Beyond Vocation: Understanding Sociocultural and Opinion-Based Determinants of STEMM Career Choice in Peruvian Women

Abstract

This study examines the underrepresentation of women in STEMM (Science, Technology, Engineering, Mathematics, and Medicine) within Peruvian public universities and identifies factors associated with women’s program choice. A cross-sectional survey was administered to first-term students across three public institutions spanning Peru’s Highlands, Coast, and Amazon regions. Data from 1142 students (145 women) were used for descriptive analysis of segregation, while an inferential sample (N = 152; 76 STEMM, 76 non-STEMM) was used for modeling. The instrument was an adapted “University Students’ Questionnaire on STEM Studies in Higher Education (QSTEMHE)” (Cronbach’s α = 0.89). Descriptive statistics and a penalized (Firth) binary logistic regression were used to evaluate sociodemographic, contextual/experiential, and motivational predictors of enrolling in a STEMM major. The cross-sectional design limits causal inference, and perception data are subject to self-report biases. Women accounted for 12.7% of STEMM enrolment overall, with pronounced horizontal segregation: engineering programs frequently recorded critically low female participation (≈3–5% in Civil, Mechanical, and Computer Engineering), whereas Medicine and Sanitary Engineering showed comparatively higher representation (27–38%). Perception data indicated that STEMM students more strongly rejected gender–ability stereotypes than non-STEMM peers, although a substantial proportion still reported constraining gender expectations and rigid household roles. In the penalized regression, Prior Interest in STEM (OR = 7.76; p = 0.018) and Motivation: Opportunities (OR = 2.24; p = 0.0001) significantly increased the probability of choosing STEMM. Crucially, Ethnicity emerged as a significant barrier: identifying as ‘Quechua’ (OR = 0.19; p = 0.0004) or ‘Other(s)’ (OR = 0.16; p = 0.011) significantly decreased this likelihood. Age, area of residence, and Motivation: Altruism was not significant. Findings support early, gender-responsive career guidance, mentoring, addressing intersectional ethnic barriers, and targeted financial aid to strengthen women’s participation and retention in STEMM.

1. Introduction

The underrepresentation of women in Science, Technology, Engineering, Mathematics, and Medicine (STEMM) fields remains a persistent global issue that constrains innovation potential and socioeconomic development. Worldwide, women continue to account for only one-third of the scientific community, a figure that has stagnated over the past decade [1]. Recent reports from the Organization for Economic Co-operation and Development (OECD) and the World Economic Forum confirm that although women’s participation in the general labor force is approaching parity, their representation in STEM occupations reaches just 29.2% [2,3]. While regions such as Latin America and the Caribbean display relatively higher parity in research and development (R&D) personnel at 45.3%, critical gaps persist in high-demand disciplines such as engineering and information technology [4,5]. This structural disparity not only restricts women’s professional opportunities but also deprives society of diverse perspectives, which are essential to addressing complex challenges in equitable and innovative ways [6].

In the Latin American context, this disparity is particularly nuanced. The aggregate parity figure [4,5] masks persistent vertical and horizontal segregation, where women remain concentrated in life sciences rather than in high-prestige, higher-paying fields like engineering and computer science [7]. Regional studies emphasize that this gap is cemented early in the educational pipeline, influenced by classroom dynamics and a lack of gender-responsive teaching practices. Furthermore, research in neighboring countries, such as Colombia, confirms that the persistence of sociocultural barriers—including pervasive stereotypes about women’s abilities, a lack of visible female role models in technical sectors, and the perceived conflict between demanding careers and traditional family roles—is a decisive factor in discouraging women from pursuing STEM pathways [8,9].

Scientific evidence has identified a set of individual and psychological factors that contribute to this gender gap. Cultural stereotypes regarding the need for innate brilliance selectively discourage women from pursuing careers in physics or engineering [10]. This perception is exacerbated by stereotype threat, where the fear of confirming negative prejudices can inhibit women’s performance and aspirations in quantitative domains. Complementarily, lower self-efficacy in male-coded fields and the scarcity of role models are key determinants [11]. The inclusion of Medicine within the STEMM acronym is justified by its growing symbiosis with technology, biostatistics, and biomedical engineering [12]. However, research has documented that prosocial motivations, the desire to help others directly, often lead many women with strong STEM competencies to choose medicine over engineering, revealing a complex interplay between skills, personal values, and the social perception of disciplines [13,14].

