Data Sheet 1_Automated software scoring of senior school certificate examination mathematical items in economics using a contextual similarity model.docx

Education is a key driver of national development, and technology is transforming assessment methods globally. Although traditional examination formats, such as multiple-choice and essay tests, dominate assessment, subjective human scoring often leads to inconsistencies. Artificial intelligence-based automated scoring systems offer a promising solution. Therefore, this study uses a developed artificial intelligence software based on a contextual similarity model to score mathematical items in secondary school Economics and validates the AI-generated scores against human experts. A correlational research design was adopted for this study. The population comprised 274,978 Economics students from South-west Nigeria, 1,008 of which were sampled through a multi-stage sampling procedure. An AI-based software was developed based on the waterfall model. It was validated by three software experts, and a Cronbach alpha of 0.86 was determined. The data were analysed using descriptive statistics and a Pearson Correlation Coefficient at an alpha level of 0.05 significance. The study’s findings revealed a significant relationship between the AI software and the 12 human experts in scoring Economics Mathematical items. The intra-class correlation coefficient (ICC) indicated a high level of agreement (r = 0.86), P < 0.01). The study concluded that the scoring model possesses the ability to award scores to mathematical computation items in economics that are equivalent to the scores awarded by human expert markers. The study, therefore, recommends that teachers, evaluators, and various examination bodies use this Artificial Intelligence software in scoring mathematical computation items in economics to ensure consistency and save on cost and time.