# Quick Start To set_cfa_layout" In semptools: Customizing Structural Equation Modelling Plots

```knitr::opts_chunk\$set(
collapse = TRUE,
comment = "#>",
fig.width  =  8,
fig.height =  6,
fig.align = "center"
)
```

# Introduction

The package semptools (CRAN page) contains functions that post-process an output from `semPlot::semPaths()`, to help users to customize the appearance of the graphs generated by `semPlot::semPaths()`. For the introduction to functions for doing very specific tasks, such as moving the parameter estimate of a path or rotating the residual of a variable, please refer to `vignette("semptools")`. The present guide focuses on how to use `set_cfa_layout()` to configure various aspects of an `semPaths` graph generated for a typical confirmatory factor analysis (CFA) model.

# The Initial semPaths Graph

Let us consider a CFA model. We will use `cfa_example`, a sample CFA dataset from `semptools` with 14 variables for illustration.

```library(semptools)
```

This is the CFA model to be fitted:

```mod <-
'f1 =~ x01 + x02 + x03
f2 =~ x04 + x05 + x06 + x07
f3 =~ x08 + x09 + x10
f4 =~ x11 + x12 + x13 + x14
'
```

Fitting the model by `lavaan::cfa()`

```library(lavaan)
fit <- lavaan::cfa(mod, cfa_example)
```

This is the plot from `semPlot::semPaths()`:

```library(semPlot)
p <- semPaths(fit, whatLabels="est",
sizeMan = 3.25,
node.width = 1,
edge.label.cex = .75,
style = "ram",
mar = c(10, 5, 10, 5))
```

The default layout is sufficient to have a quick examination of the results. We will see how `set_cfa_layout()` can be used to do the following tasks to post-process the graph:

• Change the order of the indicators.

• Change the order of the factors.

• Change the curvature of the inter-factor covariances.

• Rotate the graph.

# Order the Indicators and Factors

Suppose we want to do this:

• Order the factors this way, from the left to the right:

• `f2, f1, f4, f3`
• Order the indicators this way, from the left to the right:

• `x04, x05, x06, x07, x01, x02, x03, x11, x12, x13, x14, x08, x09, x10`
• We would like to place the factors this way:

• `f2` above the center of `x04`, `x05`, `x06`, and `x07`.

• `f1` above the center of `x01`, `x02`, and `x03`.

• `f4` above the center of `x11`, `x12`, `x13`, and `x14`.

• `f3` above the center of `x08`, `x09`, and `x10`.

To do this, we create two vectors, one for the argument `indicator_order` and the other for the argument `indicator_factor`.

• `indicator_order` is a string vector with length equal to the number of indicators, with the desired order. In this example, it will be like this:
```indicator_order  <- c("x04", "x05", "x06", "x07",
"x01", "x02", "x03",
"x11", "x12", "x13", "x14",
"x08", "x09", "x10")
```
• `indicator_factor` is a string vector with length equal to the number of indicators. The elements are the names of the latent factors, denoting which indicators will be used to compute the mean positions to place the latent factors:
```indicator_factor <- c( "f2",  "f2",  "f2",  "f2",
"f1",  "f1",  "f1",
"f4",  "f4",  "f4",  "f4",
"f3",  "f3",  "f3")
```

The `set_cfa_layout()` function needs at least three arguments:

• `semPaths_plot`: The `semPaths` plot.

• `indicator_order`: The vector for the order of indicators.

• `indicator_factor`: The vector for the placement of the latent factors.

They do not have to be named if they are in this order.

We now use `set_cfa_layout()` to post-process the graph:

```p2 <- set_cfa_layout(p,
indicator_order,
indicator_factor)
plot(p2)
```

# Change the Curvatures of the Factor Covariances

The graph has the factors and indicators ordered as required. However, the inter-factor covariances are too close to the factors. To increases the curvatures of the covariances, we can use the argument `fcov_curve`. The default is .4. Let us increase it to 1.75.

```p2 <- set_cfa_layout(p,
indicator_order,
indicator_factor,
fcov_curve = 1.75)
plot(p2)
```

The covariances are now more readable. The exact effect of the values vary from graph to graph. Therefore, trial and error is required to find a value suitable for a graph.

We can also move all the factor loadings together using the argument `loading_position`. The default value is .5, at the middle of the paths. If we want to move the loadings closer to the indicators, we increase this number. If we want to move the loadings closer to the indicators, we decrease this number. In the following example, we move the loadings closer to the indicators, and increase the distance between them in the process.

```p2 <- set_cfa_layout(p,
indicator_order,
indicator_factor,
fcov_curve = 1.75,
plot(p2)
```

# Rotate the Model

The default orientation is "pointing downwards": latent factors on the top, pointing down to the indicators on the bottom. The orientation can be set to one of these four directions: down (default), left, up, and right. This is done by the argument `point_to`.

```p2 <- set_cfa_layout(p,
indicator_order,
indicator_factor,
fcov_curve = 1.75,
point_to = "up")
plot(p2)
```

# Pipe

Like other functions in `semptools`, the `set_cfa_layout()` function can be chained with other functions using the pipe operator, `%>%`, from the package `magrittr`, or the native pipe operator `|>` available since R 4.1.x. Suppose we want to mark the significant test results for the free parameters using `mark_sig()`:

```# If R version >= 4.1.0
p2 <- set_cfa_layout(p,
indicator_order,
indicator_factor,
fcov_curve = 1.75,
point_to = "up") |>
mark_sig(fit)
plot(p2)
```
```# if ((compareVersion(as.character(getRversion()), "4.1.0")) >= 0) {
#     p2 <- set_cfa_layout(p, indicator_order, indicator_factor, fcov_curve = 1.75,
#                         mark_sig(fit)
#   } else {
require(magrittr)
p2 <- set_cfa_layout(p, indicator_order, indicator_factor, fcov_curve = 1.75,
mark_sig(fit)
# }
plot(p2)
```

# Limitations

• Currently, if a function needs the SEM output, only lavaan output is supported.

## Try the semptools package in your browser

Any scripts or data that you put into this service are public.

semptools documentation built on Oct. 15, 2023, 5:07 p.m.