Beyond individual variables, the state of the art acknowledges the decisive influence of contextual factors, both sociocultural and institutional. Family environment and early school experiences play a fundamental role in shaping interests and career aspirations [15,16]. At the same time, the technological advances of the last decade and the increasing visibility of STEMM careers offering high salaries and global opportunities have reconfigured the landscape of economic incentives, potentially influencing women’s decisions positively [17]. At the institutional level, universities and workplaces can perpetuate structural biases that hinder women’s advancement. Although medicine is included within STEMM, recent studies show that female researchers in medicine achieve lower success rates in research funding and face barriers to leadership positions in surgical specialties [18,19]. Finally, access to financial support, such as scholarships and residencies, emerges as a decisive lever for retaining women in STEMM careers, especially those from disadvantaged socioeconomic backgrounds [20,21].

Despite extensive international evidence on the factors shaping STEMM career choice, significant gaps remain regarding the specific causes within the context of Peruvian public universities. Most existing studies have focused on Global North countries, leaving unexplored the interaction of sociodemographic, contextual, and attitudinal variables in settings with distinctive sociocultural and economic characteristics. Understanding these dynamics is essential for designing interventions and public policies that do not merely replicate foreign models but instead respond to local realities [22]. A significant gap persists regarding the specific interaction of determinants in the Peruvian public university context. This study moves beyond descriptive analysis to empirically test the factors influencing this choice. We hypothesize that (H1) motivational and prior experiential factors will demonstrate a stronger association with STEMM enrolment than sociodemographic factors. Furthermore, we hypothesize that (H2) women in STEMM, despite intellectually rejecting ability-based stereotypes, will still report significant constraints from sociocultural gender expectations. The objective of this research is therefore to test these hypotheses, identifying and explaining the factors that influence women’s choice of STEMM careers in Peruvian public universities.

1.1. Sociocognitive Career Theory and Individual Psychological Factors

Sociocognitive Career Theory (SCCT) is an important framework for understanding the motivations and psychological factors that influence women’s decisions to pursue careers in STEM (Science, Technology, Engineering, and Mathematics) and medicine. This theory posits that career choices are shaped by an interaction of personal attributes such as self-efficacy, outcome expectations, and environmental factors, including social influences and cultural contexts [23,24].

Research indicates that women face unique barriers when exploring STEM fields, including widespread negative stereotypes that undermine their self-efficacy and interest in these areas. Studies have shown that such stereotypes not only affect women’s intentions to pursue STEM careers but also mediate their broader professional choices [24,25]. The implications of these stereotypes are reflected in the findings of Reinking and Martin, who emphasize the importance of addressing social expectations and cultural pressures that marginalize women in STEM, suggesting that educational content reform and cultural change are necessary to close the gender gap [26].

Moreover, different cultural contexts can significantly shape women’s career decisions and their perseverance in STEM fields. For example, several studies focusing on Emirati women suggest that recognizing female role models and advocating for women’s contributions in STEM careers can encourage future generations to follow these paths. Role models play an essential role in shaping young women’s career aspirations, as they reinforce self-efficacy and provide social validation [27,28]. Additionally, the social context in which women operate can either facilitate or hinder their professional progress, particularly through the expectations imposed on them regarding family responsibilities [29]. Research focusing on women’s persistence after graduation reveals that both individual psychological factors and external sociocultural influences play a critical role in shaping their professional trajectories in STEM. Self-efficacy beliefs and organizational support, combined with sociocultural expectations, influence women’s participation and retention in STEM professions [30,31]. Evidence suggests that fostering a supportive environment, alongside initiatives to strengthen women’s psychological resilience and career aspirations, may promote greater female representation in STEM and medical fields.

In medicine, gender differences persist, particularly in specialties such as cardiology, where women often express concerns about work–life balance and perceive the field differently from their male counterparts [32,33]. This disparity underscores the need to address the cultural and psychological barriers that influence specialty choice among medical students, especially women, who place greater importance than men on family responsibilities when deciding on a professional career [34].

