While existing research extensively explores higher education enrollment decisions, the transition from bachelor's to master's degrees, especially in STEM, remains understudied. This paper addresses this gap by analyzing the factors influencing final-year bachelor students' intentions to pursue master's degrees, focusing on the immediate post-graduation decision. Master's programs provide opportunities for deeper specialization and advanced research skills, offering significant individual and societal benefits, including advancements in science and technology. The increasing prevalence of master's degree attainment underscores the importance of understanding the underlying mechanisms driving this trend. Building upon Hossler and Gallagher's three-phase college-going model, this study concentrates on the predisposition phase—the initial decision to pursue further education. Unlike earlier research emphasizing student persistence, this study views the master's degree decision as a distinct choice among various post-graduation options. Educational decisions are multifaceted, influenced by individual capabilities, aspirations, and environmental conditions. These encompass a wide array of individual, institutional, and economic factors, including demographics (gender, race, age, residence), academic performance, parental education, socioeconomic status, educational debt, labor market participation, university characteristics (quality, type), and economic context (expected earnings, unemployment, business innovation). The literature reveals significant heterogeneity among master's students, influenced by these varied factors. For instance, STEM fields often exhibit underrepresentation of women and minorities, while higher academic achievement generally correlates with a greater likelihood of pursuing further education. Furthermore, parental education and family income significantly impact educational attainment. Institutional factors such as university quality and type also play a crucial role. The field of study itself significantly influences master's program pursuit, with STEM students historically showing a higher propensity than non-STEM counterparts. Human capital theory posits that educational investment is driven by expected economic returns, linked to labor market factors like unemployment rates and anticipated earnings. Local economic conditions, including business demography and innovation, also significantly impact returns to education, influencing student choices. This study integrates human capital theory with cultural and social capital to explore how these factors, particularly the business innovation environment, shape the contrasting educational choices of STEM and non-STEM students. Understanding these influences is crucial for developing effective strategies to increase participation in STEM master's programs, vital for driving innovation and economic growth.

The literature review comprehensively examines existing research on factors influencing educational choices, specifically the decision to pursue master's level education. It highlights the influence of individual characteristics, such as gender, age, residence, race, ethnicity, academic performance, parental education, socioeconomic status, and educational debt. Studies referenced demonstrate how these factors contribute to the likelihood of enrolling in graduate programs. The review also explores the impact of institutional characteristics, including university quality, type of university (research-intensive vs. teaching-oriented), and field of study (STEM vs. non-STEM). Findings from past research indicate significant differences in the propensity to pursue graduate studies across various fields, with STEM often attracting a higher proportion of students. The review then delves into the role of economic context, focusing on factors such as expected earnings upon degree completion, unemployment rates, and business demography and innovation. These economic factors are analyzed within the framework of human capital theory, emphasizing the relationship between investment in education and expected financial returns. The review integrates various theoretical models of educational choice, including those that incorporate cultural and social capital alongside human capital. Existing research, including the authors' prior work, is cited to support the study's rationale and proposed model. The literature review concludes by identifying a key gap in existing research: the interaction between the business innovation environment and the educational choices of STEM and non-STEM students.

