Differences in Academic Persistence Intentions among STEM Undergraduates in South Korea: Analysis of Related and Influencing Factors

Abstract

In recent years, a decline in employment rates has been observed among science, technology, engineering, and mathematics (STEM) graduates in South Korea, influenced by diverse factors, including economic, social, and policy-related factors. Understanding the reasons behind this decline in STEM employment requires an exploration of academic persistence intentions in STEM and their interconnected relationships with relevant factors. This study aimed to comprehensively examine differences in academic persistence intentions, environmental factors, achievement-related factors, and career motivation among 2393 STEM undergraduates in Korea. Furthermore, this study investigated the factors affecting academic persistence intentions. Data were analyzed using different statistical methods, including factorial multivariate analysis of variance, Pearson’s correlation analysis, and multiple regression models conducted by adding variables of interest. The findings revealed significant differences in academic persistence intentions, environmental factors, achievement-related factors, and career motivation as perceived by STEM undergraduates, based on individual background and university characteristics. Moreover, 53.9% of academic persistence intentions were explained by 10 variables (gender, career direction after graduation, career field to enter after graduation, start period of career path preparation [career-related factors], contextual supports and career barriers [environmental factors], engineering self-efficacy, outcome expectations, major interest [achievement-related factors], and career motivation). Specifically, career motivation contributed the most to the academic persistence intentions of STEM undergraduates, followed by engineering self-efficacy and career barriers. The findings underscore the necessity for customized educational and policy support, considering individual background and university characteristics, to effectively address the challenges faced by STEM graduates in the changing employment landscape.

1. Introduction

In recent years, it has been observed that the employment rate of science, technology, engineering, and mathematics (STEM) graduates in South Korea has declined (as of 2020 compared to 2015, men 70.2% → 64.7%, women 64.1% → 58.1%), with employment rates also revealing an exacerbation of the gender gap. This phenomenon is evident across most STEM majors and regions of the country [1]. Additionally, the recent global impact of COVID-19 on women’s employment in STEM is predicted to further widen the gender gap in employment rates [2]. Employment rates can be influenced by various economic, social, and policy factors such as domestic and international economic conditions, changes in technology and social structure, regional factors, and government employment policies. However, in addition to the macro level, factors at the individual level, including college experiences, career decision-making-related factors, and academic persistence intentions in STEM should also be considered. For example, a positive educational experience at university has a favorable impact on future career advancement in STEM fields [3,4]. Furthermore, career decisions and nondecisions in related fields as well as academic persistence in STEM are not aspects that can be addressed in an instant and are influenced by diverse factors; indeed, learners in the 21st century are influenced by more diverse factors in this regard [5].

In South Korea, students undergo a structured and demanding education process starting with six years of elementary school, followed by six years of middle and high school. High school students can choose between liberal arts and science tracks, with a strong emphasis on academic achievement and college entrance exams due to fierce competition for admission to prestigious universities. Many people strive to enter major cities for better educational opportunities, leading to a concentration of resources in urban areas and challenges for rural students in accessing high-quality education. The Korean educational system tends to prioritize academic achievement over diverse development, particularly for those pursuing STEM fields. High school students aiming for STEM majors focus on subjects such as mathematics and science and undergo rigorous preparation for university admission. Admission is more competitive in STEM fields than in liberal arts fields; however, it offers more opportunities and prepares students for leadership roles in the technology and science fields.

Given that STEM is a field in which women are underrepresented, and academic persistence intentions and actual academic persistence in STEM are a global concern, numerous studies exploring these topics have been conducted. For instance, several studies on the gender gap in academic persistence in STEM [6,7] and persistence in STEM after graduating from college [8,9,10,11] have consistently found that women are more likely to leave STEM fields. Conversely, no gender differences have been found in the intention to leave STEM fields [12]. Furthermore, gender differences in academic persistence among STEM majors do not significantly contribute to the overall gender gap. In this context, to attain a comprehensive understanding of the gender gap in STEM, further research on various stages of the lifecycle would be valuable [13].

The academic persistence intention of undergraduates refers to their intention to continue their current major and enroll in a relevant educational institution in the next semester to complete their studies [14,15,16], which also applies to STEM undergraduates. As the intention to perform a certain action often becomes a prerequisite for the actual occurrence of that behavior, students’ intention to persist in their studies can be considered a predictor of actual academic persistence. Regarding factors affecting persistence in STEM, numerous studies have explored critical contextual, cultural, and cognitive factors affecting students’ entrance into, persistence in, and degree completion within STEM disciplines [6,7,17,18,19]. For example, factors affecting persistence in STEM include good high school grade point average (GPA) [20], good GPA in university [6], high levels of parental education and income, high levels of parental encouragement [6,21], self-efficacy [7,22,23], and college experiences [6,21,24]. Additionally, in the process of making a career choice, according to the social cognitive career theory, representative environmental factors affecting career goal achievement and academic persistence are contextual supports and career barriers [18,25,26]. Furthermore, in the STEM field, engineering self-efficacy is considered an integral factor for achieving and maintaining successful academic performance [27,28]. From another perspective, university institutional characteristics such as size, control, type, and campus climate can affect academic persistence in STEM [29], and students who select STEM majors at four-year colleges and universities are more likely to persist than those in other types of institutions [6,20]. Overall, students’ academic choices related to STEM education have gained increasing scholarly attention in the last decade [17,18,30].

Nonetheless, to understand and address the factors underlying the decline in employment rates in the STEM field, it is crucial to not only promote undergraduates’ persistence in STEM during college but also comprehensively examine differences in academic persistence intentions based on diverse individual backgrounds (i.e., gender, grade, and GPA) and university characteristics (i.e., major field, university location, and departments with a significant gender employment gap). Although prior studies have shed light on some of the factors that influence persistence in STEM fields, the relationships between STEM undergraduates’ academic persistence intentions and environmental factors, achievement-related factors, and career motivation have not been fully explained from a more integrated perspective, despite their importance. To address this gap, the present study involved an empirical analysis to examine differences in academic persistence intentions and related factors, such as environmental factors (i.e., contextual supports and career barriers), achievement-related factors (i.e., engineering self-efficacy, outcome expectations, and major interest), and career motivation among STEM undergraduates based on individual background and university characteristics. Second, the factors affecting academic persistence intentions in STEM were investigated, focusing on diverse variables.

The findings of this study are expected to provide insights into related education and program development for STEM major undergraduates, inform career guidance efforts for this demographic, and present implications for policy support. This study also seeks to provide suggestions for enhancing employment rates in STEM fields and mitigating gender disparities in employment. The specific research questions are as follows.

• How do academic persistence intentions and related factors such as environmental factors (contextual supports and career barriers), achievement-related factors (engineering self-efficacy, outcome expectations, and major interest), and career motivation among STEM students differ by individual background (gender, grade, and GPA) and university institutional characteristics (major, university location, and departments with a significant gender employment gap)?
• Which factors are significant predictors of STEM undergraduates’ academic persistence intentions?

