simsem (version 0.2-8)

plotPowerFitNonNested: Plot power of rejecting a non-nested model based on a difference in fit index

Description

Plot the proportion of the difference in fit indices from one model that does not in the range of sampling distribution from another model (reject that the dataset comes from the second model) or indicates worse fit than a specified cutoff. This plot can show the proportion in the second model that does not in the range of sampling distribution from the first model too.

Usage

plotPowerFitNonNested(dat2Mod1, dat2Mod2, dat1Mod1=NULL, dat1Mod2=NULL, 
cutoff = NULL, usedFit = NULL, alpha = 0.05, contN = TRUE, contMCAR = TRUE, 
contMAR = TRUE, useContour = TRUE, logistic = TRUE, onetailed = FALSE)

Arguments

dat2Mod1

SimResult that saves the simulation of analyzing Model 1 by datasets created from Model 2

dat2Mod2

SimResult that saves the simulation of analyzing Model 2 by datasets created from Model 2

dat1Mod1

SimResult that saves the simulation of analyzing Model 1 by datasets created from Model 1

dat1Mod2

SimResult that saves the simulation of analyzing Model 2 by datasets created from Model 1

cutoff

A vector of priori cutoffs for the differences in fit indices.

usedFit

Vector of names of fit indices that researchers wish to plot.

alpha

A priori alpha level

contN

Include the varying sample size in the power plot if available

contMCAR

Include the varying MCAR (missing completely at random percentage) in the power plot if available

contMAR

Include the varying MAR (missing at random percentage) in the power plot if available

useContour

If there are two of sample size, percent completely at random, and percent missing at random are varying, the plotCutoff function will provide 3D graph. Contour graph is a default. However, if this is specified as FALSE, perspect

logistic

If logistic is TRUE and the varying parameter exists (e.g., sample size or percent missing), the plot based on logistic regression predicting the significance by the varying parameters is preferred. If FALSE, the ove

onetailed

If TRUE, the function will use the cutoff from one-tail test. If FALSE, the funciton will use the cutoff from two-tailed test.

Value

• NONE. Only plot the fit indices distributions.

See Also

• SimResultfor simResult that used in this function.
• getCutoffNonNestedto find the cutoffs of the differences in fit indices for non-nested model comparison
• plotCutoffNonNestedto visualize the cutoffs of the differences in fit indices for non-nested model comparison
• getPowerFitNonNestedto find the power in rejecting the non-nested model by the difference in fit indices cutoffs

Examples

n1 <- simNorm(0, 0.1)
u79 <- simUnif(0.7, 0.9)

loading.A <- matrix(0, 8, 2)
loading.A[1:3, 1] <- NA
loading.A[4:8, 2] <- NA
LX.A <- simMatrix(loading.A, 0.7)
latent.cor <- matrix(NA, 2, 2)
diag(latent.cor) <- 1
RPH <- symMatrix(latent.cor, "u79")
RTD <- symMatrix(diag(8))
CFA.Model.A <- simSetCFA(LY = LX.A, RPS = RPH, RTE = RTD)

error.cor.mis <- matrix(NA, 8, 8)
diag(error.cor.mis) <- 1
RTD.Mis <- symMatrix(error.cor.mis, "n1")
CFA.Model.A.Mis <- simMisspecCFA(RTE = RTD.Mis)

loading.B <- matrix(0, 8, 2)
loading.B[1:4, 1] <- NA
loading.B[5:8, 2] <- NA
LX.B <- simMatrix(loading.B, 0.7)
CFA.Model.B <- simSetCFA(LY = LX.B, RPS = RPH, RTE = RTD)

SimData.A <- simData(CFA.Model.A, 500)
SimData.B <- simData(CFA.Model.B, 500)

SimModel.A <- simModel(CFA.Model.A)
SimModel.B <- simModel(CFA.Model.B)

# The actual number of replications should be greater than 10.
Output.A.A <- simResult(10, SimData.A, SimModel.A)
Output.A.B <- simResult(10, SimData.A, SimModel.B)
Output.B.A <- simResult(10, SimData.B, SimModel.A)
Output.B.B <- simResult(10, SimData.B, SimModel.B)

plotPowerFitNonNested(Output.B.A, Output.B.B, dat1Mod1=Output.A.A, dat1Mod2=Output.A.B)
plotPowerFitNonNested(Output.B.A, Output.B.B, cutoff=c(AIC=0, BIC=0))