[ABE-L] Seminários do grupo de Modelos de Variáveis Latentes - Paula Fariña

[Marcelo Andrade]

Bom dia,

Gostaríamos de convidar a todos para o próximo seminário do grupo de
pesquisa de Modelos de Variáveis Latentes (MVL) da USP <
https://mvl.icmc.usp.br> cadastrado no CNPq. Seguem as informações:

*Data:* quinta-feira, 01 de dezembro.

*Horário:* 15:00 (Brasília).

*Nome:* Paula Fariña

*Título:* A computarized Adaptive Tool to Assist Statistical Learning

*Resumo:*
This research explores the potential of the Computerized Adaptive Learning
(CAL) technology to support an active learning class in a statistical
inference course. The distinguishing feature of these systems is that each
student receives a different set of items: the student's skill/knowledge is
measured from the responses of the items. Then, a new item is given to each
student based on their skill/knowledge previously measured. CAL system
proposed here includes a bank of 50 items of the topic: ‘hypothesis
testing’.

The platform employed to apply the CAL system is at the second prototype
stage. It is still under construction now. The first prototype consisted in
using the open-source Concerto platform to apply the tool. The tool was
applied on two sections of an Econometric course belonging to the
Industrial Engineer carrier at the Universidad Diego Portales (26 and 21
students) the second semester 2022. The scheme of the application
considered: 5 minutes explanation, 20 minutes using the test, 20 minutes of
teacher’s feedback, 20 minutes using the test, and 10 minutes answering a
satisfaction survey. In the first section, the feedback included a formal
explanation of the topic, while in the second section, the teacher only
answered students’ questions. The second prototype is an app build in R
that employs the Glicko ranking system to rank both, student’s abilities,
and item’s difficulties.

Preliminary results on the first prototype suggest that the activity was
enjoyed by 50% of the students and that satisfaction was similar in both
sections. Besides, clear differences were observed between sections: in the
section with expositive feedback the students reduced answering times and
the tool moved to more difficult stems. On the other side, the students
that received ‘answering questions’ as feedback did not improve their
results after the this. The second prototype will be proved in three
sections of the course “Statistical Inference” (last week of November 2022).


*Bio:* Paula Fariña is an Economist at the Universidad de San Andrés in
Buenos Aires, Argentina (2001), and PhD. in Statistics at the Universidad
Católica de Chile (2010). In 2011, she did a postdoctoral research at the
Centro de Estudios de Políticas y Práticas en Educación (CEPPE). Since 2012
she has been an assistant professor at the Industrial Egineering School,
Facultad de Ingeniería y Ciencias, Universidad Diego Portales. Her research
work has been focused on psychometrics, educational statistics, discrete
choice models, statistics applied to social sciences. Her present areas of
interest are also Bayesian statistics  and data science.

*Link da videoconferência:* https://meet.google.com/ujj-vyvs-ocs


As palestras do grupo de MVL ocorrem mensalmente de forma remota e são
publicadas na página: https://mvl.icmc.usp.br/seminars/.
Contamos com a participação de vocês.

Atenciosamente,
-------------- Próxima Parte ----------
Um anexo em HTML foi limpo...
URL: <http://lists.ime.usp.br/pipermail/abe/attachments/20221201/5f680675/attachment.htm>