2. Literature Review

2.1. Academic Persistence Intentions

To increase the employment rates of graduates in STEM and boost the women workforce, it is essential to encourage students to enter STEM fields; however, it is also critical to ensure that undergraduates who have chosen a major in STEM do not drop out or stop majoring and maintain their career path in the field. Previous research has addressed the issue of college student attrition and investigated actual academic persistence, with academic persistence intentions identified as an effective predictor of the latter [14,31,32].

Concerning the factors affecting academic persistence intentions, Gayles and Ampaw [6] found that educational experiences, including faculty interactions and peer social involvement, influenced academic persistence in STEM. Gray and Hackling [33] reported that satisfaction with the learning environment and academic experiences can affect persistence intentions, mediated by self-efficacy. Additionally, career factors such as motivation, exploration, and self-efficacy are closely linked to academic persistence intentions [34,35]. As predictive variables influencing the academic persistence of engineering students, reported factors are based on the social cognitive career theory [18,25,26]: Academic persistence intentions are found to be influenced by environmental factors, including contextual supports and career barriers, with major self-efficacy, encompassing individual interest and outcome expectations, identified as a crucial factor affecting career development and academic persistence [25,26,36,37].

However, relatively few studies have examined how career barriers affect academic persistence intentions in STEM, including overall career development [25,26]. Considering the social and academic environments in STEM, women may experience more difficulties, which may be linked to their decision not to persist [6,23,29,38]. Furthermore, although academic persistence intentions can be influenced by factors such as the scale, type, and campus climate of the university [6], little is known about the differences in academic persistence intentions and the related influencing factors based on university institutional characteristics.

2.2. Contextual Supports and Career Barriers

In the process of making career choices, contextual supports and career barriers can be among the environmental factors affecting career goal achievement and academic persistence [25,39]. Contextual supports, a broader concept than social support, refer to the various types of support that individuals receive from their immediate environment in the context of academic and career-related issues [39]. Environmental factors, including the influence of others (e.g., social support, contextual supports, role models, and mentors), reportedly have a positive effect on engineering self-efficacy and academic positivity [25,40], as well as major persistence goals and outcome expectations [41,42]. This is because contextual supports increase engineering self-efficacy, which in turn increases the goal of continuing the major through outcome expectations or major (engineering) interest [43]. Similarly, engineering students’ social interactions (between professors and students or among students) can also increase their major-related self-efficacy, which has a positive effect on major persistence [44]. If students perceive positive support in social interactions, their self-efficacy in engineering tasks increases, promoting their intentions to continue their major. One study [34] found that women generally receive higher levels of support than men, with their engaging actively in campus life and valuing friendships. Given these findings, the generalization of gender differences in the context of environmental factors is premature. However, contextual supports have a stronger impact on women’s self-efficacy than that of men [23]. Nonetheless, women in STEM face challenges owing to a lack of role models and mentors. Successful role models and mentors enhance career motivation, inspire academic success visions, and encourage major immersion [3,7]. A prior study [30] indicated that contextual supports and career barriers also contribute significantly to these intentions to persist. Furthermore, Archer et al. [45] introduced the concept of science capital, encompassing economic, social, and cultural factors that are particularly relevant to science.

Career barriers, as a construct, encompass both internal psychological factors and external environmental factors related to the phenomenon in which an individual’s career path is interrupted and career aspirations or motivation are frustrated or suppressed [46]. Perceived career barriers hinder the achievement of job or career plans [47]. Career barriers can have a direct or indirect negative impact on the academic persistence intentions of STEM students to continue their studies [25,36,41] and, in particular, have a direct effect on engineering self-efficacy, regardless of gender [25]. Furthermore, career barriers significantly impact career decisions; the higher an individual perceives these barriers, the more challenging it becomes to make decisions, thus limiting career choices [47,48]. The awareness of career barriers is negatively correlated with the level of career decision-making, career decision self-efficacy, career exploration, and career preparation behavior [46].

Concerning gender differences in career barriers, prior research has found that compared to men, women in STEM fields often perceive higher career barriers [49]. This is attributed to the male-centric STEM education environment, which potentially leads to a decline in women’s confidence and major self-efficacy, resulting in major changes, abandonment of their field of study, and an overall reinforcement of career barriers. However, a contrasting study [38] reported no differences in career barriers, including with regard to confidence, expectations, and interest, between men and women in STEM fields. Therefore, it is necessary to provide interventions such as career education and counseling to improve the career identity and career resilience of women STEM undergraduates to overcome psychological and environmental career barriers and facilitate career preparation behavior.

2.3. Engineering Self-Efficacy, Outcome Expectations, and Major Interest

Achievement-related factors, such as engineering self-efficacy, outcome expectations, and major interest of STEM students, are closely related to academic persistence intentions [50,51]. Self-efficacy is domain-specific and represents an individual’s confidence in their ability to perform specific behaviors or tasks [52]. Numerous studies have identified it as a predictor of academic achievement and performance [53,54,55] and it is closely linked to persistence in career-related activities within a major as well as academic achievement [56,57]. Similarly, engineering self-efficacy, which reflects confidence in one’s ability in STEM fields, particularly as a scientist or engineer [53,58], is linked to academic persistence, increased commitment to an engineering career [50], and notable performance [43,51].

Furthermore, previous studies have found that strong self-efficacy in engineering is a predictor of academic achievement and major interest [59,60]; student achievement, motivation, and persistence [61]; academic intent to persist in engineering [62]; academic success [63,64]; and academic achievement manifested in higher grades [27]. Some researchers attribute specific factors to low engineering self-efficacy: the high dropout rate of underrepresented minority students in engineering programs [65] and low interest and retention in engineering among women [7,34]. Contradictory research results have been reported regarding gender differences in engineering self-efficacy; some prior studies [7,34,66] revealed statistically significant differences in engineering self-efficacy by gender, whereas others [62,67] found no significant differences. Furthermore, research suggests that freshmen are not as confident as seniors in their belief regarding the likelihood of achieving success in their engineering major [34].

Outcome expectations in STEM, a concept based on the expectancy-value theory [68,69], are related to the value of STEM and expectations of success in STEM fields. This theory emphasizes the significance of intrinsic motivation, where success expectations reflect students’ beliefs about their performance in a given task, and subjective task value in this context encompasses importance (attainment value), interest (intrinsic value), and utility (utility value). Individual engagement in a behavior is influenced by the probability or expectancy of achieving a goal through that behavior and the value assigned to that goal. Students with high expectations and confidence generally demonstrate better academic achievement and performance than those with low expectations. They tend to persist in challenging tasks for longer periods [25,42,68]. Engineering self-efficacy contributes to engineering students’ outcome expectations, major interest [25,42], and sense of belonging in the major [44]. According to the expectancy-value model, expectations of success derive from confidence, shaping learners’ motivation for academic decision-making, while assessments of difficulty are influenced by advice from peers, family, and professors. Values and expectations of success play a crucial role in motivating and sustaining engagement in STEM activities and careers. As a predictive factor, compared to men, women engineering students tend to exhibit lower values and expectations of success [68,69,70]. Gender differences in major interest in STEM are sometimes pointed out as a potential factor contributing to major attrition, although an alignment between one’s major and personal interests does not always occur [70].

