Survival time in higher education program after a dropout using modified survival function: A retrospective study to predict average graduation time and factors leading to early dropout.

Abstract

  Introduction: The case recurrence survival model in terms of case retention after at least one dropout is still difficult to investigate, and some concrete framework is required to derive the survival model in order to achieve more precise results and reduce the magnitude of bias that may occur during dropout, reappearance, and retention either until the last or dropout again.   Design: Retrospective, longitudinal study   Place and duration: The study was conducted in the School of Mathematical Sciences, College of Computing, Informatics, and Mathematics, Universiti Teknologi MARA, Malaysia, from December 15, 2023, to April 14, 2024.   Material and method: Data for undergraduate program students spanning eight years was retrieved from the College of Dentistry, Imam Abdulrahman Bin Faisal University. It included students of any gender or any age group with at least 50% attendance in the first semester following the commencement of the program. Students who were expelled from the program based on violation or disciplinary action, deportation, imprisonment for criminal acts, or those with special needs or disabilities were excluded. Three possible events of survival function: Retention: Attended the program in continuation till the end. Dropout: Discontinuation of the program for a period of one or more semesters (or >6 months) with a zero-grade point average (GPA). Retention after one dropout: Resumption of program after one dropout and retained in continuation till the end. All relevant information was entered into the data worksheet of SPSS-29.0 (IBM product, USA). Syntax programming of the survival algorithm was developed using the statistical programming software R version 4.2.1, and survival parameters were generated.   Results: The survival probability of the existing model compared to the modified function showed minimal differences. The survival rate was 94% in the first year of study, with a gradual decline of 1%–3% annually, reaching 91.6% by the end of the fifth year. The average survival time for the existing survival function was 4.666 ± 6.70 years, whereas the modified function exhibited a higher average of 5.584 ± 8.63 years. Similarly, the mean graduation time was slightly higher for the modified function (6.10 ± 0.302 years) compared to the existing model (6 ± 0.0 years). Due to data confidentiality, only two variables were included as covariates in the Cox regression analysis: gender and reason for dropout. Among these, the reason for dropout was identified as a significant factor influencing student survival. Model performance, assessed using the R² value, indicated that the modified survival model was more accurate and preferable compared to the existing model (i.e., 0.903 vs. 0.808).   Conclusion: It was concluded that dropout cases, which were censored in the existing survival model, played a significant role in estimating students’ survival time and the program’s graduation time. Hence, the modified function can be preferred when the first event to time doesn’t represent the final outcome. Keywords: Survival analysis, student dropout, higher education, retention, recurrence