To test the proposed two-level model, the study integrated data from various sources. Data on student intentions to pursue master's degrees in the 2020/2021 academic year was collected via a questionnaire survey administered in 2019 to 502 final-year bachelor students across ten Romanian public universities. The sample comprised 250 STEM and 252 non-STEM students. University quality was assessed using performance scores from a national university ranking. The survey also gathered information on individual and background characteristics, including academic performance, parental education, subjective family income, gender, age, area of residence, employment status, and work experience. Students also provided their subjective expectations regarding the economic benefits of a master's degree. The second level of the model incorporates regional-level data on business innovation, specifically the proportion of enterprises introducing product and/or process innovations, sourced from the Romanian National Institute for Statistics. A multilevel logistic regression analysis was employed, acknowledging the hierarchical data structure (individuals nested within universities, located within regions). The analysis proceeded in stages. The first stage involved estimating a baseline random intercept model with no explanatory variables to determine the suitability of a multilevel approach. The second stage introduced individual-level variables (student characteristics) to assess their impact on master's enrollment intentions. The third stage incorporated regional-level variables (business innovation indicators) to evaluate their influence. The fourth stage included cross-level interactions to explore the possibility that the effect of individual-level variables might vary across regions. Random intercept models were used initially, assuming constant coefficients across regions. Subsequently, random slope models were employed, allowing both the intercept and the coefficient of the field of study (STEM vs. non-STEM) to vary across regions. Likelihood ratio tests were performed to assess the significance of random slopes and cross-level interactions. The detailed specification of the models, including the logit transformation and interpretation of coefficients, is provided in the paper.

The study's findings reveal several key aspects of master's program enrollment intentions. Overall, 53.6% of students expressed intention to enroll in a master's program, with STEM students showing a higher propensity (62%) compared to non-STEM students (45.2%). Significant regional variations were also observed. Multilevel modeling confirmed the suitability of this approach, indicating substantial between-region (2.77%) and between-university (19.75%) variation in enrollment intentions. Random intercept models incorporating individual-level characteristics showed that higher grades, higher parental education (father's education), full-time employment, and shorter work experience positively influenced enrollment intentions. Interestingly, perceived economic benefits of master's degrees (wages and unemployment) did not significantly impact intentions. University performance was also found to have a positive impact. When regional-level variables were introduced, the proportion of enterprises introducing product innovations exhibited a negative effect on master's enrollment intentions. This suggests that more dynamic business environments may provide alternative incentives, diverting students from further education. Crucially, a significant cross-level interaction was observed between field of study and business innovation. The impact of business innovation on enrollment intentions was more pronounced for STEM students than non-STEM students, indicating that STEM students' decisions are more sensitive to local economic conditions. Random slope models, which allowed the effect of STEM vs. non-STEM to vary across regions, further strengthened this conclusion, demonstrating greater regional variation in enrollment intentions for STEM students. This highlights the complex interplay between individual choices and regional context in shaping educational decisions. The study also revealed regional variations in enrollment propensities; the Bucharest-Ilfov region exhibited the highest probability.

The findings address the research questions by demonstrating that the effects of STEM vs. non-STEM majors on master's enrollment intentions differ across regions, and that the business innovation context interacts with field of study in shaping these intentions. The significant regional variation highlights the influence of local economic opportunities on students' decisions. The negative effect of business innovation intensity on enrollment, particularly among STEM students, suggests that attractive immediate employment opportunities in innovative sectors may outweigh the perceived long-term benefits of graduate education. This challenges the traditional view of a straightforward relationship between human capital investment and economic returns, underscoring the importance of contextual factors. The findings contribute to the literature on educational choices by highlighting the complexities of individual decisions within varied economic and regional contexts. The multilevel modeling approach allows for a nuanced understanding of the interplay between individual, institutional, and regional factors. The study's emphasis on the differential influence of business innovation on STEM and non-STEM students provides valuable insights for policy-makers.

This study advances our understanding of master's degree enrollment decisions by demonstrating the significant interplay between individual characteristics, institutional factors, and the regional business innovation environment. The findings emphasize the higher propensity of STEM students to pursue master's degrees, but also reveal that this propensity is particularly sensitive to the attractiveness of immediate employment opportunities in regions with highly innovative business ecosystems. Further research should explore longitudinal data to track long-term career outcomes of STEM and non-STEM graduates and employ mixed methods approaches incorporating qualitative data to gain a deeper understanding of student motivations.

The study's geographical focus on Romania limits the generalizability of the findings to other contexts. The reliance on self-reported data may introduce biases. The cross-sectional nature of the data prevents causal inferences. Future research should address these limitations by employing longitudinal studies, incorporating data from diverse countries, and utilizing mixed-methods approaches.