2.4. Career Motivation

Career motivation, as a type of motivation, refers to an attitude of immersion in one’s career and continuous development of one’s career to achieve one’s career goals [71,72], which is closely related to persistence in STEM [71,73]. In other words, it refers to individuals’ internal power affecting their career decisions and behaviors. The career motivation theory (CMT) [71,73] delineates the structural relationships between college students’ career-related variables, such as situational conditions, career motivation, career decisions, and behavior, and is effective in systematically understanding and explaining career decisions and behaviors [74]. Since its introduction, numerous related studies have provided implications for career education. In general, career motivation comprises three components: career identity (direction of career motivation), career insight (inducing career motivation), and career resilience (sustainability of career motivation). It is influenced by diverse situational conditions and affects career-related decisions and behavior [71,73]. Given the gender differences in career development, as highlighted in prior research [75], analyses have consistently focused on career motivation and its related variables.

Some representative factors affecting career motivation include social support and self-leadership [76], interactions (between professors and students and between colleagues) [24], higher GPA, and high socioeconomic status [77], which are part of situational conditions and might result in the formation of career motivation. In addition, a male-dominated STEM education environment may have a negative impact on the development of career motivation, especially among women students. This is because it is difficult for women engineering students to feel sufficiently supported and welcomed both inside and outside the classroom, and they are easily alienated from the educational environment [78]. Consequently, there are negative impacts on their career motivation related to entering the field of study, as well as a decline in confidence and academic performance, avoidance of engineering, and dropout. Figure 1 depicts the research model established through an analysis of previous studies.

3. Methods

3.1. Data Sources and Sample

The data analyzed in this study were collected as part of an institutional project conducted by the Korea Foundation for Women in Science, Engineering and Technology (WISET). The dataset contains information related to individual backgrounds (gender, grade, and GPA) and university characteristics (major, university location, and departments with significant gender employment gaps). The research variables included environmental factors (contextual supports and career barriers), achievement-related factors (engineering self-efficacy, outcome expectations, and major interest), career motivation, and academic persistence intentions among STEM students in South Korea.

The target population was undergraduate students currently enrolled in four-year universities in both the metropolitan region (Seoul, Incheon, and Gyeonggi) and non-metropolitan regions of South Korea. The sample consisted of 2393 undergraduate students.

Table 1. Regional distribution of universities and survey responses.

Regions and Cities		Number of Universities	Number of Responses	Ratio (%)
Metropolitan regions	Seoul	25	968	40.45
	Gyeonggi	10	170	7.1
	Incheon	3	59	2.47
	Total	38	1197	50.02
Non-metropolitan regions	Gangwon	1	337	14.08
	Daejeon and Chungcheong (Chungbuk, Chungnam, and Sejong)	3	166	6.94
	Gwangju and Honam (Jeonbuk and Jeonnam)	2	272	11.37
	Busan, Daegu, Ulsan, and Yeongnam (Gyeongbuk and Gyeongnam)	6	406	16.97
	Jeju	1	15	0.63
	Total	13	1196	49.98
Total		51	2393	100

presents the regional distribution of universities and survey responses.

Among them, 1189 men students (49.7%) and 1204 women students (50.3%) were included. Approximately 41.9% were freshmen and sophomores, while 58.1% were juniors and seniors. In Korea, most men serve in the military around the ages of 20–22, usually after completing their first or second year of university. This service is a significant factor for men undergraduates as it affects the continuation of their university education. Therefore, the classification of academic years into freshmen and sophomores (1–2 years) and juniors and seniors (3–4 years) takes this into account. Students with low achievement (based on GPA) accounted for 27.8% of the sample, whereas high achievers accounted for 72.2%. The “high” grades are composed of 3.5 to 4.5 points (out of 4.5) and 3.3 to 4.3 points (out of 4.3). The “low (poor)” are composed of less than 3.0 points (out of 4.5), 3.0 to 3.5 points (out of 4.5), less than 2.7 points (out of 4.3), and 2.7 to 3.3 points (out of 4.3). Students in engineering comprised 48.8% of the sample, whereas students in natural sciences accounted for 51.2%. Based on region, the sample was divided equally (50% each from the metropolitan region and non-metropolitan regions). Finally, students in departments with a low gender employment gap constituted 66.8% of the sample, while those in departments with a high gender employment gap constituted 33.2%. Departments with a high gender employment gap in the natural sciences include environmental science, life science, fisheries, forestry, and horticulture; whereas departments with a high gender employment gap in engineering include materials engineering, chemical engineering, materials science, and textiles (WISET, 2022).

3.2. Measures

First, to assess academic persistence intentions, a shortened 4-item scale [79], which was developed in Korea based on prior research [15] and whose validity has been verified, was used. It consists of four questions, and each question is answered on a 5-point Likert-type scale (1 = not at all; 5 = very much). This scale showed good reliability in the current sample, with a Cronbach’s alpha of 0.732. Second, to assess environmental factors, composed of contextual supports and career barriers, a shortened 8-item scale and a shortened 11-item scale, respectively, were used; the scales have been translated into Korean and their validity has been verified in Korea [80]. In general, to measure contextual supports and career barriers, instruments developed by Lent and colleagues [39,41] are widely utilized, for which there is substantial evidence of reliability and validity. In the present study, the instrument used for contextual supports consists of three subdimensional factors: social support, instrumental support, and financial support. The instrument for career barriers is also composed of three subdimensional factors: physical constraint, social influence, and discrimination. Each question is answered on a 5-point Likert-type scale (1 = not at all; 5 = very much). These scales showed good reliability in the current sample, with Cronbach’s alpha values of 0.828 and 0.907 for contextual supports and career barriers, respectively. Third, achievement-related factors comprise three variables: engineering self-efficacy, outcome expectations, and major interest. The instrument used to assess engineering self-efficacy in the current study was a shortened 19-item scale from a prior study [81], developed based on previous research [52,53]. The instrument for engineering self-efficacy consists of three subdimensional factors: effort and satisfaction, major aptitude, and goal and self-confidence. Each question is answered on a 5-point Likert-type scale (1 = not at all; 5 = very much). This scale demonstrated good reliability in the current sample, with a Cronbach’s alpha of 0.83. To assess outcome expectations and major interest, a shortened 8-item scale and a shortened 6-item scale, respectively, based on Lent et al.’s [25,41] works, translated into Korean, and whose validity has been confirmed in Korea [79], were used. Each question is answered on a 5-point Likert-type scale (1 = not at all; 5 = very much). These scales showed good reliability in the current sample, with Cronbach’s alpha values of 0.855 and 0.822 for outcome expectations and major interest, respectively. Fourth, to assess career motivation, a shortened 11-item scale developed based on prior research [71], translated into Korean, and whose validity has been confirmed in Korea [82] was used. It consists of 11 questions based on three subdimensional factors: career identity, career insight, and career resilience, and each question is answered on a 5-point Likert-type scale (1: not at all; 5: very much). This scale showed good reliability in the current sample, with a Cronbach’s alpha of 0.846.

