1EdTech QTI Glossary of Terms for Usage Data Item Statistics
Vocabulary Details
This document provides a human readable description of the vocabulary terms defined in the 1EdTech Vocabulary Definition and Exchange (VDEX) file "itemstatisticsglossaryv1p0.xml". The details of the vocabulary are:
Vocabulary Identifier: | http://www.imsglobal.org/qti/qtiv2p1/itemstatisticsglossaryv1p0 |
Vocabulary Type: | glossaryOrDictionary |
Registered Status: | true |
Vocabulary Language: | en-US |
Author: | Colin Smythe, 1EdTech (UK) |
Issue Date: | 2013-01-07 |
Version: | 1.0 |
Status: | Final Release |
Base Specification: | Question and Test Interoperability (QTI) Version 2.1 |
Keywords: | QTIv2.1, Usage Data, Item Statistics Glossary |
Publisher: | 1EdTech Consortium Inc. |
Copyright: | 2012 (c) 1EdTech Consortium Inc. |
Description: | The set of terms that would be used to collect the item statistics. This is the value that will be entered in the 'name' element. |
History: | Version 1.0: The first formal release of this QTI Glossary. |
License: | Users of products or services that include this document are hereby granted a worldwide, royalty-free, non-exclusive license to use this document. |
Vocabulary Terms
The set of terms for the vocabulary are described in the table below.
Identifier | Caption | Description |
---|---|---|
A-Param | IRT a parameter | Index of item discrimination based on item response theory. |
AIS | Average Item Score | A group-dependent measure of item difficulty. Equal to the mean of the item scores for a group of individuals. For a rights-only dichotomously scored test, equivalent to the percent answering the item correctly (P-value). |
B-Param | IRT b parameter | Index of item difficulty based on item response theory. |
C-Param | IRT c parameter | The lower asymptote of the item response function. Generally an index of guessing i.e. the probability of getting the item correct for very low ability examinees. |
D-Param | IRT d parameter | A measure of the dispersion of the item computed using the IRT method. |
Fifths_Table_Highest | Fifths_Table_Highest | Count of examinees in the highest fifth of the test score distribution who selected the specified option. |
Fifths_Table_Lowest | Fifths_Table_Lowest | Count of examinees in the lowest fifth of the test score distribution who selected the specified option. |
Fifths_Table_Middle | Fifths_Table_Middle | Count of examinees in the middle fifth of the test score distribution who selected the specified option. |
Fifths_Table_Second_Highest | Fifths_Table_Second_Highest | Count of examinees in the second highest fifth of the test score distribution who selected the specified option. |
Fifths_Table_Second_Lowest | Fifths_Table_Second_Lowest | Count of examinees in the second lowest fifth of the test score distribution who selected the specified option. |
P-value | P-value | A group-dependent measure of item difficulty. It is equal to the percent answering the item correctly. |
PHI | Phi coefficient | A group-dependent measure of item discrimination, Phi is the correlation of two dichotomous variables. In item analysis, it is generally the correlation between getting an item right or wrong and some dichotomous variable e.g. pass, fail. |
PTbis | r-pointbiserial | A group-dependent measure of item discrimination, r-pointbiserial is the point biserial correlation between the score on a dichotomous item and a criterion which is a variable that can take many possible values, such as total score on the test. |
Polyserial | Polyserial | The polyserial correlation is the generalization of the biserial correlation for use with either dichotomous or polytomous items. For example, it may be the generalized form of the correlation with a criterion. It is the correlation of the item score with some continuous variable, usually the total score on the test. |
Score_Conversion | Score Conversion Table | Raw scores and their corresponding converted values e.g. the values to be used in score reports. |
rbis | r-biserial | A group-dependent measure of item discrimination, r-biserial is an estimate of the biserial correlation between a latent continuous variable hypothesised to underlie performance on a dichotomous item (i.e. item score) and a criterion which is a variable that can take many possible vales, such as total score on the test. |