op mal reassembly of shadow tests in cat
play

Op#mal Reassembly of Shadow Tests in CAT Seung W. Choi, - PowerPoint PPT Presentation

Op#mal Reassembly of Shadow Tests in CAT Seung W. Choi, Karin Mollering, Jie Li, & Wim J. van der Linden McGraw-Hill Educa#on CTB 2015


  1. Op#mal ¡Reassembly ¡of ¡Shadow ¡Tests ¡in ¡CAT ¡ Seung ¡W. ¡Choi, ¡Karin ¡Mollering, ¡Jie ¡Li, ¡& ¡ Wim ¡J. ¡van ¡der ¡Linden ¡ ¡ McGraw-­‑Hill ¡Educa#on ¡CTB ¡ 2015 ¡Na#onal ¡Conference ¡on ¡Student ¡Assessment, ¡June ¡22-­‑24, ¡San ¡Diego, ¡CA ¡

  2. Context ¡ I. “The ¡[CAT] ¡delivery ¡system ¡must ¡support ¡ at ¡least ¡ 250,000 ¡simultaneous ¡ testers.” ¡ II. “Contractor’s ¡[CAT] ¡solu#on ¡must ¡be ¡a ¡ mul)-­‑stage ¡approach ¡and ¡allows ¡ for ¡different ¡item ¡selec#on ¡rou#nes/algorithms ¡to ¡be ¡applied ¡to ¡each ¡ sec#on ¡if ¡desired.” ¡ III. “Assessments ¡will ¡be ¡ par))oned ¡into ¡parts/segments ¡that ¡will ¡target ¡ specific ¡areas ¡of ¡content. ¡[…] ¡Different ¡selec#on ¡rules ¡could ¡then ¡be ¡ applied ¡to ¡either ¡consecu#ve ¡or ¡pre-­‑iden#fied ¡item ¡subsets.” ¡

  3. Objec#ve ¡ • Even ¡in ¡the ¡age ¡of ¡abundant ¡and ¡fast ¡compu#ng ¡resources, ¡ concurrency ¡requirements ¡for ¡large-­‑scale ¡online ¡tes#ng ¡ programs ¡s#ll ¡put ¡an ¡uninterrupted ¡delivery ¡of ¡computer-­‑ adap#ve ¡tests ¡at ¡risk. ¡ ¡ • For ¡the ¡par#cularly ¡compu#ng-­‑intensive ¡shadow-­‑test ¡ approach ¡to ¡CAT ¡(Shadow ¡CAT) ¡we ¡therefore ¡propose ¡ effec#ve ¡strategies ¡reducing ¡computa#onal ¡demands ¡without ¡ compromising ¡measurement ¡quality. ¡

  4. Shadow ¡Test ¡Approach ¡to ¡CAT ¡ Maximum ¡ Content ¡ Measurement ¡ Psychometric ¡ Precision ¡ Constraints ¡ Sequen#al ¡ Simultaneous ¡ Selec#on ¡ Selec#on ¡ Shadow ¡CAT ¡