Table 2. Variable source, number of items, and reliability of survey questions.

Variables	Subfactors	Source	Number of Items	Cronbach’s α
Academic persistence intentions	-	Lee (2015), [79]	4	0.732
Contextual supports	Social support	Lee et al. (2008), [80]	3	0.764
	Instrumental support		3	0.749
	Financial support		2	0.656
	Total		8	0.828
Career barriers	Physical constraint	Lee et al. (2008), [80]	5	0.812
	Social influence		2	0.862
	Discrimination		4	0.879
	Total		11	0.907
Engineering self-efficacy	Effort and satisfaction	Lee (2009), [81]	8	0.832
	Major aptitude		6	0.796
	Goal and self-confidence		5	0.778
	Total		19	0.83
Outcome expectations	-	Lee (2015), [79]	8	0.855
Major interest	-	Lee (2015), [79]	6	0.822
Career motivation	Career identity	Kang et al. (2016), [82]	3	0.745
	Career insight		5	0.724
	Career resilience		3	0.614
	Total		11	0.846

shows the variable source, number of items, and reliability of survey questions.

3.3. Data Analysis

The data collected for this study were processed using SPSS (version 28.0) for frequency analysis, EFA, reliability analysis, factorial multivariate analysis of variance (MANOVA), correlation analysis, and multiple regression models conducted by adding variables of interest. AMOS (version 20.0) was used to perform CFA. The three major analysis methods used in this study were MANOVA to analyze the differences in academic persistence intentions and the related variables (environmental factors, achievement-related factors, and career motivation) by individual background and university characteristics, Pearson correlation analysis to investigate the relationships between these variables, and multiple regression models conducted by adding variables of interest to examine the factors affecting academic persistence intentions.

4. Results

4.1. Differences in Academic Persistence Intentions, Environmental Factors, Achievement-Related Factors, and Career Motivation of STEM Students by Group

Table 3. Mean and SD of STEM undergraduates based on the seven variables measured.

Variables		Individual Background						University Characteristics						Total
	Stat	Gender		Grade		GPA		Major Field		University Location		Departments with a Significant Gender Employment Gap
		Men	Women	Freshmen and Sophomores	Juniors and Seniors	Low (Poor)	High (Excellent)	Engineering	Natural Sciences	Metropolitan Area	Non-Metropolitan Area	Departments with a Low Gender Employment Gap	Departments with a High Gender Employment Gap
1.	M	3.83	3.89	3.76	3.94	3.71	3.92	3.86	3.86	3.84	3.88	3.83	3.92	3.86
	SD	0.6	0.61	0.62	0.59	0.61	0.6	0.59	0.63	0.59	0.63	0.61	0.6	0.61
2.1.	M	3.84	3.86	3.74	3.93	3.71	3.91	3.87	3.83	3.81	3.9	3.84	3.88	3.85
	SD	0.65	0.7	0.71	0.63	0.68	0.66	0.66	0.69	0.66	0.68	0.65	0.72	0.67
2.2.	M	3.63	3.54	3.54	3.62	3.41	3.66	3.57	3.6	3.63	3.54	3.61	3.53	3.59
	SD	0.71	0.83	0.76	0.78	0.81	0.75	0.8	0.75	0.68	0.85	0.74	0.83	0.77
2.3.	M	3.5	3.42	3.44	3.48	3.35	3.51	3.47	3.46	3.52	3.41	3.5	3.39	3.46
	SD	0.86	0.86	0.83	0.88	0.85	0.86	0.85	0.87	0.84	0.88	0.86	0.86	0.86
2.	M	3.68	3.63	3.59	3.7	3.51	3.71	3.66	3.65	3.67	3.64	3.67	3.62	3.66
	SD	0.57	0.62	0.61	0.58	0.61	0.58	0.6	0.59	0.58	0.61	0.58	0.63	0.6
3.1.	M	3.11	3.17	3.21	3.09	3.24	3.1	3.09	3.19	3.1	3.18	3.11	3.2	3.14
	SD	0.87	0.79	0.79	0.85	0.72	0.87	0.81	0.85	0.87	0.79	0.86	0.78	0.83
3.2.	M	2.52	2.44	2.59	2.4	2.48	2.48	2.29	2.66	2.53	2.43	2.49	2.45	2.48
	SD	1.1	1.16	1.14	1.12	1.05	1.16	1.07	1.16	1.1	1.16	1.11	1.17	1.13
3.3.	M	2.43	2.58	2.6	2.44	2.52	2.5	2.45	2.56	2.49	2.52	2.5	2.52	2.51
	SD	1.05	0.97	1	1.02	0.96	1.03	1	1.02	1.01	1.01	1.01	1.01	1.01
3.	M	2.76	2.82	2.87	2.73	2.84	2.77	2.71	2.86	2.78	2.8	2.78	2.82	2.79
	SD	0.86	0.79	0.82	0.82	0.73	0.86	0.8	0.84	0.85	0.8	0.84	0.8	0.83
4.1.	M	3.75	3.74	3.66	3.81	3.59	3.81	3.75	3.75	3.75	3.74	3.74	3.78	3.75
	SD	0.55	0.57	0.57	0.54	0.58	0.54	0.57	0.55	0.56	0.56	0.57	0.53	0.56
4.2.	M	3.75	3.59	3.67	3.67	3.61	3.69	3.71	3.63	3.75	3.59	3.71	3.59	3.67
	SD	0.61	0.62	0.65	0.61	0.71	0.58	0.65	0.59	0.6	0.63	0.63	0.53	0.62
4.3.	M	3.66	3.66	3.56	3.73	3.46	3.73	3.61	3.7	3.7	3.62	3.65	3.67	3.66
	SD	0.62	0.63	0.65	0.59	0.7	0.58	0.66	0.59	0.61	0.64	0.64	0.6	0.62
4.	M	3.72	3.66	3.63	3.73	3.55	3.75	3.69	3.7	3.73	3.65	3.7	3.68	3.69
	SD	0.5	0.54	0.53	0.52	0.56	0.5	0.54	0.51	0.51	0.54	0.53	0.51	0.52
5.	M	3.67	3.67	3.66	3.68	3.55	3.71	3.67	3.66	3.7	3.64	3.69	3.63	3.67
	SD	0.61	0.63	0.6	0.63	0.64	0.6	0.59	0.65	0.62	0.62	0.6	0.65	0.65
6.	M	3.62	3.56	3.59	3.59	3.45	3.64	3.58	3.6	3.68	3.5	3.61	3.57	3.6
	SD	0.63	0.7	0.65	0.68	0.69	0.65	0.68	0.65	0.58	0.73	0.66	0.65	0.66
7.1.	M	3.7	3.68	3.61	3.74	3.59	3.73	3.67	3.71	3.7	3.68	3.68	3.71	3.69
	SD	0.66	0.69	0.67	0.68	0.7	0.66	0.67	0.68	0.64	0.7	0.68	0.66	0.67
7.2.	M	3.74	3.78	3.68	3.82	3.63	3.81	3.74	3.79	3.77	3.76	3.75	3.8	3.76
	SD	0.56	0.58	0.6	0.54	0.6	0.55	0.58	0.56	0.55	0.58	0.56	0.59	0.57
7.3.	M	3.67	3.68	3.61	3.72	3.5	3.74	3.63	3.71	3.71	3.63	3.67	3.67	3.67
	SD	0.62	0.67	0.63	0.66	0.66	0.63	0.64	0.66	0.61	0.68	0.63	0.69	0.65
7.	M	3.71	3.72	3.64	3.77	3.58	3.77	3.69	3.74	3.73	3.7	3.71	3.74	3.72
	SD	0.51	0.54	0.54	0.5	0.55	0.5	0.53	0.52	0.51	0.54	0.52	0.54	0.53