1.2. Horizontal Segregation and Goal Congruence in a Sociocultural Context

Horizontal segregation in careers, particularly within STEM and medicine, reflects the unequal distribution of men and women across professions, with women often concentrated in lower-paying and less prestigious positions. This segregation is shaped by diverse sociocultural and psychological factors, particularly goal congruence, which influences women’s participation in traditionally male-dominated fields. The concept of goal congruence emphasizes how the alignment between an individual’s personal values and the values embedded in a career can affect the likelihood of pursuing and persisting in that career, especially for women in STEM fields [35,36]. The role of educational experiences in shaping women’s career choices in STEM cannot be underestimated. Beekman and Ober report that early mathematical performance significantly influences girls’ later interest in STEM fields, highlighting a sociocultural origin of gender disparities in these areas [37]. These findings corroborate the research of Moakler and Kim, who argue that confidence and demographic factors play a crucial role in women’s selection of university majors in STEM, indicating that improving self-concept and promoting support from female role models can enhance women’s representation in these fields [37,38].

In addition, environmental factors, intertwined with personal challenges, influence women’s professional trajectories in STEM. Studies demonstrate that organizational culture can significantly affect women’s experiences, thereby shaping their retention. The findings of Babalola et al. suggest that a supportive work environment fosters a sense of belonging, which is essential for motivating women to enter and remain in STEM careers [39]. Conversely, toxic workplace climates characterized by incivility or lack of support often lead to adverse professional outcomes for women, perpetuating horizontal segregation [40].

In medicine, women face specific challenges related to balancing work and personal life, particularly in demanding specialties such as surgery or emergency medicine. Research highlights that, for many medical students, family considerations strongly affect their choice of specialty, reflecting social expectations that caregiving roles may be prioritized by women over professional advancement [32,38]. The intersection of gender, ethnicity, and socioeconomic status adds further complexity to discussions on women’s participation in STEM and medicine. Studies indicate that organizational and socio-structural factors play a significant role in shaping the experiences of women from diverse backgrounds after graduation, influencing their career choices in STEM fields [31]. Furthermore, research exploring systemic barriers faced by women in different contexts highlights that prevailing social stereotypes often dictate young women’s career aspirations [41,42].

2. Materials and Methods

2.1. Study Design and Participants

This research employed an observational, cross-sectional design with an explanatory scope. The objective was to analyze the relationships between existing variables without deliberate manipulation of variables, to explain the factors determining the choice of STEMM careers. The study population consisted of female university students enrolled in the first academic cycle of the 2025-I term at three public universities representing Peru’s main geographic regions: the Universidad Nacional del Centro del Perú (Andean Highlands), the Universidad Nacional de Trujillo (Coastal Region), and the Universidad Nacional de San Martín (Amazon Rainforest).

These institutions were purposively selected based on their offering of comparable STEMM careers across the three geographic regions. Inclusion criteria for participants were (a) being a student at a public university, (b) being enrolled in the first academic cycle, and (c) being actively studying at the time of the survey. Exclusion criteria included refusal to provide consent or withdrawal from the study. No monetary compensation was offered; participants were informed that their contribution was important for developing public policies for women’s inclusion in STEM, as promoted by UNESCO.

The study utilized two different samples. First, for the descriptive analysis of horizontal segregation, data were collected from the official university database of first-year STEMM students (N = 1142), which included 145 women. Second, for the inferential analysis (logistic regression), a specific sample was established. An a priori sample size calculation was performed based on the population of STEMM women (N = 145), determining that a minimum sample of 72 was required to achieve a 90% confidence level and a 7% margin of error. The study successfully sampled 76 of these STEMM women, who were then compared with a 1:1 contrast group of 76 non-STEMM women.

2.2. Instrument and Data Collection

Data were collected using a single, integrated virtual questionnaire administered via Google Forms. The instrument was an adaptation of the “University Students’ Questionnaire on STEM Studies in Higher Education (QSTEMHE) [43]. The adapted questionnaire consisted of 41 questions divided into three sections, as detailed in the instructions provided to participants:

• Sociodemographic and Contextual Factors (12 items): This section included variables such as age, university, career, ethnicity, and prior experiences.
• Open-Ended Opinion Questions (5 items): These questions qualitatively explored perceptions of gender roles and professions.
• Attitudinal Factors (24 items): This section measured perceptions on stereotypes, gender roles, and the value of science using a 5-point response scale (ranging from 1 = Totally disagree to 4 = Totally agree, with an option for 5 = Don’t know).

