TL;DR
This paper proposes a quantitative framework to evaluate how "quality development training"—a form of experiential learning—impacts the entrepreneurial success of college students. By combining the Analytic Hierarchy Process (AHP) with a "dominant feature recognition model," the authors move beyond theoretical discussion to provide a mathematical tool for measuring the soft skills and professional knowledge essential for startup survival.
What Problem Does This Paper Address?
Despite high enthusiasm, the success rate for college student entrepreneurs in China is remarkably low—averaging only about 1%. While universities offer "quality development" (outward-bound style training and experiential workshops), most existing research on its effectiveness is purely theoretical or focused on specific job placements.
There is a significant gap in empirical quantification: How do we actually measure "competency"? The authors tackle the challenge of turning abstract traits like "perseverance" and "innovation ability" into a weighted, scientific index system that can evaluate both individuals and teams.
Methods
The researchers utilize a multi-layered mathematical approach to build their evaluation model:
- The Iceberg Model: Competency is divided into Explicit (professional, management, and general knowledge) and Implicit (charisma, psychological quality, perseverance, and soft skills).
- Analytic Hierarchy Process (AHP): Experts were consulted to perform pairwise comparisons between 12 different indices to establish a "criteria layer" and an "alternative layer." This helps assign objective weights to different competencies.
- Dominant Feature Recognition Model: To account for individual student preferences and strengths, the authors used a model based on L2-norm distance and solved it using Lagrange multipliers. This allows the system to identify which specific features "dominate" a student's entrepreneurial profile.
- Validation: They applied Harman’s one-factor test to check for Common Method Variance (CMV), ensuring the survey data wasn't biased by the research environment or data source.
- Cluster Analysis: Finally, they used rapid clustering to categorize students into different competency levels.
Datasets
The study relies on empirical survey data collected from college students. While the total sample size ($m$) is used to build the initial matrix, the paper specifically highlights a case study of a 5-member entrepreneurial team (labeled Members A through E) to demonstrate the model's practical application in a real-world team setting.
Key Insights and Ideas
- The Weight of Skills: The model found that "Training of entrepreneurial skills and management/marketing abilities" carried the highest weight (0.425) compared to "Learning of cultural knowledge" (0.3229) and "Improvement of intrinsic qualities" (0.2521). This suggests that while "grit" matters, practical skill application is the strongest predictor of competency.
- Implicit vs. Explicit: The "Iceberg" framing is crucial. The authors argue that while professional knowledge is the visible tip, the submerged implicit qualities—like psychological tolerance and risk management—are what actually prevent the "ship" from sinking.
- Mathematical Objectivity: By using Lagrange multipliers to solve the recognition model, the authors provide a way to minimize subjective bias in self-reporting, creating a more "scientific" benchmark for educators.
Key Results
- Data Validity: The extracted principal components in the CMV test explained 75.24% of all evaluation indices, indicating the data was robust and free of serious systematic error.
- Consistency: The Consistency Ratio (CR) for all categories was well within acceptable limits (e.g., CR = 0.0152 for intrinsic qualities, where < 0.3 is the passing threshold), proving the internal logic of the weighting system.
- Team Dynamics: In the case study, the model successfully identified that Members A and B were the natural leaders with "strongly strong" and "slightly strong" competencies, respectively. Conversely, Member E was identified as "slightly weak," correlating with a lack of dominant features that support entrepreneurial activities.
- Clustering: The study categorized 51.2% of the surveyed students into "Class 1," representing the baseline competency level for the population studied.
Limitations and Future Work
The authors acknowledge that while the model is scientific, it is currently applied to a specific context (Yanching Institute of Technology). Future work could involve:
- Longitudinal Tracking: Measuring if students who score high on this index actually achieve higher business success rates 5-10 years post-graduation.
- Broader Samples: Expanding the dataset across different types of universities (e.g., technical vs. liberal arts) to see if the "dominant features" shift.
- Automation: Integrating this model into an AI-driven platform for real-time student feedback during entrepreneurship workshops.
Summary updated