Data on students’ perceived educational value of artificial intelligence, smart learning, perceived academic and career progress, and mental well-being in higher education
Description
This dataset contains anonymized survey data collected from university students in Vietnam who had prior experience using artificial intelligence in learning. The deposit includes the original dataset of 1.274 complete responses and an analysis-ready dataset used for the revised PLS-SEM measurement model. The variables cover perceived educational value of artificial intelligence (PEVAI), smart learning (SL), perceived academic and career progress (PACP), mental well-being (MWB), and demographic characteristics. PEVAI items assess students’ perceived usefulness and educational value of AI-supported systems. The code SL3_SC identifies the original smart-learning item containing explicit social-comparison wording. PACP represents self-perceived academic and career progress rather than objective academic performance. All construct items were rated on a five-point Likert scale ranging from 1 (strongly disagree) to 5 (strongly agree). The files contain no names, student identification numbers, email addresses, or other direct personal identifiers. Supporting materials include the codebook, the English and Vietnamese questionnaires, a README describing variable coding and data processing, descriptive statistics, and PLS-SEM measurement-model outputs. The data may be reused for psychometric assessment, group comparisons, educational technology research, and analyses of relationships among AI-related perceptions, smart learning, academic and career progress, and mental well-being. Because convenience sampling was used, the dataset should not be interpreted as nationally representative of all university students in Vietnam.
Files
Steps to reproduce
1. Download the README, codebook, raw dataset, analysis-ready dataset, questionnaires, and analysis-output files. Each row in the data files represents one respondent. 2. Construct items are coded on a five-point Likert scale: 1 = strongly disagree, 2 = disagree, 3 = neither agree nor disagree, 4 = agree, and 5 = strongly agree. 3. Demographic variables are coded as follows: GENDER (0 = male, 1 = female); YEAR_STUDY (2 = second year, 3 = third year, 4 = fourth year, 5 = other year of study); UNIV_TYPE (0 = private university, 1 = public university); MAJOR (1 = education, 2 = economics, 3 = language and linguistics, 4 = engineering and technology, 5 = other); and AI_USE_DURATION (1 = less than 1 year, 2 = from 1 to under 2 years, 3 = 2 years or more). 4. To reproduce the measurement-model analysis, import the analysis-ready data file into SmartPLS [insert software version]. Specify PEVAI using PEVAI1–PEVAI8; SL using SL1, SL2, SL3_SC, SL4, and SL5; PACP using the indicators documented in the final codebook; and MWB using MWB1–MWB6. All constructs are modeled reflectively. 5. Run the PLS algorithm using the settings reported in the manuscript. Assess outer loadings, Cronbach’s alpha, rho_A, composite reliability, average variance extracted, the Fornell–Larcker criterion, and the heterotrait–monotrait ratio. 6. Run bootstrapping using the number of subsamples reported in the manuscript to obtain confidence intervals for HTMT and the relevant model estimates. 7. Use the raw dataset to reproduce the respondent profile and item-level descriptive statistics. The README and data-processing log document variable renaming, retained indicators, excluded indicators or observations, and the rationale for each processing decision. No individual response value was manually altered.