To ensure validity in the Peruvian context, the instrument underwent a two-stage validation process. First, it was submitted for expert judgment by 3 experts to assess content validity and contextual relevance. Second, the internal consistency of the 24 attitudinal items (Part 3) was assessed using the data from the current sample. This analysis yielded a strong Cronbach’s alpha coefficient of 0.89, indicating high reliability for the scale. A confirmatory factor analysis (CFA) was not performed as part of this study’s validation.

2.3. Study Variables

The variables were defined and operationalized for subsequent statistical analysis. The dependent variable was the choice of a STEMM career, coded as a dichotomous variable (1 = chooses a STEMM career; 0 = does not choose a STEMM career). The independent variables consisted of a set of predictors grouped into three categories: (1) sociodemographic factors (age, area of residence, ethnicity); (2) contextual and prior experience factors (prior interest in STEM, participation in STEM activities, technical studies, career choice option, family members in STEM); and (3) attitudinal and motivational factors. For the latter category, specific questionnaire items were aggregated into composite variables, as detailed in

Table 1. Grouping of items for motivation variables.

Composite Variable	Items Included
Altruism	To give back and help society To improve the quality of life for society
Academic Interest	Attraction to the field of study Cultural enrichment
Career/Employment	To find a job High salaries To start my own company Social recognition
External Influence	The will of the family Other friends have chosen these studies Family tradition
Opportunities	To meet interesting people It’s an option for travel The university is close to my home Possibility of working in a team Possibility of working on projects
Others	Another reason

.

2.4. Statistical Analysis

Data processing and analysis were conducted using the R programming language and environment (version 4.4.1, Race for Your Life, 2024-06-14 ucrt), developed by the R Core Team, which provides specialized tools for statistical analysis and data visualization. The analysis was performed in two phases. First, a descriptive analysis was carried out to characterize the descriptive sample (N = 1142), calculating frequencies and percentages for categorical variables and measures of central tendency for numerical variables. Subsequently, to test hypotheses and identify determining factors, a penalized binary logistic regression (Firth’s method) model was fitted using the inferential sample (N = 152; 76 STEMM women and 76 non-STEMM women). This approach was chosen to resolve quasi-complete separation caused by sparse data in the ‘Ethnicity’ variable, utilizing the ‘logistf’ package. This model allowed for the estimation of the probability that a student would choose a STEMM career (1) vs. a non-STEMM career (0) based on the independent variables. Results were interpreted through odds ratios (ORs) with their corresponding 95% confidence intervals and p-values to determine the statistical significance of each predictor.

3. Results

3.1. Descriptive Analysis and Gender Distribution in the Sample

Table 2. Distribution of students by gender in STEMM majors by university and geographic region.

University (Geographic Region)	STEMM Field	No. of Men	No. of Women	Total Students	Percentage of Women (%)
Universidad Nacional del Centro del Peru (Andean Highlands Region)	Mechanical Engineering	63	2	65	3.07%
	Electrical Engineering	67	3	70	4.28%
	Civil Engineering	39	2	41	5.12%
	Mining Engineering	32	4	36	11.11%
	Metallurgical Engineering	42	8	50	16.00%
	Systems Engineering	30	7	37	18.91%
	Human Medicine	28	10	38	26.32%
Universidad Nacional de Trujillo (Coastal Region)	Computer Engineering	99	4	103	3.88%
	Mining Engineering	52	5	57	8.77%
	Materials Engineering	99	4	103	9.30%
	Civil Engineering	55	6	61	9.83%
	Metallurgical Engineering	43	5	48	10.41%
	Systems Engineering	71	10	81	12.34%
	Mechatronics Engineering	45	7	52	13.46%
	Mechanical Engineering	42	8	50	16.00%
	Human Medicine	75	34	109	32.07%
	Sanitary Engineering	31	13	44	38.24%
Universidad Nacional de San Martin (Amazon Rainforest Region)	Civil Engineering	29	1	30	3.33%
	Systems Engineering	31	3	34	8.82%
	Human Medicine	24	9	33	27.27%
Total		997	145	1142	12.70%

details the composition of the sample, which comprised a total of 1142 students enrolled in STEMM programs. Of these, 997 were men, and 145 were women, representing an overall female participation rate of 12.70%. This marked underrepresentation was consistent across the three geographical regions studied: Highlands (UNCP), Coast (UNT), and Amazon (UNSM).

