MedSurvey (version 1.0)

chisq.BRR: To adjust model fit statistics

Description

This function is used to adjust model fit statistics for complex surveys with balanced repeated replications (Oberski, 2014; Satorra & Muthen, 1995) .

Usage

chisq.BRR(model, lavaan.fit, data, mwgtname, repwgtnames,
  fayfactor = 0.5, estimator = c("ML"), test = c("satorra.bentler"))

Arguments

model

The model being fitted. It is written in lavaan model syntax (Rosseel, 2012).

lavaan.fit

The model fit results using 'ML' estimator with sample main weights, but without adjusting the fit statistics or standard errors for complex surveys. Note that it is a lavaan object.

data

The raw data including the variables of interest and the survey weights. It should be a dataset or dataframe.

mwgtname

The variable name indicating the sample main weight in the dataset. See balanced repeated replications method (Wolter, 2007) for more information about the main weight.

repwgtnames

The variable names indicating the set of replicate weights in the dataset. See balanced repeated replications method (Wolter, 2007) for more information about the replicate weight.

fayfactor

The fayfactor used in the standard error calculation by fay's method (Fay & Train, 1995; Judkins, 1990) for balanced repeated replications. Fayfactor is a value between 0 and 1. The default is 0.5.

estimator

The method used to estimate the model. 'ML' is the default option and the only available option for current version. It is not required.

test

The method used to generate adjusted standard errors. 'satorra.bentler' is the default option and the only available option for current version. It is not required.

Value

The model fit results as a lavaan object (Rosseel, 2012) with the adjusted model fit statistics.

References

Fay, R. E., & Train, G. F. (1995). Aspects of survey and model-based postcensal estimation of income and poverty characteristics for states and counties. In Proceedings of the Section on Government Statistics, American Statistical Association, Alexandria, VA (pp. 154-159).

Judkins, D. R. (1990). Fay<U+2019>s method for variance estimation.Journal of Official Statistics,6(3), 223-239.

Oberski, D. (2014). lavaan. survey: An R package for complex survey analysis of structural equation models. Journal of Statistical Software, 57(1), 1-27. DOI:10.18637/jss.v057.i01

Rosseel, Y. (2012). Lavaan: An R package for structural equation modeling and more. Version 0.5<U+2013>12 (BETA). Journal of statistical software, 48(2), 1-36. DOI:10.18637/jss.v048.i02

Satorra, A., & Muthen, B. (1995). Complex sample data in structural equation modeling. Sociological methodology, 25(1), 267-316.

Wolter, K. (2007). Introduction to variance estimation. New York, NY: Springer.

Examples

# NOT RUN {
R <- 160
wgtnames <- paste("repwgt", seq(0,R,by=1), sep="")
mwgtname=wgtnames[1]
repwgtnames=wgtnames[2:(R+1)]
fayfactor=0.5

############ Example 1
model1 <- ' # outcome
             numcg ~ u2*1 + c*workban + b*sp_adltban
           # mediator
             sp_adltban ~ u1*1 + a*workban
           # indirect effect (a*b)
             ab := a*b
           # total effect
             total := c + (a*b)
          '
fit <- lavaan::sem(model=model1, data=MedData, estimator='ML', test='standard')
chisq.BRR(model1,fit,MedData,mwgtname, repwgtnames)
#
# lavaan 0.6-3 ended normally after 24 iterations
#
# Optimization method                           NLMINB
# Number of free parameters                          7
#
# Number of observations                          3922
#
# Estimator                                         ML      Robust
# Model Fit Test Statistic                       0.000       0.000
# Degrees of freedom                                 0           0
# Scaling correction factor                                     NA
# for the Satorra-Bentler correction#'

############ Example 2

model3 <- ' # outcome
            numcg ~ u0*1 + c*workban + b1*sp_adltban + b2*sp_kidsban
           # mediator
              sp_adltban ~ u1*1 + a1*workban
              sp_kidsban ~ u2*1 + a2*workban
           # indirect effect (a*b)
              a1b1 := a1*b1
             a2b2 := a2*b2
           # total effect
             total := c + (a1*b1) + (a2*b2)
          '

fit <- lavaan::sem(model=model3, data=MedData, estimator='ML', test='standard')
chisq.BRR(model3,fit,MedData,mwgtname, repwgtnames)

#       MedSurvey 1.0
#
# lavaan 0.6-3 ended normally after 27 iterations
#
# Optimization method                           NLMINB
# Number of free parameters                         11
#
# Number of observations                          3922
#
# Estimator                                         ML      Robust
# Model Fit Test Statistic                     305.248      70.973
# Degrees of freedom                                 1           1
# P-value (Chi-square)                           0.000       0.000
# Scaling correction factor                                  4.301
# for the Satorra-Bentler correction

# }
# NOT RUN {
# }

Run the code above in your browser using DataCamp Workspace