr(>Chi)    
#> 1        22     643.09                          
#> 2        20      95.78  2   547.31 < 2.2e-16 ***
#> ---
#> Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
## For glm with family = "binomial"
mod0 <- plm(
  exampleAB_score$Christiano, 
  formula = values ~ 1, 
  family = "binomial", 
  var_trials = "trials"
)
mod1 <- plm(
  exampleAB_score$Christiano, 
  trend = FALSE, 
  family = "binomial", 
  var_trials = "trials"
)
anova(mod0, mod1)
#> Analysis of Deviance Table
#> 
#> Model 1: values ~ 1
#> Model 2: values ~ 1 + phaseB + interB
#>   Resid. Df Resid. Dev Df Deviance  Pr(>Chi)    
#> 1        29    267.683                          
#> 2        27     27.124  2   240.56 < 2.2e-16 ***
#> ---
#> Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
## For multilevel models:
mod0 <- hplm(Leidig2018, trend = FALSE, slope = FALSE, level = FALSE)
mod1 <- hplm(Leidig2018, trend = FALSE)
mod2 <- hplm(Leidig2018)
anova(mod0, mod1, mod2)
#>   Model df      AIC      BIC    logLik   Test  L.Ratio p-value
#> 1     1  3 6255.509 6272.876 -3124.754                        
#> 2     2  5 5826.509 5855.454 -2908.254 1 vs 2 433.0002  <.0001
#> 3     3  6 5827.604 5862.339 -2907.802 2 vs 3   0.9044  0.3416
## For mplm
mod0 <- mplm(
  Leidig2018$`1a1`, 
  update = . ~  1, dvar = c("academic_engagement", "disruptive_behavior")
)
mod1 <- mplm(
  Leidig2018$`1a1`, 
  trend = FALSE, 
  dvar = c("academic_engagement", "disruptive_behavior")
)
mod2 <- mplm(
  Leidig2018$`1a1`, 
  dvar = c("academic_engagement", "disruptive_behavior")
)

anova(mod0, mod1, mod2)
#> Analysis of Variance Table
#> 
#> Model 1: y ~ 1
#> Model 2: y ~ 1 + phaseB + interB
#> Model 3: y ~ 1 + mt + phaseB + interB
#>   Res.Df Df Gen.var.  Pillai approx F num Df den Df    Pr(>F)    
#> 1     82     0.52962                                             
#> 2     80 -2  0.44785 0.33250   7.8764      4    158 8.167e-06 ***
#> 3     79 -1  0.44628 0.03166   1.2750      2     78    0.2852    
#> ---
#> Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1