person_item_maximization

internal

Provides tools to adjust estimates of learning for guessing-related
bias in educational and survey research. Implements standard guessing
correction methods and latent class models that leverage informative
pre-post test transitions to account for guessing behavior.
The package helps researchers obtain more accurate estimates of actual
learning when respondents may guess on closed-ended knowledge items.
For theoretical background and empirical validation, see Cor and Sood (2016)
<https://gsood.com/research/papers/guess.pdf>.

Gaurav Sood

guess

Adjust Estimates of Learning for Guessing

Ken Cor

person_item_maximization function

<dl><dt>pre</dt>
<dd>numeric pre-test matrix</dd>
<dt>post</dt>
<dd>numeric post-test matrix</dd>
<dt>complete</dt>
<dd>logical matrix of complete response pairs</dd>
<dt>expected</dt>
<dd>output from person_item_expectation</dd>
<dt>gamma</dt>
<dd>current item guessing probabilities</dd>
<dt>epsilon</dt>
<dd>boundary used to keep probabilities in the open interval</dd></dl>

Arguments

Person/item EM maximization step — person_item_maximization

<dl>

<dt>pre</dt>
<dd>numeric pre-test matrix</dd>


<dt>post</dt>
<dd>numeric post-test matrix</dd>


<dt>complete</dt>
<dd>logical matrix of complete response pairs</dd>


<dt>expected</dt>
<dd>output from person_item_expectation</dd>


<dt>gamma</dt>
<dd>current item guessing probabilities</dd>


<dt>epsilon</dt>
<dd>boundary used to keep probabilities in the open interval</dd>

</dl>

person_item_maximization: Person/item EM maximization step

Description

Usage

Value

Arguments