Automatically uses RM (dichotomous data) or PCM (polytomous data) depending on data structure.
Usage
RIitemfit(
data,
simcut,
output = "table",
sort = "items",
cutoff = c(0.001, 0.999),
...
)Arguments
- data
Dataframe with response data
- simcut
Object output from
RIgetfit()- output
Optional "dataframe" or "quarto"
- sort
Optional "infit" or "outfit"
- cutoff
Default
c(.001,.999)- ...
Options passed on to
kbl_rise()for table creation
Details
Uses iarm::out_infit() to calculate conditional mean square fit statistics
for all items. See Müller (2020, DOI: 10.1186/s40488-020-00108-7) for details.
Note: only uses complete cases! This is explicitly mentioned in the automatic
table caption text.
Cutoff threshold values from simulation data (using option simcut) are
used with the quantile() function with .001 and .999 values to filter out
extremes. Actual cutoff values are shown in the output.
Simulated datasets that have zero responses in any response category that should have data will automatically be removed/skipped from analysis, which means that final set of iterations may be lower than specified by user.
Optional sorting (only) for table output with conditional highlighting based
on simulation cutoff values, either sort = "infit" or sort = "outfit.
Optional conditional highlighting of misfit based on rule-of-thumb values for
infit MSQ according to Smith et al. (1998), since Müller (2020) showed that
these can be fairly accurate for conditional infit and thus useful for a
quick look at item fit. Set cutoff = "Smith98 to use.