Item selection function based on network distance criterion.

This item selection function delivers the item farthest in the network from the items a respondent has already answered, with edges weighted by the inverse of their entry in the item-item adjacency matrix. Ties are broken using the maximum information criterion.

Usage

select_max_dist(
  pers,
  item,
  resp,
  resp_cur = NULL,
  adj_mat = NULL,
  n_candidates = 1
)

Arguments

pers

A dataframe of current respondent ability estimates.

item

A dataframe of current item parameter estimates.

resp

A long-form dataframe of all potential pre-simulated item responses.

resp_cur

A long-form dataframe of administered item responses.

adj_mat

An item-item adjacency matrix, where each entry is the count of individuals who have respondent to both item i and item j. See documentation for construct_adj_mat

n_candidates

A parameter that allows the assembly of a pool of $N$ farthest items, before selecting the next item according to maximum information. Allows users to balance exposure patterns away from increased network density and toward more efficient estimation.

Value

A long-form dataframe of all previously administered item responses with the new responses from this iteration appended to the end.