Premium
Designing a formative e‐assessment: Latent class analysis of early reading skills
Author(s) -
Sainsbury Marian,
Benton Tom
Publication year - 2011
Publication title -
british journal of educational technology
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.79
H-Index - 95
eISSN - 1467-8535
pISSN - 0007-1013
DOI - 10.1111/j.1467-8535.2009.01044.x
Subject(s) - formative assessment , test (biology) , class (philosophy) , reading (process) , psychology , computer science , mathematics education , strengths and weaknesses , latent class model , artificial intelligence , linguistics , machine learning , biology , paleontology , social psychology , philosophy
Computer‐based testing, or e‐assessment, has the potential to deliver immediate results for the benefit of schools. This paper describes a project that aimed to exploit this potential by designing e‐assessments where the results were intended for use by teachers in planning the next steps in teaching and learning: low‐stakes, formative assessment. A pair of tests assessing early reading was developed, with the test items based on a range of distinct skills, including phonological segmentation, rhyming and word recognition. The tests were administered on a screen with the questions presented aurally and visually. In trials, a total of 1345 test results were obtained from pupils aged 5–7 years from 26 schools. Latent class analysis was used to identify patterns of performance within the data. Four latent classes were distinguished, each characterised by a pattern of responses related to the different test items. The strengths and weaknesses in early reading skills implied by each of these latent classes were described in terms of formative ‘profiles’ provided for teachers in an online reporting package together with indicators for the next steps in teaching. The research resulted in an automated marking and analysis system that can be genuinely formative.