The analysis by field of study reveals strong horizontal segregation. Disciplines such as Civil Engineering at UNSM (3.33%), Mechanical Engineering at UNCP (3.07%), and Computer Engineering at UNT (3.88%) exhibited the lowest levels of female participation. In contrast, Human Medicine consistently presented the highest proportion of women across the three institutions, reaching 32.07% at UNT. The most notable case was Sanitary Engineering at UNT, which, with 38.24% female enrollment, emerged as the program with the highest gender parity in the sample.

Figure 1 illustrates the percentage distribution of gender across each field of study at an aggregated level, ranking disciplines from highest to lowest female participation. This ordering makes the existing gap evident. At the top of the chart, Sanitary Engineering (29.5%) and Human Medicine (29.4%) stand out as the only fields where female representation approaches one-third of the total student body. The next group of engineering disciplines (Mechatronics, Metallurgical, and Systems Engineering) falls within a considerably lower range, with approximately 13% female participation. In addition, six engineering majors—including Mining, Mechanical, Civil, and Electrical Engineering—show female enrollment below 10%, reaching critical lows of 3.9% in Materials and Computer Engineering. This pattern reinforces the conclusion that female underrepresentation is not uniform across the STEMM spectrum but is instead critically concentrated within certain engineering disciplines.

3.2. Comparative Analysis of Perceptions on Gender and Science

Figure 2 presents a comparative analysis of perceptions regarding gender stereotypes in STEM between female students enrolled in STEMM and non-STEMM programs. The data reveal notable divergences in the rejection of gendered assumptions, with STEMM students showing a consistently stronger opposition to statements linking scientific interest and aptitude to men. For instance, 74% of STEMM students disagreed with the notion that girls are less interested in science, compared to only 49% of their non-STEMM peers. Similarly, while nearly half of non-STEMM students (45%) viewed STEM studies as more attractive to men, this perception was shared by only 29% of STEMM students, underscoring the more critical stance of the latter group towards traditional stereotypes.

Other items illustrate additional nuances. The belief that boys dominate STEM because they are “more geeky” was endorsed by 34% of non-STEMM students but largely rejected among STEMM students (46%). Perceptions of household roles also differed, with non-STEMM respondents more frequently agreeing that boys engage in practical activities with parents, whereas STEMM students expressed greater disagreement (39%). Interestingly, both groups showed partial alignment regarding hobbies related to science and technology, though responses among STEMM students were more dispersed, with a higher proportion reporting uncertainty. Taken together, these findings confirm that exposure to STEMM fields fosters stronger resistance to gender stereotypes, yet also reveal persistent ambiguities linked to broader sociocultural expectations.

Figure 3 explores students’ perceptions of the impact of gender expectations and roles on their personal lives, revealing a more complex and nuanced pattern compared to the clear stereotype rejection shown in the previous figure. Notably, STEMM students were more likely to report feeling constrained by gender norms: 33% agreed that “gender expectations limit them,” compared to 26% of non-STEMM students. A similar tendency emerged regarding rigid household roles, with 33% of STEMM students acknowledging their existence, versus 29% among non-STEMM peers.

The data also indicate that experiences of being mocked for transgressing gender norms were reported at comparable levels, though STEMM students showed slightly higher disagreement overall (41% strongly disagreed) than non-STEMM students (36%). With respect to gender labels, approximately one-third of both groups agreed that such labels restrict them, but STEMM students expressed more polarized views, with higher proportions at both extremes of agreement and disagreement. Collectively, these findings suggest that while STEMM students challenge stereotypes about intellectual abilities, they remain acutely aware of the social and cultural constraints imposed by gender roles, underscoring the persistence of structural expectations beyond academic environments.

Figure 4 examines students’ beliefs about gender ideology and its relationship to abilities and affinity for STEM fields, revealing a profound ideological divide between STEMM and non-STEMM students. Across items, STEMM students systematically rejected stereotypes questioning women’s innate scientific abilities, while a significant proportion of non-STEMM students continued to endorse such views. For instance, 83% of STEMM students disagreed with the claim that university studies are more important for men, compared with 72% of non-STEMM respondents. Likewise, 86% of STEMM students dismissed the idea that girls are not as good as boys in science, technology, engineering, and medicine, a higher proportion than the 71% recorded among non-STEMM students.

