IPV (version 1.0.0)

ipv_est: IPV estimation

Description

IPV estimation

Usage

ipv_est(
  dat,
  name,
  include_raw = TRUE,
  include_lav = TRUE,
  include_xarrow = TRUE,
  id = "id",
  value.var = "value",
  ...
)

Value

list; $est includes the center distances and all necessary input for the IPV chart functions, $est_raw includes the factor loadings and latent correlations, $lav includes the fitted models (class: lavaan), $xarrow includes a data frame for arrows between facets in nested charts, that can be passed on directly to

nested_chart; by default, all three of these elements are provided.

$xarrow includes only those cases, where the estimate of the latent correlation between facets exceeds the estimate of the latent correlation between their respective tests, as recommended by the original authors.

Arguments

dat

data frame; raw data (see details)

name

character; name of the overall construct or test that comprises all items used

include_raw

logical; should raw estimates of factor loadings be included in the output?; defaults to TRUE

include_lav

logical; should lavaan objects of the fitted models be included in the output?; defaults to TRUE

include_xarrow

logical; should an object for the drawing of arrows in nested plots be returned?; defaults to TRUE

id

character; name of the case identifying variable in long format; defaults to "id"

value.var

character; name of the variable in long format that contains measurement values; defaults to "value"

...

further arguments passed to lavaan::cfa (or one step further to lavaan::lavOptions).

Details

the data given to dat can be either in long or in wide format.

If they are in wide format, they have to conform to the following rules: * no additional variables / columns * variables are named according to the following pattern: "test_facet_item". * If there is only one test in the data, the pattern is "facet_item". For tests without facets in a larger dataset also comprising tests with items, the pattern is "test_item". * Variable names have to be unique. Item names have to be unique at the level of the test (not only at the level of the facet) See example

If they are in long format, they have to include the columns "test", "facet", and "item", as well as a case identifying variable (id) and the measurement variable (value.var).

Examples

# An IPV that comprises the honesty/humility and the agreeableness factor of
# the HEXACO. Estimation takes some time.
if (FALSE) res <- ipv_est(
  HEXACO[ ,grep("^H|^A", names(HEXACO))],
  "HA")
nested_chart(res)

# Customize call to lavaan::cfa via ellipsis to treat missing data and use long format
# Estimation takes some time.
if (FALSE) HEXACO_long <- reshape2::melt(
  cbind(id = row.names(HEXACO)[1:1000],
  HEXACO[1:1000,1:240]),
 id.vars = "id")
HEXACO_long$test <- substr(HEXACO_long$variable, 1, 1)
HEXACO_long$facet <- substr(HEXACO_long$variable, 3, 6)
HEXACO_long$item <- substr(HEXACO_long$variable, 8, 13)
HEXACO_long$variable <- NULL
head(HEXACO_long)
res <- ipv_est(
  HEXACO_long[HEXACO_long$test %in% c("H", "A"), ],
  name = "HA", missing = "fiml")