1. academic persistence intentions; 2.1. social support, 2.2. instrumental support, 2.3. financial support, 2. contextual supports (total); 3.1. physical constraint, 3.2. social influence, 3.3. discrimination, 3. career barriers (total); 4.1. effort and satisfaction, 4.2. major aptitude, 4.3. goal and self-confidence, 4. engineering self-efficacy (total); 5. outcome expectations; 6. major interest; 7.1. career identity, 7.2. career insight, 7.3. career resilience, 7. career motivation (total).

presents the mean scores of the STEM undergraduates for each of the seven variables by individual background and university characteristics. First, the mean score for academic persistence intentions (total) was 3.86 (SD = 0.61). Second, regarding contextual factors, the mean score for contextual supports (total) was 3.66 (SD = 0.6), and that for career barriers (total) was 2.79 (SD = 0.83). Third, concerning achievement-related factors, the mean score for engineering self-efficacy (total) was 3.69 (SD = 0.52), that for outcome expectations (total) was 3.67 (SD = 0.7), and that for major interest (total) was 3.60 (SD = 0.7). Finally, the mean score for career motivation (total) was 3.72 (SD = 0.5). The results indicated an ordinary level, except for career barriers (M = 2.79; SD = 0.83).

Next, results from the MANOVA are presented in

Table 4. MANOVA results of STEM undergraduates based on the seven variables measured.

Independent Variables	Dependent Variables	Wilks’ Lambda (Λ)	F	df	Univariate
					MS	F	df
Gender	Academic persistence intentions	0.985	4.949 ***	7	0.012	0.037	1
	Contextual supports				0.050	0.158	1
	Career barriers				2.243	3.782	1
	Engineering self-efficacy				3.114	12.384 ***	1
	Outcome expectations				0.004	0.011	1
	Major interest				0.934	2.320	1
	Career motivation				1.342	5.266 *	1
Grade	Academic persistence intentions	0.987	4.380 ***	7	0.317	0.950	1
	Contextual supports				0.004	0.012	1
	Career barriers				1.250	2.107	1
	Engineering self-efficacy				0.044	0.177	1
	Outcome expectations				1.662	4.635 *	1
	Major interest				2.540	6.312 *	1
	Career motivation				0.195	0.764	1
GPA	Academic persistence intentions	0.981	6.575 ***	7	4.736	14.167 ***	1
	Contextual supports				10.813	34.413 ***	1
	Career barriers				4.557	7.684 **	1
	Engineering self-efficacy				6.876	27.339 ***	1
	Outcome expectations				9.451	26.360 ***	1
	Major interest				8.924	22.174 ***	1
	Career motivation				3.303	12.959 ***	1
Major field	Academic persistence intentions	0.991	2.855 ***	7	2.042	6.108	1
	Contextual supports				0.003	0.010	1
	Career barriers				6.891	11.618 **	1
	Engineering self-efficacy				0.013	0.050	1
	Outcome expectations				0.035	0.099	1
	Major interest				0.324	0.805	1
	Career motivation				0.010	0.039	1
University location	Academic persistence intentions	0.978	7.614 ***	7	0.094	0.282	1
	Contextual supports				0.943	3.002	1
	Career barriers				2.385 ×10 −6	0.000	1
	Engineering self-efficacy				2.014	8.007 **	1
	Outcome expectations				4.256	11.8712 **	1
	Major interest				12.510	31.082 ***	1
	Career motivation				0.551	2.161	1
Departments with a significant gender employment gap	Academic persistence intentions	0.986	4.548 ***	7	2.408	7.203 **	1
	Contextual supports				0.214	0.681	1
	Career barriers				0.273	0.460	1
	Engineering self-efficacy				0.003	0.012	1
	Outcome expectations				1.685	4.699 *	1
	Major interest				0.076	0.189	1
	Career motivation				0.446	1.751	1

* p < 0.05, ** p < 0.01, *** p < 0.001.

.

Table 4 shows that first, there were statistically significant differences in academic persistence intentions based on GPA (Λ = 0.981 at the 0.001 level) and departments with a significant gender employment gap (Λ = 0.986; 0.01 level). Second, there were statistically significant differences in contextual supports based on GPA (Λ = 0.981; 0.001 level). Third, there were statistically significant differences in career barriers based on GPA (Λ = 0.981; 0.001 level) and major field (Λ = 0.991; 0.01 level). Fourth, there were statistically significant differences in engineering self-efficacy based on gender (Λ = 0.985; 0.001 level), GPA (Λ = 0.981; 0.001 level), and university location (Λ = 0.978; 0.01 level). Fifth, there were statistically significant differences in outcome expectations based on grade (Λ = 0.987; 0.001 level), GPA (Λ = 0.981; 0.001 level), university location (Λ = 0.978; 0.01 level), and departments with a significant gender employment gap (Λ = 0.986; 0.01 level). Sixth, there were statistically significant differences in major interest based on grade (Λ = 0.987; 0.001 level), GPA (Λ = 0.981; 0.001 level), and university location (Λ = 0.978; 0.01 level). Finally, there were statistically significant differences in career motivation based on gender (Λ = 0.985; 0.001 level) and GPA (Λ = 0.981; 0.001 level).

4.2. Factors Affecting Academic Persistence Intentions

4.2.1. Relationships between Academic Persistence Intentions and Related Variables

The Pearson correlation analysis results, presented in

Table 5. Correlation results based on the measured variables.