The sharpest divergence appeared in the perception of “natural abilities.” While 46% of non-STEMM students agreed that girls have fewer innate abilities than boys, nearly seven in ten STEMM students (68%) rejected this assertion. A similar divide was observed in the belief that STEM fields are inherently “masculine,” with 61% of STEMM students rejecting this idea compared to only 49% of non-STEMM peers. Nevertheless, convergence emerged around complementary stereotypes: most of both groups agreed that most girls are “better at other things,” such as humanities and languages (67% non-STEMM; 66% STEMM). Interestingly, clear differences surfaced in assessments of specific skills, with 79% of STEMM students strongly affirming that women can develop useful software programs, compared to 63% among non-STEMM students. Taken together, these findings indicate that exposure to STEMM fields strengthens opposition to exclusionary stereotypes, though persistent cultural narratives continue to frame women’s aptitudes through gendered divisions of knowledge.

Figure 5 evaluates students’ perceptions of whether women must adopt masculine behaviors to succeed in STEMM fields, again highlighting a pronounced ideological gap between STEMM and non-STEMM respondents. The strongest divergence appears in response to the statement “To succeed in STEM: think/act like a man.” Here, 80% of STEMM students disagreed, compared to only 62% of non-STEMM students, indicating that exposure to STEMM environments fosters a more explicit rejection of masculine-coded success norms. Similarly, while 53% of non-STEMM students agreed that women in STEM must “act like men,” this perception was rejected by most STEMM students, with 59% expressing disagreement.

Broader gendered expectations also reveal contrasting attitudes. A third of non-STEMM students (34%) agreed that women should sacrifice their careers for traditional roles, a position strongly opposed by STEMM students, 62% of whom strongly disagreed. Regarding mixed-gender work environments, 55% of non-STEMM respondents considered them a source of conflict, compared with 46% of STEMM students, who showed more rejection of this notion. Finally, when asked whether women succeed only by adopting masculine behaviors, responses diverged again: while nearly half of non-STEMM students (46%) agreed, STEMM students were considerably more skeptical, with 57% disagreeing. Collectively, these results suggest that STEMM students are more resistant to gendered prescriptions of success, whereas non-STEMM students continue to endorse traditional assumptions that equate achievement in science and technology with masculine identity.

Figure 6 analyzes students’ perceptions of the overall value of science, revealing a broad consensus across groups. Both STEMM and non-STEMM students overwhelmingly recognized the positive impact of science on society and their personal lives, though the intensity of conviction differed markedly. For instance, while 94% of STEMM students agreed that science and technology will provide greater opportunities for future generations, 78% of them expressed this belief with strong agreement. By contrast, although 88% of non-STEMM students shared the same view, only 51% reported it emphatically.

Similar patterns were observed for other items. A large majority of STEMM students (71%) strongly agreed that science is useful in daily life, compared with 46% of non-STEMM respondents. Likewise, 95% of STEMM students affirmed that learning science had made them more critical overall, with 53% strongly agreeing, whereas non-STEMM students tended to agree at a moderate level (57%) rather than with strong conviction (30%). Taken together, these findings indicate that although both groups attribute high value to science, STEMM students demonstrate greater intensity of belief, reflecting a stronger internalization of science as a central component of their worldview and future expectations.

3.3. Determining Factors in the Choice of STEMM Careers

Table 3. Logistic regression model for sociodemographic variables in the choice of STEMM careers among public university students.

