26/08/2019 - Identifiability Problems and Possible Solutions in Test Score Equating - Palestrante: Jorge González (PUC-Chile)

SEMINÁRIO CONJUNTO UFSCAR/ICMC-USP
(Grupo de Modelos de Variáveis Latentes)

Data e Horário:
26/08/2019 às 14h

Local:
Auditório Luiz Antonio Favaro (sala 4-111), ICMC-USP
também será transmitido por Google Meet (acesse neste link)

Título:
Identifiability Problems and Possible Solutions in Test Score Equating

Palestrante:
Jorge González (PUC, Chile)

Resumo:
The nonequivalent groups with anchor test design (NEAT) is widely used in test equating. In this design, two groups of examinees are administered separate test forms with each test form containing a common subset of items. Because test takers from different populations are assigned only one test form, missing score data emerge by design rendering some of the score distributions unavailable. The equating literature has treated this problem from different perspectives, all of them making different assumptions in order to estimate the missing score distributions. In this talk I'll use the concept of partial identification of probability  distributions to offer an alternative to this traditional practice in NEAT equating. Comparisons with current methods of equating under the NEAT design are made and illustrated.