NotesFAQContact Us
Collection
Advanced
Search Tips
Back to results
Peer reviewed Peer reviewed
PDF on ERIC Download full text
ERIC Number: EJ1238381
Record Type: Journal
Publication Date: 2019-Dec
Pages: 27
Abstractor: As Provided
ISBN: N/A
ISSN: EISSN-2330-8516
EISSN: N/A
Available Date: N/A
A Modified "a"-Stratified Method for Computerized Adaptive Testing. Research Report. ETS RR-19-10
Gu, Lixiong; Ling, Guangming; Qu, Yanxuan
ETS Research Report Series, Dec 2019
Research has found that the "a"-stratified item selection strategy (STR) for computerized adaptive tests (CATs) may lead to insufficient use of high a items at later stages of the tests and thus to reduced measurement precision. A refined approach, unequal item selection across strata (USTR), effectively improves test precision over the STR by allowing more items to be selected from the strata with higher a-parameter values. However, both approaches ignore the contribution of items' c-parameters to the information. This study proposes another procedure--maximum information STR (MISTR)--that groups items based on the maximum amount of Fisher information an item can provide. This information is a function of its "a"- and "c"-parameters. MISTR can be further modified to select more items from strata with high a-parameter values (unequal MISTR [UMISTR]). This study evaluated and compared MISTR, UMISTR, STR, and USTR on two aspects of the CAT performance: (a) quality of [theta] estimation and (b) effectiveness in item pool usage. The results showed that both the MISTR and UMISTR approaches produced more precise ability estimation than the STR approach when the test length was longer and when an item-exposure-control procedure was used. The UMISTR produced slightly less precise ability estimation than USTR but led to fewer underused items, indicating a more balanced use of the item pool. These findings suggest that MISTR and UMISTR can be viable alternatives to STR and USTR.
Educational Testing Service. Rosedale Road, MS19-R Princeton, NJ 08541. Tel: 609-921-9000; Fax: 609-734-5410; e-mail: RDweb@ets.org; Web site: https://www.ets.org/research/policy_research_reports/ets
Publication Type: Journal Articles; Reports - Research
Education Level: N/A
Audience: N/A
Language: English
Sponsor: N/A
Authoring Institution: N/A
Grant or Contract Numbers: N/A
Author Affiliations: N/A