Development of a Longitudinal Method to Measure Attrition Intentions

This research paper describes a novel method to longitudinally investigate factors active in the attrition decision making process of engineering graduate students. Across disciplines, researchers demonstrate a variety of factors that influence attrition. To date, research has depended on cross-sectional and qualitative models to investigate the factors influencing attrition. Previous studies have demonstrated the importance of various factors such as demographic information, satisfaction, and advisor relationships to the intention to drop out or actual departure from graduate programs. These studies were based on data collection at one time point. However, students take time to consider many factors before making the decision to leave engineering programs indicating the departure decision process may fluctuate over time. Therefore, a method that captures the departure decision making process over time is necessary to predict attrition risk. The method developed and described in this paper employed the Graduate Attrition Decisions (GrAD) model to identify variables necessary to measure. The GrAD model identified themes of attribution from graduate students' narratives. The departure decision process involves advisor relationship, support network, quality of life and work, cost perceptions by others, and goals. The method is designed to measure the GrAD variables over a 1-year period using multivariate time series analysis. Questions regarding perceived stress and critical events to observe fluctuation over time will be included. The method gathers data through text message-based (SMS) surveys and SMS invitations to web-based surveys. Data collection can occur at varying time spans to measure factors daily, weekly, monthly, and semesterly. The paper will detail the process of developing the questions, SMS system, and time series analysis. This paper provides a framework for the future research to engage this longitudinal data collection method. The method will allow the development of a model based to understand the trajectory and fluctuation of graduate students' attrition decision making process.