  5. Standard ¡Shadow ¡CAT ¡ . ¡. ¡. ¡ Previously ¡administered ¡item ¡ 1 ¡ n-­‑1 ¡ n ¡ 2 ¡ 3 ¡ 4 ¡ Item ¡Posi#on ¡

  6. Universal ¡Shadow ¡Test ¡Assembler* ¡ 1 ¡ ¡If ¡shadow ¡test ¡is ¡reassembled ¡ S i ¡ ¡= ¡ ¡ 0 ¡ ¡if ¡shadow ¡test ¡is ¡frozen ¡ 𝑗 =1,2,3,… 𝑜 ¡ 1 ¡ 2 ¡ 3 ¡ 4 ¡ 5 ¡ 6 ¡ 7 ¡ 8 ¡ 9 ¡ 10 ¡ 𝑜 =10 ¡ 1 ¡ 1 ¡ 1 ¡ 1 ¡ 1 ¡ 1 ¡ 1 ¡ 1 ¡ 1 ¡ 1 ¡ Standard ¡ShadowCAT ¡ LOFT ¡ 1 ¡ 0 ¡ 0 ¡ 0 ¡ 0 ¡ 0 ¡ 0 ¡ 0 ¡ 0 ¡ 0 ¡ 1 ¡ 0 ¡ 0 ¡ 0 ¡ 0 ¡ 1 ¡ 0 ¡ 0 ¡ 0 ¡ 0 ¡ 2-­‑Stage ¡on-­‑the-­‑fly ¡MST ¡ Adap#ve ¡+ ¡Fixed ¡ 1 ¡ 1 ¡ 1 ¡ 1 ¡ 1 ¡ 0 ¡ 0 ¡ 0 ¡ 0 ¡ 0 ¡ Fixed ¡+ ¡Adap#ve ¡ 1 ¡ 0 ¡ 0 ¡ 0 ¡ 0 ¡ 1 ¡ 1 ¡ 1 ¡ 1 ¡ 1 ¡ * ¡van ¡der ¡Linden ¡and ¡Diao ¡(2014) ¡– ¡Using ¡a ¡Universal ¡Shadow-­‑Test ¡Assembler ¡with ¡Mul#stage ¡Tes#ng ¡

  7. Reassembly ¡Policies ¡ • Freeze ¡Period ¡Policy ¡ • Theta ¡Threshold ¡Policy ¡ • Hybrid ¡Policy ¡

  8. Freeze ¡Period ¡Policy ¡ Freeze ¡Policy ¡ (t_freeze ¡= ¡x) ¡ Item ¡ 0 ¡ 1 ¡ 2 ¡ 3 ¡ Posi#on ¡ 1 ¡ Assemble ¡ Assemble ¡ Assemble ¡ Assemble ¡ 2 ¡ Assemble ¡ Freeze ¡ Freeze ¡ Freeze ¡ 3 ¡ Assemble ¡ Assemble ¡ Freeze ¡ Freeze ¡ 4 ¡ Assemble ¡ Freeze ¡ Assemble ¡ Freeze ¡ 5 ¡ Assemble ¡ Assemble ¡ Freeze ¡ Assemble ¡ 6 ¡ Assemble ¡ Freeze ¡ Freeze ¡ Freeze ¡ 7 ¡ Assemble ¡ Assemble ¡ Assemble ¡ Freeze ¡ 8 ¡ Assemble ¡ Freeze ¡ Freeze ¡ Freeze ¡ 9 ¡ Assemble ¡ Assemble ¡ Freeze ¡ Assemble ¡ 10 ¡ Assemble ¡ Freeze ¡ Assemble ¡ Freeze ¡ 11 ¡ Assemble ¡ Assemble ¡ Freeze ¡ Freeze ¡ 12 ¡ Assemble ¡ Freeze ¡ Freeze ¡ Freeze ¡ 13 ¡ Assemble ¡ Assemble ¡ Assemble ¡ Assemble ¡ 14 ¡ Assemble ¡ Freeze ¡ Freeze ¡ Freeze ¡

  9. Theta ¡Threshold ¡Policy ¡ Threshold ¡Policy ¡ (threshold ¡= ¡y) ¡ Item ¡ 0 ¡ 0.1 ¡ 0.3 ¡ 0.5 ¡ Posi#on ¡ 1 ¡ Assemble ¡ Assemble ¡ Assemble ¡ Assemble ¡ 2 ¡ Assemble ¡ > ¡0.1 ¡ > ¡0.3 ¡ > ¡0.5 ¡ 3 ¡ Assemble ¡ > ¡0.1 ¡ > ¡0.3 ¡ > ¡0.5 ¡ 4 ¡ Assemble ¡ > ¡0.1 ¡ > ¡0.3 ¡ > ¡0.5 ¡ 5 ¡ Assemble ¡ > ¡0.1 ¡ > ¡0.3 ¡ > ¡0.5 ¡ 6 ¡ Assemble ¡ > ¡0.1 ¡ > ¡0.3 ¡ > ¡0.5 ¡ 7 ¡ Assemble ¡ > ¡0.1 ¡ > ¡0.3 ¡ > ¡0.5 ¡ 8 ¡ Assemble ¡ > ¡0.1 ¡ > ¡0.3 ¡ > ¡0.5 ¡ 9 ¡ Assemble ¡ > ¡0.1 ¡ > ¡0.3 ¡ > ¡0.5 ¡ 10 ¡ Assemble ¡ > ¡0.1 ¡ > ¡0.3 ¡ > ¡0.5 ¡ 11 ¡ Assemble ¡ > ¡0.1 ¡ > ¡0.3 ¡ > ¡0.5 ¡ 12 ¡ Assemble ¡ > ¡0.1 ¡ > ¡0.3 ¡ > ¡0.5 ¡ 13 ¡ Assemble ¡ > ¡0.1 ¡ > ¡0.3 ¡ > ¡0.5 ¡ 14 ¡ Assemble ¡ > ¡0.1 ¡ > ¡0.3 ¡ > ¡0.5 ¡

