testOMF {cSEM} | R Documentation |

## Test for overall model fit

### Description

### Usage

```
testOMF(
.object = NULL,
.alpha = 0.05,
.fit_measures = FALSE,
.handle_inadmissibles = c("drop", "ignore", "replace"),
.R = 499,
.saturated = FALSE,
.seed = NULL,
...
)
```

### Arguments

`.object` |
An R object of class cSEMResults resulting from a call to |

`.alpha` |
An integer or a numeric vector of significance levels.
Defaults to |

`.fit_measures` |
Logical. (EXPERIMENTAL) Should additional fit measures
be included? Defaults to |

`.handle_inadmissibles` |
Character string. How should inadmissible results
be treated? One of " |

`.R` |
Integer. The number of bootstrap replications. Defaults to |

`.saturated` |
Logical. Should a saturated structural model be used?
Defaults to |

`.seed` |
Integer or |

`...` |
Can be used to determine the fitting function used in the calculateGFI function. |

### Details

Bootstrap-based test for overall model fit originally proposed by Beran and Srivastava (1985). See also Dijkstra and Henseler (2015) who first suggested the test in the context of PLS-PM.

By default, `testOMF()`

tests the null hypothesis that the population indicator
correlation matrix equals the population model-implied indicator correlation matrix.
Several discrepancy measures may be used. By default, `testOMF()`

uses four distance
measures to assess the distance between the sample indicator correlation matrix
and the estimated model-implied indicator correlation matrix, namely the geodesic distance,
the squared Euclidean distance, the standardized root mean square residual (SRMR),
and the distance based on the maximum likelihood fit function.
The reference distribution for each test statistic is obtained by
the bootstrap as proposed by Beran and Srivastava (1985).

It is possible to perform the bootstrap-based test using fit measures such
as the CFI, RMSEA or the GFI if `.fit_measures = TRUE`

. This is experimental.
To the best of our knowledge the applicability and usefulness of the fit
measures for model fit assessment have not been formally (statistically)
assessed yet. Theoretically, the logic of the test applies to these fit indices as well.
Hence, their applicability is theoretically justified.
Only use if you know what you are doing.

If `.saturated = TRUE`

the original structural model is ignored and replaced by
a saturated model, i.e., a model in which all constructs are allowed to correlate freely.
This is useful to test misspecification of the measurement model in isolation.

### Value

A list of class `cSEMTestOMF`

containing the following list elements:

`$Test_statistic`

The value of the test statistics.

`$Critical_value`

The corresponding critical values obtained by the bootstrap.

`$Decision`

The test decision. One of:

`FALSE`

(**Reject**) or`TRUE`

(**Do not reject**).`$Information`

The

`.R`

bootstrap values; The number of admissible results; The seed used and the number of total runs.

### References

Beran R, Srivastava MS (1985).
“Bootstrap Tests and Confidence Regions for Functions of a Covariance Matrix.”
*The Annals of Statistics*, **13**(1), 95–115.
doi:10.1214/aos/1176346579.

Dijkstra TK, Henseler J (2015).
“Consistent and Asymptotically Normal PLS Estimators for Linear Structural Equations.”
*Computational Statistics & Data Analysis*, **81**, 10–23.

### See Also

`csem()`

, `calculateSRMR()`

, `calculateDG()`

, `calculateDL()`

, cSEMResults,
`testMICOM()`

, `testMGD()`

, `exportToExcel()`

### Examples

```
# ===========================================================================
# Basic usage
# ===========================================================================
model <- "
# Structural model
eta2 ~ eta1
eta3 ~ eta1 + eta2
# (Reflective) measurement model
eta1 =~ y11 + y12 + y13
eta2 =~ y21 + y22 + y23
eta3 =~ y31 + y32 + y33
"
## Estimate
out <- csem(threecommonfactors, model, .approach_weights = "PLS-PM")
## Test
testOMF(out, .R = 50, .seed = 320)
```

*cSEM*version 0.5.0 Index]