	1	2.1.	2.2.	2.3.	2	3.1.	3.2.	3.3.	3	4.1.	4.2.	4.3.	4	5	6	7.1.	7.2.	7.3.	7	a.	b.	c.	d.	e.	f.
2.1.	0.573 **	1
2.2.	0.332 **	0.427 **	1
2.3.	0.311 **	0.342 **	0.496 **	1
2	0.516 **	0.754 **	0.846 **	0.747 **	1
3.1.	−0.229 **	−0.219 **	−0.087 **	−0.250 **	−0.225 **	1
3.2.	−0.328 **	−0.319 **	0.052 *	−0.01	−0.114 **	0.568 **	1
3.3.	−0.254 **	−0.245 **	0.059 **	0.016	−0.069 **	0.598 **	0.721 **	1
3	−0.299 **	−0.288 **	−0	−0.110 **	−0.162 **	0.864 **	0.829 **	0.897 **	1
4.1.	0.656 **	0.575 **	0.470 **	0.337 **	0.593 **	−0.178 **	−0.219 **	−0.158 **	−0.206 **	1
4.2.	0.469 **	0.381 **	0.477 **	0.378 **	0.530 **	−0.080 **	−0.04	0.019	−0.04	0.638 **	1
4.3.	0.521 **	0.479 **	0.468 **	0.320 **	0.546 **	−0.096 **	−0.04	−0.075 **	−0.087 **	0.708 **	0.574 **	1
4	0.626 **	0.545 **	0.541 **	0.396 **	0.637 **	−0.133 **	−0.108 **	−0.079 **	−0.122 **	0.889 **	0.850 **	0.876 **	1
5	0.562 **	0.494 **	0.500 **	0.464 **	0.619 **	−0.192 **	−0.135 **	−0.061 **	−0.148 **	0.601 **	0.601 **	0.543 **	0.667 **	1
6	0.519 **	0.436 **	0.492 **	0.357 **	0.552 **	−0.072 **	−0.01	0.02	−0.03	0.644 **	0.701 **	0.580 **	0.736 **	0.623 **	1
7.1.	0.533 **	0.418 **	0.386 **	0.350 **	0.491 **	−0.139 **	−0.138 **	−0.121 **	−0.151 **	0.561 **	0.514 **	0.556 **	0.624 **	0.527 **	0.533 **	1
7.2.	0.571 **	0.470 **	0.391 **	0.279 **	0.490 **	−0.114 **	−0.165 **	−0.150 **	−0.160 **	0.674 **	0.450 **	0.619 **	0.663 **	0.534 **	0.521 **	0.604 **	1
7.3.	0.461 **	0.404 **	0.450 **	0.360 **	0.520 **	−0.146 **	−0.061 **	−0.077 **	−0.116 **	0.588 **	0.486 **	0.568 **	0.627 **	0.562 **	0.568 **	0.535 **	0.575 **	1
7	0.622 **	0.513 **	0.478 **	0.381 **	0.587 **	−0.154 **	−0.150 **	−0.142 **	−0.171 **	0.725 **	0.565 **	0.689 **	0.755 **	0.636 **	0.633 **	0.827 **	0.895 **	0.806 **	1
a.	0.046 *	0.017	−0.057 **	−0.046 *	−0.04	0.040 *	−0.04	0.072 **	0.041 *	−0.01	−0.130 **	−0	−0.056 **	0.002	−0.041 *	−0.02	0.037	0.008	0.013	1
b.	0.110 **	0.136 **	0.045 *	0.03	0.090 **	−0.050 *	−0.057 **	−0.04	−0.054 **	0.113 **	−0.02	0.125 **	0.081 **	0.008	−0.01	0.083 **	0.121 **	0.070 **	0.112 **	0.03	1
c.	0.155 **	0.136 **	0.145 **	0.079 **	0.157 **	−0.075 **	0	−0.01	−0.04	0.176 **	0.060 **	0.196 **	0.164 **	0.119 **	0.127 **	0.094 **	0.141 **	0.166 **	0.158 **	0.076 **	0.167 **	1
d.	0	−0.02	0.027	0.069 **	0.03	−0.01	0.014	0.041 *	0.018	−0	0.058 **	0.034	0.036	0.109 **	0.011	0.061 **	−0	0.059 **	0.039	−0.01	0.04	0.011	1
e.	0.404 **	0.327 **	0.182 **	0.165 **	0.287 **	−0.120 **	−0.235 **	−0.100 **	−0.158 **	0.365 **	0.332 **	0.240 **	0.356 **	0.326 **	0.282 **	0.281 **	0.220 **	0.168 **	0.263 **	−0.03	0.172 **	0.083 **	0.183 **	1
f.	0.015	−0.02	−0.068 **	0.001	−0.04	−0.089 **	−0.133 **	−0.116 **	−0.125 **	−0.049 *	−0.074 **	−0.145 **	−0.104 **	−0.062 **	−0.114 **	−0.073 **	−0.070 **	−0.116 **	−0.099 **	−0.054 **	0.183 **	−0.044 *	−0.070 **	0.026	1

*p < 0.05, ** p < 0.01; 1. academic persistence intentions; 2.1. social support, 2.2. instrumental support, 2.3. financial support, 2. contextual supports (total); 3.1. physical constraint, 3.2. social influence, 3.3. discrimination, 3. career barriers (total); 4.1. effort and satisfaction, 4.2. major aptitude, 4.3. goal and self-confidence, 4. engineering self-efficacy (total); 5. outcome expectations; 6. major interest; 7.1. career identity, 7.2. career insight, 7.3. career resilience, 7. career motivation (total); a. gender, b. grade, c. GPA; d. career direction after graduation, e. career field to enter after graduation, f. start period of career path preparation.

, indicate significant correlations between the seven variables as well as their subfactors. Most of the major variables included in this study showed significant correlations (r = ±0.041~±0.755), with the results also indicating significant correlations among their subfactors.

4.2.2. Factors Affecting Academic Persistence Intentions

The basic statistics for the variables included in the multiple regression models, developed to examine the factors affecting academic persistence intentions, are presented in

Table 6. Descriptive statistics for the variables included in the multiple regression models.

Variables		M	SD	Min	Max	Skewness	Kurtosis
Dependent variable	Academic persistence intentions	3.86	0.61	1	5	−0.22	3.08
Independent variables	Contextual supports (Total)	3.66	0.6	1	5	−0.36	3.21
	Career barriers (Total)	2.79	0.83	1	5	−0.23	3.3
	Engineering self-efficacy (Total)	3.69	0.52	1	5	0.2	3.03
	Outcome expectations	3.67	0.62	1	5	0.46	3.76
	Major interest	3.59	0.66	1	5	0.55	3.78
	Career motivation	3.72	0.53	1.27	5	−0.03	0.42
	Variables			N		%
	Gender	Men		1189		49.7
		Women		1204		50.3
	Grade	Freshmen, sophomores		1002		41.9
		Juniors, seniors		1391		58.1
	GPA	Low (poor)		665		27.8
		High (excellent)		1728		72.2
	Career direction after graduation	Employment		1764		73.7
		Graduate school admission in domestic universities		291		12.2
		Graduate school admission in overseas universities		23		1
		Undecided		315		13.2
	Career field to enter after graduation	Major related field		1933		80.8
		Major unrelated field		218		9.1
		Undecided		242		10.1
	Start period of career path preparation	Freshmen		172		7.2
		Sophomores		347		14.5
		Juniors		1105		46.2
		Seniors		769		32.1

. The mean of the dependent variable academic persistence intentions is 3.86 (SD = 0.61), and the independent variables include environmental factors (contextual supports and career barriers), achievement-related factors (engineering self-efficacy, outcome expectations, and major interest), career motivation, career direction after graduation, career field to enter after graduation, and the start period of career path preparation.

