mirt (version 1.42)

reverse.score: Reverse score one or more items from a response matrix

Description

Reverse score specific items given empirical range or specific scoring range.

Usage

reverse.score(data, which, range = NULL, append = ".RS")

Value

returns the original data object with the specified items reverse scored replacing the original scoring scheme

Arguments

data

an object of class data.frame, matrix, or table with the response patterns

which

names of items in data that should be rescored. If missing the all columns in data will be reverse scored

range

(optional) a named list to specify the low and high score ranges. Specified names must match the names found in data, and each element of this list should contain only two values. If items specified in which are omitted from this list then the empirical min/max information will be used instead

append

character vector indicating what to append to the item names that have been rescored

Author

Phil Chalmers

References

Chalmers, R., P. (2012). mirt: A Multidimensional Item Response Theory Package for the R Environment. Journal of Statistical Software, 48(6), 1-29. tools:::Rd_expr_doi("10.18637/jss.v048.i06")

Examples

a <- rlnorm(20)
a[c(1,5,10)] <- -a[c(1,5,10)]
diffs <- t(apply(matrix(runif(20*4, .3, 1), 20), 1, cumsum))
diffs <- -(diffs - rowMeans(diffs))
d <- diffs + rnorm(20)
dat <- simdata(a,d,itemtype='graded', N=300)
head(dat)

if (FALSE) {
# fitted model has negative slopes due to flipped scoring
mod <- mirt(dat)
coef(mod, simplify=TRUE)$items
plot(mod, type = 'itemscore')
}

# reverse the scoring for items 1, 5, and 10 only using empirical min/max
revdat <- reverse.score(dat, c('Item_1', 'Item_5', 'Item_10'))
head(revdat)

# compare
apply(dat[,c(1,5,10)], 2, table)
apply(revdat[,c(1,5,10)], 2, table)

if (FALSE) {
# slopes all positive now
mod2 <- mirt(revdat)
coef(mod2, simplify=TRUE)$items
plot(mod2, type = 'itemscore')
}

# use different empirical scoring information due to options not used
  # 0 score not observed for item 1, though should have been rescored to a 4
dat[dat[,1] == 0, 1] <- 1
table(dat[,1])

# 4 score not observed for item 5, though should have been rescored to a 0
dat[dat[,5] == 4, 5] <- 3
table(dat[,5])

# specify theoretical scoring values in the range list
revdat2 <- reverse.score(dat, c('Item_1', 'Item_5', 'Item_10'),
                              range = list(Item_1 = c(0,4), Item_5 = c(0,4)))
head(revdat2)
table(dat[,1])
table(revdat2[,1])

table(dat[,5])
table(revdat2[,5])