Predictor Variable	Category	Coefficient (β)	Std. Error	Odds Ratio (OR)	p-Value
Sociodemographic Factors
Age	—	0.059	0.126	1.0607	0.6508
Area of Residence	Urban	−0.319	0.509	0.7271	0.5433
	Marginal Urban	−0.462	0.894	0.6298	0.6317
Ethnicity	Amazon	−0.659	1.913	0.5174	0.7273
	Other(s)	−1.836	0.755	0.1595	0.0108 *
	Quechua	−1.684	0.482	0.1857	0.0004 *
Contextual & Prior Experience
Prior Interest in STEM	Yes	2.048	0.924	7.7556	0.0178 *
Prior STEM Activities	Yes	0.117	0.511	1.1237	0.8271
Previous Technical Studies	Yes	−0.440	0.862	0.6443	0.6284
Career Choice Option	Second	−0.362	0.410	0.6966	0.3967
	Third	−1.507	0.805	0.2216	0.0545
Family in STEM	Yes	0.651	0.654	1.9182	0.3495
Opinion & Motivational Factors
Motivation: Altruism	—	−0.307	0.272	0.7356	0.2759
Motivation: Academic Interest	—	−0.574	0.340	0.5632	0.0987
Motivation: Career/Employment	—	0.005	0.192	1.0050	0.9800
Motivation: External Influence	—	−0.369	0.470	0.6917	0.4481
Motivation: Opportunities	—	0.806	0.222	2.2379	0.0001 *
Motivation: Others	—	−0.447	0.798	0.6397	0.5942
(Intercept)	—	−2.545	2.452	0.0785	0.3139

Note: * p < 0.05; p < 0.10. Reference Categories: Area of Residence (Rural), Ethnicity (Mestizo), Prior Interest/Activities (No), Technical Studies (No), Career Choice Option (First), Family in STEM (No).

presents the full results of the logistic regression model adjusted to predict the choice of a STEMM career. This table details the fundamental statistical parameters for each predictor variable. The coefficient (β) indicates the direction and change in the log odds of the probability; a positive value, such as Prior Interest in STEM (β = 2.048), is associated with an increased probability of choosing a STEMM career, whereas a negative value, such as Ethnicity: Quechua (β = −1.684), is associated with a decreased probability. The p-value serves as the key indicator of statistical significance. In this analysis, four variables show a statistically significant relationship (p < 0.05) with career choice: Ethnicity: Other(s) (p = 0.0108), Ethnicity: Quechua (p = 0.0004), Prior Interest in STEM (p = 0.0178), and Motivation: Opportunities (p = 0.0001). Additionally, the variables Career Choice Option: Third (p = 0.0545) and Motivation: Academic Interest (p = 0.0987) are considered marginally significant (p < 0.10). The remaining factors, such as age or area of residence, do not demonstrate a statistically significant effect in this model.

Table 4. Odds ratios from logistic regression predicting the choice of a STEMM career.

Predictor Variable	Odds Ratio (OR)	95% Lower Confidence Interval	95% Lower Confidence Interval
Sociodemographic Factors
Age	1.0607	0.8206	1.3891
Area of Residence (Ref: Rural)
Urban	0.7271	0.2487	2.0097
Marginal Urban	0.6298	0.0862	4.0198
Ethnicity †
Amazonia	0.5174	0.0027	18.0725
Other(s)	0.1595	0.0262	0.6716
Quechua	0.1857	0.0665	0.4779
Contextual & Prior Experience Factors
Prior Interest in STEM (Ref: No)
Yes	7.7556	1.3809	85.2742
Prior STEM Activities (Ref: No)
Yes	1.1237	0.3889	3.2183
Previous Technical Studies (Ref: No)
Yes	0.6443	0.0982	3.8899
Career Choice Option (Ref: First)
Second	0.6966	0.2978	1.6110
Third	0.2216	0.0338	1.0278
Family in STEM (Ref: No)
Yes	1.9182	0.4827	7.6539
Opinion & Motivational Factors
Motivation: Altruism	0.7356	0.4163	1.2772
Motivation: Academic Interest	0.5632	0.2701	1.1111
Motivation: Career/Employment	1.0050	0.6789	1.4954
Motivation: External Influence	0.6917	0.2620	1.8109
Motivation: Opportunities	2.2379	1.4556	3.6470
Motivation: Others	0.6397	0.1209	3.3962

Note: OR = odds ratio; CI = confidence interval. Model (N = 152). † Coefficients and CIs for ‘Ethnicity’ were stabilized using Firth’s penalized regression to correct for quasi-separation.

translates the model coefficients into odds ratios (ORs), which facilitate the interpretation of the magnitude of each predictor’s effect. An OR greater than 1 indicates that the probability of the event (choosing a STEMM career) increases, while an OR less than 1 indicates a decrease. Within this dimension, the most influential factor in the study is Prior Interest in STEM (OR = 7.76). This means that prior interest increases the likelihood of choosing a STEMM career by almost 8 times. The 95% confidence interval for this variable [1.38, 85.27] does not include the value 1.0, confirming its statistical significance. On the other hand, selecting the career as the third option (OR = 0.22) has a strong negative, marginally significant association. Variables such as Prior STEM Activities, Previous Technical Studies, and having Family in STEM are not significant, as their confidence intervals clearly include the value 1.0, indicating that the absence of effect cannot be ruled out.