Among the independent variables, career direction after graduation concerns the direction of students’ career paths after graduation; employment had the highest weight (73.7%), followed by domestic graduate school enrollment (12.2%), overseas graduate school enrollment (1.0%), and an undecided (undetermined) status (13.2%). Meanwhile, the variable career field to enter after graduation shows that advancement into major-related fields accounts for 80.8%, followed by major-unrelated fields at 9.1. Regarding the start period of career path preparation, 46.2% started in the third year (junior) and 32.1% in the fourth year (senior), indicating a high proportion of late-stage career decision-making at university. To specifically investigate the factors influencing academic persistence intentions, additional multiple regression models were conducted by stepwise adding variables of interest, based on six models, the results of which are presented in

Table 7. Results of multiple regression models conducted by stepwise adding variables of interest.

Independent Variables		Model 1			Model 2			Model 3			Model 4			Model 5			Model 6
		B	SE	ß	B	SE	ß	B	SE	ß	B	SE	ß	B	SE	ß	B	SE	ß
Constant		3.638	0.029		3.269	0.043		1.831	0.070		2.367	0.082		1.384	0.084		1.135	0.085
1	1	0.041	0.025	0.033	0.058	0.023	0.047 *	0.081	0.020	0.066 ***	0.087	0.020	0.071 ***	0.099	0.018	0.081 ***	0.085	0.017	0.070 ***
	2	0.109	0.026	0.086 ***	0.028	0.024	0.022	0.010	0.022	0.008	0.011	0.021	0.009	0.032	0.019	0.025	0.009	0.019	0.007
	3	0.188	0.028	0.138 ***	0.156	0.026	0.115 ***	0.081	0.023	0.059 **	0.077	0.023	0.056 **	0.041	0.020	0.030 *	0.037	0.020	0.027
2	①				−0.105	0.026	−0.076 ***	−0.088	0.023	−0.064 ***	−0.079	0.023	−0.057 **	−0.079	0.020	−0.057 ***	−0.080	0.020	−0.058 ***
	②				0.627	0.030	0.405 ***	0.444	0.027	0.287 ***	0.408	0.027	0.264 ***	0.250	0.025	0.162 ***	0.271	0.024	0.175 ***
	③				0.004	0.028	0.003	0.037	0.025	0.025	0.002	0.025	0.001	0.061	0.022	0.042 **	0.075	0.021	0.051 ***
3	a.							0.439	0.018	0.429 ***	0.413	0.017	0.404 ***	0.084	0.020	0.082 ***	0.060	0.020	0.059 **
4	b.										−0.141	0.012	−0.191 ***	−0.139	0.011	−0.189 ***	−0.125	0.011	−0.170 ***
5	c.													0.372	0.028	0.320 ***	0.231	0.031	0.199 ***
	d													0.165	0.021	0.167 ***	0.121	0.021	0.123 ***
	e													0.081	0.020	0.088 ***	0.053	0.020	0.058 **
6	f																0.289	0.026	0.249 ***
R2 (Adj. R2)		0.032 (0.031)			0.186 (0.184)			0.351 (0.349)			0.385 (0.383)			0.516 (0.514)			0.539 (0.537)
△R2		0.032			0.154			0.165			0.035			0.131			0.023
F		26.705 ***			90.962 ***			184.146 ***			186.849 ***			230.807 ***			232.018 ***

*p < 0.05, ** p < 0.01, *** p < 0.001; 1. gender, 2. grade, 3. GPA; ① career direction after graduation, ② field to enter after graduation, ③ start period of career path preparation, a. contextual supports (total), b. career barriers (total), c. engineering self-efficacy (total), d. outcome expectations, e. major interest, f. career motivation.

.

In the first model, individual background variables (gender, grade, and GPA) were entered as first-stage variables, explaining 3.2% of the academic persistence intentions (p < 0.001). In the second model, career-related variables (career direction after graduation, field to enter after graduation, and start period of career path preparation) were introduced as second-stage variables. The explanatory power increased by 15.4%, explaining 18.6% of academic persistence intentions (p < 0.001). In the third model, the variable of contextual supports was introduced as the third-stage variable. The explanatory power increased by 16.5%, explaining 35.1% of academic persistence intentions (p < 0.001). In the fourth model, the variable of career barriers was introduced as the fourth-stage variable. The explanatory power increased by 3.5%, explaining 38.5% of academic persistence intentions (p < 0.001). In the fifth model, achievement-related variables (engineering self-efficacy, outcome expectations, and major interest) were introduced as fifth-stage variables. The explanatory power increased by 13.1%, explaining 51.6% of academic persistence intentions (p < 0.001). In the final model, career motivation was introduced, resulting in a 2.3% increase in explanatory power and explaining 53.9% of academic persistence intentions (p < 0.001).

Based on the multiple regression analysis results, 10 variables—gender, career direction after graduation, field to enter after graduation, start period of career path preparation, contextual supports, career barriers, engineering self-efficacy, outcome expectations, major interest, and career motivation—were found to affect academic persistence intentions, and the R² was found to be statistically significant at 53.9 (F = 232.018, p < 0.001). Regarding the relative contribution of each independent variable, career motivation (β = 0.249) showed the highest contribution, followed by engineering self-efficacy (β = 0.199), field to enter after graduation (β = 0.175), career barriers (β = −0.170), outcome expectations (β = 0.123), gender (β = 0.070), contextual supports (β = 0.059), major interest (β = 0.058), career direction after graduation (β = −0.058), and start period of career path preparation (β = 0.051). Among these variables, career barriers and career direction were found to negatively affect academic persistence intentions.

5. Discussion

The main findings of this study are as follows. First, academic persistence intentions were found to be significantly higher among students with high GPAs and those belonging to departments with a high gender employment gap. This finding indicating differences in academic persistence intentions by GPA is consistent with the results of previous studies [6], which have reported that a good GPA at university influences academic persistence in STEM. This may be because high-achieving students are more likely to have high engineering self-efficacy and a strong intention to persist in their academic pursuits. Furthermore, while research reporting differences in academic persistence intentions based on the variable “departments with a significant gender employment gap” remains scarce, these differences noted in the current study may be explained by the STEM education and employment environment unique to Korea. Given that this variable is one of the factors reflecting rapidly changing societal trends and that there is a lack of relevant prior research, further studies focusing on its impact on academic persistence intentions in STEM fields are required.

Second, statistically significant differences were noted in contextual supports, one of the environmental factors, based on GPA. Additionally, statistically significant differences were observed in career barriers, another environmental factor, based on GPA and major. Contextual supports were found to be significantly higher among students with high GPAs. This finding regarding differences in contextual supports based on GPA aligns with those of previous studies [25,40,43,44]. This suggests that contextual supports influence engineering self-efficacy and high GPA, and that high engineering self-efficacy affects academic persistence intentions. Furthermore, career barriers were found to be significantly higher among students with low GPAs and those studying natural sciences. Differences in career barriers according to GPA can be explained by considering career barriers as environmental factors that impact engineering self-efficacy and academic achievement, as reported in prior research [36]. However, the discussion regarding differences in career barriers by major is limited by the fact that research exploring career barriers based on the major within STEM fields is scant. Nonetheless, the differences noted in this study may have been influenced by sample characteristics or social and contextual changes over time. Subsequent research should explore these variables further.

