Create a Data.frame for Interrater Agreement

Creates a data.frame for a formula rating ~ subjects | raters with the subjects in rows and the raters in columns as base structure for interrater agreement (IRA) functions.

Usage

raterFrame(formula, data, subset, na.action, dropSubj = FALSE)

Arguments

formula: something like rating ~ subjects | raters
data: the data
subset: potential subset, evaluated in the long-format data
na.action: what should happen with missings, applied per subject (i.e. to the rows of the wide result)
dropSubj: logical; whether to drop the subject column (default FALSE)

Value

a data.frame of class "raterFrame" with subjects in rows and raters in columns. The name of the subject column is kept in the "subject" attribute (NA if it was dropped).

Details

Assessments made by raters are typically - and appropriately - stored and organized in databases. Data originating from databases are usually in long format. Converting this long format into a wide format suitable for analysis is a complex task, as it is not just a matter of simply displaying consecutive values in a new column (as in usual "from-long-to-wide" operations), but of assigning the values to the correct subjects, which requires a sequential join.
The present function supports this process by converting long-format data into a wide format that can be used by subsequent inter-rater agreement functions. Missing values are marked as NAs.

na.action is applied to the wide frame, i.e. per subject: na.omit removes subjects with at least one missing rating. The "na.action" attribute of the result carries a "values" attribute with the identifiers of the omitted subjects; it is absent if nothing was omitted.

Examples


d.long <- data.frame(
     expand.grid(subj=as.character(1:5), rater=LETTERS[1:3]),
     rating = c(1, 4, 5, 7, 2, 2, 5, 6, 7, 1, 1, 4, 6, 6, 2))

# default rater frame
raterFrame(rating ~ subj | rater, data=d.long)
#>   subj A B C
#> 1    1 1 2 1
#> 2    2 4 5 4
#> 3    3 5 6 6
#> 4    4 7 7 6
#> 5    5 2 1 2

# introduce some NAs
raterFrame(rating ~ subj | rater, data=d.long[-c(3, 6), ])
#>   subj  A  B C
#> 1    1  1 NA 1
#> 2    2  4  5 4
#> 3    3 NA  6 6
#> 4    4  7  7 6
#> 5    5  2  1 2

# omit cases containing NAs
raterFrame(rating ~ subj | rater, data=d.long[-c(3, 6), ],
           na.action=na.omit)
#>   subj A B C
#> 2    2 4 5 4
#> 4    4 7 7 6
#> 5    5 2 1 2

# omit the subject column
raterFrame(rating ~ subj | rater, data=d.long, dropSubj=TRUE)
#>   A B C
#> 1 1 2 1
#> 2 4 5 4
#> 3 5 6 6
#> 4 7 7 6
#> 5 2 1 2