ScholarLens Back to search
1964•ETS Research Bulletin SeriesRequires access

ITEM ANALYSIS AND TEST RELIABILITY

Charles T. Myers

Open publisher page 4 citations

Abstract

ABSTRACT This study compares four different item‐analysis procedures with respect to their effectiveness in identifying items which would make up reliable and valid tests. The four procedures were: (1) the item‐test biserial correlation, (2) the item‐test covariance divided by the test standard deviation (or the item reliability index), (3) the “standard length” of the item, and (4) the ratio of the item validity index to the item reliability index. These procedures were applied to a 32‐item, multiple‐choice test with an essay test criterion. Results for a 300‐case item‐analysis sample and for three 100‐case cross‐validation samples are presented which show each of the four procedures to be about equally effective.

About this research paper

What this paper is about

ABSTRACT This study compares four different item‐analysis procedures with respect to their effectiveness in identifying items which would make up reliable and valid tests. The four procedures were: (1) the item‐test biserial correlation, (2) the item‐test covariance divided by the test standard deviation (or the item reliability index), (3) the “standard length” of the item, and (4) the ratio of the item validity index to the item reliability index. These procedures were applied to a 32‐item, multiple‐choice test with an essay test criterion. Results for a 300‐case item‐analysis sample and for three 100‐case cross‐validation samples are presented which show each of the four procedures to be about equally effective.

Why it matters

OpenAlex reports 4 citations for this work. Citation counts describe recorded attention and do not establish research quality.

Key contribution

A contribution statement is not available in the OpenAlex record.

Method / approach

Method details are not available in the OpenAlex metadata.

Main findings

Findings are not separately available in the OpenAlex metadata.

Limitations

Limitations are not available in the OpenAlex metadata.

Applications

Application details are not available in the OpenAlex metadata.

Available abstract

ABSTRACT This study compares four different item‐analysis procedures with respect to their effectiveness in identifying items which would make up reliable and valid tests. The four procedures were: (1) the item‐test biserial correlation, (2) the item‐test covariance divided by the test standard deviation (or the item reliability index), (3) the “standard length” of the item, and (4) the ratio of the item validity index to the item reliability index. These procedures were applied to a 32‐item, multiple‐choice test with an essay test criterion. Results for a 300‐case item‐analysis sample and for three 100‐case cross‐validation samples are presented which show each of the four procedures to be about equally effective.

Key concepts: Statistics, Reliability (semiconductor), Test (biology), Item analysis, Index (typography), Standard deviation, Mathematics, Computer science

Related papers

Related paper 1Related paper 2Related paper 3
Back to paper searchBrowse research topicsOriginal source
ITEM ANALYSIS AND TEST RELIABILITY — Research Paper | ScholarLens