  10. Theta ¡Threshold ¡Policy ¡ (true ¡θ ¡= ¡2.5; ¡θ ¡threshold ¡= ¡0.1) ¡ ¡

  11. Hybrid ¡Policy ¡ Hybrid ¡Policy ¡ (t_freeze ¡= ¡x ¡& ¡threshold ¡= ¡y) ¡ Item ¡ 0 ¡ 1 ¡ 2 ¡ 3 ¡ Posi#on ¡ 1 ¡ Assemble ¡ Assemble ¡ Assemble ¡ Assemble ¡ 2 ¡ Assemble ¡ Freeze ¡ Freeze ¡ Freeze ¡ 3 ¡ Assemble ¡ ? ¡ Freeze ¡ Freeze ¡ 4 ¡ Assemble ¡ Freeze ¡ ? ¡ Freeze ¡ 5 ¡ Assemble ¡ ? ¡ Freeze ¡ ? ¡ 6 ¡ Assemble ¡ Freeze ¡ Freeze ¡ Freeze ¡ 7 ¡ Assemble ¡ ? ¡ ? ¡ Freeze ¡ 8 ¡ Assemble ¡ Freeze ¡ Freeze ¡ Freeze ¡ 9 ¡ Assemble ¡ ? ¡ Freeze ¡ ? ¡ 10 ¡ Assemble ¡ Freeze ¡ ? ¡ Freeze ¡ 11 ¡ Assemble ¡ ? ¡ Freeze ¡ Freeze ¡ 12 ¡ Assemble ¡ Freeze ¡ Freeze ¡ Freeze ¡ 13 ¡ Assemble ¡ ? ¡ ? ¡ ? ¡ 14 ¡ Assemble ¡ Freeze ¡ Freeze ¡ Freeze ¡

  12. Empirical ¡Study ¡ • Item ¡pool ¡ – ¡1,000 ¡items ¡(918 ¡MC ¡and ¡82 ¡CR ¡items ¡) ¡ • Test ¡specifica)ons ¡ – ¡Smarter ¡Balanced ¡Math ¡test ¡blueprint ¡ • CAT ¡Length ¡ – ¡32 ¡items ¡ • Sample ¡– ¡θ ¡at ¡fixed ¡intervals ¡(from ¡-­‑2.5 ¡to ¡2.5) ¡with ¡500 ¡ replica#ons ¡per ¡θ ¡(N ¡= ¡5,500) ¡ • Freeze ¡interval ¡ – ¡t freeze ¡= ¡{3, ¡7, ¡15, ¡31} ¡ • Theta ¡threshold ¡ – ¡θ threshold ¡ = ¡{0.1, ¡0.3, ¡0.5, ¡0.7, ¡0.9} ¡ • Hybrid ¡-­‑ ¡t freeze ¡= ¡{3, ¡7} ¡& ¡θ threshold ¡= ¡{0.1, ¡0.3, ¡0.5, ¡0.7, ¡0.9} ¡ • Benchmarks ¡– ¡Standard ¡Shadow ¡CAT ¡& ¡LOFT ¡at ¡true ¡θ ¡ • Evalua)on ¡criteria ¡ – ¡Refresh ¡rates, ¡RMSE, ¡Bias ¡

  13. Item ¡Pool ¡Informa#on ¡Func#on ¡

  14. Test ¡Blueprint ¡Constraints ¡

  15. Results ¡

Download Presentation
Download Policy: The content available on the website is offered to you 'AS IS' for your personal information and use only. It cannot be commercialized, licensed, or distributed on other websites without prior consent from the author. To download a presentation, simply click this link. If you encounter any difficulties during the download process, it's possible that the publisher has removed the file from their server.

Recommend


More recommend