item_analysis {irt}

Item Analysis Function

Description

Item Analysis Function

Usage

item_analysis(
  resp,
  criterion = NULL,
  ip = NULL,
  stats = c("n", "pval", "pbis", "bis", "pbis_adj", "bis_adj"),
  suppress_output = FALSE
)

Arguments

resp

A Response_set-class object, matrix or data.frame containing the item responses.

criterion

Provide a continuous criterion variable such as a total raw score, or theta score that will be used in the calculation of correlation calculations. If this value is NULL, the total score will be used.

ip

An Itempool-class object. This will help function in two ways. First, if the resp is a Response_set-class object, the function will help the responses to be arranged in the same order as ip. Second, if there are polytomous items in the data, ip will help finding the maximum values of each item. Otherwise, the maximum values each item can take will be calculated using data, which may be fallible.

stats

A vector of string containing the columns/statistics to be calculated. 'item_id' column will be added by default. Some or all of the following columns can be added to the output: c("n", "pval", "pbis", "bis", "pbis_adj", "bis_adj"). Please see the 'value' section below to see the details of these columns. By default, all of the columns above will be calculated.

suppress_output

If TRUE, the function will suppress console output. Default value is FALSE

Value

A data.frame with following columns:

'item_id'

Item ID.

'n'

Number of examinees responded this item.

'pval'

p-value, proportion of examinees correctly answered items. If there are polytomous items in the data, p-value will be calculated by dividing the mean of the scores for the item by the maximum possible score of the item.

'pval_unadj'

Unadjusted p-value, this is the mean of item scores that is not adjusted for the maximum possible score as 'pval' column does. For dichotomous items, this will be the same as 'pval' column.

'pbis'

Point biserial correlation.

'bis'

Biserial correlation.

'pbis_adj'

Point biserial correlation between item and total score without this item. Note that this stat is only available when criterion is NULL.

'bis_adj'

Biserial correlation between item and total score without this item. Note that this stat is only available when criterion is NULL.

Author(s)

Emre Gonulates

Examples

theta <- rnorm(100)
ip <- generate_ip(n = 20)
resp <- sim_resp(ip = ip, theta = theta, prop_missing = .2)
# Item analysis based on total scores
item_analysis(resp)
# Item analysis based on theta scores
item_analysis(resp, criterion = theta)