Understanding dynamic learning profiles in AI-supported environments in Vietnam: Evidence from a three-wave latent transition analysis

This dataset accompanies the study “Understanding dynamic learning profiles in AI-supported environments in Vietnam: Evidence from a three-wave latent transition analysis.” It provides a comprehensive longitudinal dataset designed to examine how university students’ learning processes evolve in AI-supported environments over time. The dataset was collected from undergraduate students in Vietnam and follows a three-wave longitudinal design, including measurements at the beginning (T1), middle (T2), and end (T3) of a semester. The primary aim is to capture dynamic changes in key psychological and behavioral constructs related to AI-supported learning. The dataset includes three core constructs measured across all waves: AI self-efficacy, cognitive offloading, and metacognitive regulation. Each construct is operationalized using four Likert-scale items (1–7), allowing for both item-level and construct-level analyses. In addition, the dataset contains composite scores for each construct at each time point to support descriptive and auxiliary analyses. Background variables are also included, such as prior academic performance (GPA), gender, and field of study (STEM vs. non-STEM). Academic performance outcomes are captured through end-of-semester GPA, enabling analysis of the relationship between learning processes and performance. Missing data due to wave attrition are coded as 999, which is compatible with Mplus and other statistical software. The repository includes a structured codebook that provides detailed descriptions of all variables, along with Mplus syntax files and selected output files to support full reproducibility of the analyses. The analytical framework of the study includes descriptive statistics, measurement model evaluation, latent profile analysis (LPA) at each wave, and profile comparison analyses using covariates and academic outcomes. The dataset is particularly suited for researchers interested in longitudinal modeling, latent variable analysis, and the role of AI in higher education learning processes. This dataset is shared to promote transparency, reproducibility, and further research on AI-supported learning in higher education contexts, particularly in emerging educational systems such as Vietnam.本数据集随附于题为《越南人工智能支持学习环境中的动态学习特征：基于三波潜在转移分析的实证证据》的学术研究。本数据集为综合性纵向数据集，旨在探究高校学生在人工智能支持学习环境中的学习过程随时间的演变规律。 本数据集的调研对象为越南高校本科生，采用三波纵向研究设计，分别于学期初始（T1）、期中（T2）及期末（T3）开展测量。本研究的核心目标是捕捉与人工智能支持学习相关的关键心理与行为构念的动态变化。 本数据集包含所有测量波次均覆盖的三大核心构念：人工智能自我效能感（AI self-efficacy）、认知卸载（cognitive offloading）与元认知调控（metacognitive regulation）。每一项构念均通过4个李克特量表条目（Likert-scale items，1~7分）进行操作化定义，支持条目级与构念级双重分析。此外，数据集还包含各时间点下各构念的合成得分，以支撑描述性与辅助性分析。 数据集同时包含背景变量，如前期学业表现（平均学分绩点，Grade Point Average，GPA）、性别及研究领域（理工科（Science, Technology, Engineering, Mathematics，STEM）与非理工科）。学业表现结果通过期末GPA进行采集，可用于分析学习过程与学业表现之间的关联。 因追踪流失产生的缺失数据以999进行编码，该编码格式兼容Mplus及其他统计软件。本数据集仓库包含结构化编码手册，对所有变量进行详细说明，同时附带Mplus语法文件与部分输出文件，以确保分析过程可完全复现。 本研究的分析框架涵盖描述性统计、测量模型评估、各波次的潜在剖面分析（Latent Profile Analysis，LPA）以及利用协变量与学业结果开展的剖面比较分析。本数据集尤其适用于关注纵向建模、潜变量分析以及人工智能在高等教育学习过程中作用的研究者。 本数据集公开共享，旨在推动高等教育场景下人工智能支持学习相关研究的透明度、可复现性及进一步拓展，尤其针对越南这类新兴教育系统的相关研究。