Leveraging past data to support interactive automated feedback in undergraduate research proposals

Chalal, ML ORCID: https://orcid.org/0000-0002-2136-8862, Bezai, NE ORCID: https://orcid.org/0000-0002-5982-8983, Medjdoub, B ORCID: https://orcid.org/0000-0002-3402-4479 and Crossley, B, 2025. Leveraging past data to support interactive automated feedback in undergraduate research proposals. The International Journal of Assessment and Evaluation, 32 (1), pp. 185-209. ISSN 2327-7920

Preview

Text 2430097_Chalal.pdf - Post-print Download (1MB) | Preview

Abstract

In academia, educators have been generating substantial amounts of written feedback for many years, yet much of it remains archived and underutilized, limiting its potential to inform and enhance teaching and assessment practices. This study uses past feedback data to develop an interactive, automated feedback tool aimed at addressing feedback inconsistencies during the research proposal stage of an undergraduate research module. To achieve this, we employed a mixed-methods methodology in two phases. First, we developed an automated feedback tool using content analysis of five years' historical feedback data, creating a database of scenarios aligned with research proposal assessment criteria. In phase 2, we applied this tool to provide feedback to 50% of forty-two BSc Architectural Technology students at Nottingham Trent University, United Kingdom, while the remainder received manual feedback. Students' perceptions of feedback relevance and usefulness were assessed via a survey questionnaire across both groups. Mann-Whitney tests were used to analyze the results, comparing the effectiveness of automated versus manual feedback. The analysis of the results suggested that automated feedback from FeedAssist was seen more relevant and useful due to its detailed, contextualized nature with educator personalization. Furthermore, it improved students' key research competencies compared to previous years before its implementation. This study emphasizes that neither approach is inherently superior; rather, each has its own merit and applicability, depending on the specific context and circumstances. FeedAssist effectively balances automation with personalization, offering significant advantages for large cohorts and limited-resource contexts. The study's originality lies in using past feedback data to develop an automated tool that allows educators to guide feedback generation through interactive historical scenarios.

Item Type:	Journal article
Publication Title:	The International Journal of Assessment and Evaluation
Creators:	Chalal, M.L., Bezai, N.E., Medjdoub, B. and Crossley, B.
Publisher:	Common Ground Research Networks
Date:	30 June 2025
Volume:	32
Number:	1
ISSN:	2327-7920
Identifiers:	Number Type 10.18848/2327-7920/CGP/v32i01/185-209 DOI 2430097 Other
Divisions:	Schools > School of Architecture, Design and the Built Environment
Record created by:	Laura Borcherds
Date Added:	28 Apr 2025 08:06
Last Modified:	28 Apr 2025 08:06
URI:	https://irep.ntu.ac.uk/id/eprint/53464

Actions (login required)

Edit View

Statistics

Views

Downloads