Third, regarding achievement-related factors (engineering self-efficacy, outcome expectations, and major interest), there were statistically significant differences in engineering self-efficacy according to gender, GPA, and university location. Engineering self-efficacy was significantly higher among men students, students with high GPAs, and those attending universities in metropolitan areas. These findings are consistent with those in the existing literature. For example, the derived differences in engineering self-efficacy by gender align with the results of prior studies [7,66] reporting higher engineering self-efficacy among men students. Furthermore, numerous studies have reported that strong and positive engineering self-efficacy is significantly correlated with academic achievements manifested in higher grades [27,59,60]. However, research examining differences in engineering self-efficacy based on university characteristics is limited. However, as university location is an environmental factor that influences engineering self-efficacy [25,40,44], the differences noted in this study can be interpreted as being influenced by the specific characteristics of the universities. Moreover, in Korea, the prevailing academic elitism centered around metropolitan areas and regional biases might act as significant contextual factors influencing engineering self-efficacy and the career development process of non-metropolitan university students. Therefore, to enhance engineering self-efficacy among STEM undergraduates, more systematic faculty support interventions should be developed and provided at the university level. Next, outcome expectations were significantly higher among junior and senior students, students with high GPAs, those attending universities in metropolitan areas, and those belonging to departments with a low gender employment gap. Lastly, major interest was significantly higher among junior and senior students, students with high GPAs, and those attending universities in metropolitan areas. That students from metropolitan regions exhibited higher levels of outcome expectations in STEM and major interest can be attributed to the more stable educational environment and advantageous position in the job market created by metropolitan universities. The recent domestic economic downturn, job reductions, unstable employment, and youth unemployment have also added to the real-world challenges faced by contemporary university students. Similarly, environmental conditions, such as the development of infrastructure centered around metropolitan areas, economic stability, and advantageous positions in the labor market, suggest that students in metropolitan regions may have a relatively favorable environment for academic achievement and career exploration. Consequently, non-metropolitan students encounter more difficulties in entering the job market than their counterparts from metropolitan areas and are considered to possess lower qualifications. Additionally, the finding on the differences in outcome expectations based on departments with a significant gender employment gap in Korea might be closely associated with social and economic fluctuations. In Korea, departments with a significant gender employment gap in STEM natural sciences include environmental sciences, life sciences, fisheries, forestry, and horticulture. Similarly, in engineering departments, the gender employment gap is notable in fields such as advanced materials engineering, chemical engineering, materials science, and textiles [1]. Further studies considering changing socioeconomic variables are needed, and more attention should be paid to the interpretation of results on the gender employment gap.

Fourth, career motivation was significantly higher among women students and students with higher GPA. The finding of this study revealing a positive correlation between higher grades and increased career motivation is consistent with prior research [77,83]. These results imply that students with higher career motivation actively engage in major-related classes and achieve higher GPAs [83]. Previous studies have consistently shown that senior college students in Korea exhibit higher levels of career motivation than junior college students [84,85]. This is because, for university students, career preparation behaviors tend to be undertaken urgently as they approach their senior years rather than being systematically initiated during their junior years, resulting in heightened career motivation. However, in this study, no differences were found in career motivation by gender and grade. Therefore, further in-depth studies are required.

Finally, the impact on academic persistence intentions was found to have an explanatory power of 53.9% based on gender, grade, GPA, contextual supports, career barriers, engineering self-efficacy, outcome expectations, major interest, career motivation, career direction, alignment of major with career field, and start period of career path preparation. Especially noteworthy is the finding that career motivation, with the highest contribution among various affecting factors, significantly influences academic persistence intentions. This result aligns with those of previous studies [25,34,35] demonstrating the associations between academic persistence intentions and factors such as career motivation, career exploration behaviors, and career and occupational self-efficacy. This relationship can be explained by the interplay between environmental variables such as contextual supports, career motivation, and academic persistence. In general, learners who perceive support from their surrounding environment tend to exhibit higher motivation, which influences their academic persistence. Correspondingly, learners who believe that they receive positive assistance from those around them show higher career motivation in STEM, thereby enhancing their willingness to persist in academic pursuits.

Based on these results, several educational implications can be derived. First, enhancing career motivation requires the development of individualized and systematic education programs customized for STEM majors, supported by national initiatives for career development. Specifically, as career motivation influences career preparatory behaviors and the career landscape varies across academic disciplines, including career preparatory actions, employment support programs, and regular employment, government policies for career development and employment support should be diversified rather than kept uniform. Second, to enhance contextual supports for STEM majors, various strategies such as mentorship programs with professors, mentoring programs with seniors, study group support programs, and parent information sessions can be more actively implemented at the departmental level in universities. Third, instructional and learning strategies are essential to enhance engineering self-efficacy. In engineering colleges, the learning process generally follows a hierarchical knowledge structure, and if students do not grasp the prerequisite knowledge, they encounter barriers in progressing to the next stages. Many students experience a decline in engineering self-efficacy for their major due to failures in pre-learning, which hinders subsequent major-related learning stages.

This study is noteworthy in its investigation of differences in the academic persistence intentions of 2393 STEM undergraduates and the related factors based on individual background and university characteristics, also involving a comprehensive analysis of the factors influencing academic persistence intentions. However, there were some limitations to this study, providing directions for future research. First, the findings have limited generalizability to all undergraduate students majoring in STEM fields, although the sample size was large and drawn from large universities in Korea. In future work, there is a need to conduct a more in-depth analysis that considers both individual- and university-level variables through multilevel modeling (MLM) or mediating effect verification. This would permit further analyses of the effects of educational interventions in the relationship between academic persistence intentions and the related variables. Second, the data were collected using self-report measures and are quantitative; thus, potential biases cannot be dismissed and the results may not reflect objective realities. Future research needs to add qualitative measurement methods, such as direct measurements through external evaluators and interviews (i.e., a mixed-method research design), to support the present findings with stronger and more concrete evidence. Third, as the academic persistence of STEM undergraduates is not short-term, it would be meaningful to conduct longitudinal research analyzing changes over time in subsequent studies. This would enable a more detailed analysis of the factors influencing academic persistence intentions based on learners’ career development stages. Finally, at this stage of the research, it is not sufficient to draw causal relationships between academic persistence intentions and related factors. Future studies should examine more variables that are possibly related to the academic persistence intentions of STEM undergraduates. Accordingly, more appropriate means of improving STEM students’ academic persistence intentions, with more convincing research conclusions, could be derived.

6. Conclusions

The current findings indicate that the factors related to academic persistence intentions among STEM undergraduates vary based on individual background and university characteristics. Additionally, career motivation, engineering self-efficacy, and career barriers were found to influence the majority of academic persistence intentions among STEM undergraduates. Therefore, to bring about positive changes in academic persistence intentions, it is crucial to explore approaches that enhance career motivation and engineering self-efficacy and address career barriers, all while aligning with the curriculum. Overall, the findings highlight the general importance of academic persistence intentions among STEM undergraduates and contribute to the body of research on the associated factors.