power4mome (version 0.1.1)

fit_model: Fit a Model to a List of Datasets

Description

Get the output of sim_data() and fit a model to each of the stored datasets.

Usage

fit_model(
  data_all = NULL,
  model = NULL,
  fit_function = "lavaan",
  arg_data_name = "data",
  arg_model_name = "model",
  arg_group_name = "group",
  ...,
  fit_out = NULL,
  parallel = FALSE,
  progress = FALSE,
  ncores = max(1, parallel::detectCores(logical = FALSE) - 1)
)

Value

An object of the class fit_out, which is a list of the output of fit_function (lavaan::sem()

by default). If an error occurred when fitting the model to a dataset, then this element will be the error message from the fit function.

Arguments

data_all

The output of sim_data(), or a sim_data class object.

model

The model to be fitted. If NULL, the default, the model stored in data_all, which should be the data generation model, will be used.

fit_function

The function to be used to fit the model. Can also be a string: "lavaan" (the default) for lavaan::sem(), and "lm" or many_lm for lmhelprs::many_lm(). Other functions can also be used if necessary.

arg_data_name

The name of the argument of fit_function expecting the dataset. Default is "data".

arg_model_name

The name of the argument of fit_function expecting the model definition. Default is "model".

arg_group_name

The name of the argument of fit_function expecting the name of the group variable. Used only for multigroup models. Default is "group".

...

Optional arguments to be passed to fit_function when fitting the model.

fit_out

If set to a fit_out object (the output of fit_model()), then all missing arguments will be retrieved from fit_out. That is, users can use fit_model(data_all = new_data, fit_out = old_out) to re-fit a model originally fitted in old_out on a new list of dataset (new_data). No need to include all other arguments.

parallel

If TRUE, parallel processing will be used to fit the models. Default is FALSE.

progress

If TRUE, the progress of model fitting will be displayed. Default is `FALSE.

ncores

The number of CPU cores to use if parallel processing is used.

The role of <code>fit_model()</code>

This function is used by the all-in-one function power4test(). Users usually do not call this function directly, though developers can use this function to customize the model fitting step in power analysis.

Details

By default, the function fit_model()

• extracts the model stored in the output of sim_data(),

• fits the model to each dataset simulated using fit_function, default to "lavaan" and lavaan::sem() will be called,

• and returns the results.

If the datasets were generated from a multigroup model when calling sim_data(), a multigroup model is fitted.

See Also

See power4test() for the all-in-one function that uses this function, and sim_data() for the function generating datasets for this function.

Examples

# Specify the population model

mod <-
"m ~ x
 y ~ m + x"

# Specify the effect sizes (population parameter values)

es <-
"
y ~ m: m
m ~ x: m
y ~ x: n
"

# Generate several simulated datasets

data_all <- sim_data(nrep = 5,
                     model = mod,
                     pop_es = es,
                     n = 100,
                     iseed = 1234)

# Fit the population model to each datasets

fit_all <- fit_model(data_all)
fit_all[[1]]

# Fit the population model using the MLR estimator

fit_all_mlr <- fit_model(data_all,
                         estimator = "MLR")
fit_all_mlr[[1]]

# Fit a model different from the population model,
# with the MLR estimator

mod2 <-
"m ~ x
 y ~ m"

fit_all_mlr2 <- fit_model(data_all,
                          mod2,
                          estimator = "MLR")
fit_all_mlr2[[1]]