Opportunities (OR = 2.24) emerges as another strong predictor that more than doubles the probability of being in a STEMM field, and its 95% confidence interval [1.46, 3.65] confirms this significance. In contrast, Motivation: Altruism (OR = 0.74) was not statistically significant in this final model (p = 0.2759), as its confidence interval [0.42, 1.28] clearly crosses 1.0. A key finding of this robust model is the significant association of sociodemographic factors, specifically Ethnicity. While Age and Area of Residence were not significant, identifying as ‘Quechua’ (OR = 0.19, CI [0.07, 0.48]) or ‘Other(s)’ (OR = 0.16, CI [0.03, 0.67])—relative to the reference group No—significantly decreased the likelihood of choosing a STEMM career. This finding, made possible by the penalized regression, highlights a critical structural barrier that was previously uninterpretable.

4. Discussion

4.1. Sub-Representation and Horizontal Segregation in STEMM

The persistent underrepresentation of women in STEMM fields within public universities in Peru is evident, with an overall female participation of only 12.7%. This outcome reflects international evidence indicating that women remain significantly underrepresented in science and technology careers despite global progress in gender equality (OECD; UNESCO). The fact that participation is consistently below 10% in engineering majors such as Mechanical, Civil, Electrical, and Computer Engineering underscores the structural dimension of this disparity, aligning with the concept of horizontal segregation, whereby women are disproportionately absent from traditionally male-dominated and high-prestige disciplines.

The contrast between engineering fields and disciplines such as Medicine (27–32%) and Sanitary Engineering (38%) further illustrates the uneven distribution of female participation, suggesting that motivational and cultural factors play a decisive role in shaping career choice. Previous studies have shown that women are more likely to pursue careers aligned with communal or prosocial goals, such as medicine, where the value of helping others is socially recognized [32,38]. This pattern reflects a cultural framing of “female-friendly” disciplines versus “male-coded” hard sciences, reinforcing stereotypes that continue to influence professional trajectories [41,42].

The systemic nature of this segregation has also been documented in other contexts [44,45]. Ballen et al. demonstrated that even within STEM classrooms, women participate less than men, reflecting broader patterns of exclusion [44,45]. Similarly, UNESCO and OECD reports highlight cultural and institutional barriers as drivers of this uneven distribution [46,47]. The Peruvian case is thus not isolated but rather embedded within a global structure of gendered educational pathways, where women are channeled into health-related or “softer” STEMM fields while remaining underrepresented in engineering and technology.

Globalization processes have been shown to generate mixed outcomes. Although they may expand opportunities for women in the labor market, they also risk reinforcing inequalities in contexts with persistent structural barriers [48]. The extreme gender gaps observed in engineering in Peru (e.g., 3–5% in Civil and Mechanical Engineering) are consistent with the argument of [49] that labor market norms and hiring biases perpetuate male dominance in technical sectors, particularly in developing economies.

The evidence, therefore, reinforces the urgency of addressing horizontal segregation not only through access policies but also via deeper transformations in institutional cultures and curricula. As recommended by UNESCO and OECD, fostering inclusive educational environments and diversifying teaching staff may be decisive for challenging stereotypes and enabling the sustained participation of women in engineering and technology [50].

4.2. Gender Stereotypes and Student Perceptions

The comparative analysis of student perceptions reveals a marked divergence between women enrolled in STEMM and those in non-STEMM fields. Female STEMM students show a stronger rejection of traditional gender stereotypes, with 74% disagreeing with the statement that girls are less interested in science, compared to only 49% among their non-STEMM peers. Similarly, perceptions of STEM as “male-attractive” were less prevalent among STEMM students (29%) than among those outside STEMM (45%). These findings reflect the theoretical framework of stereotype threat and self-efficacy, as women who engage directly in STEMM fields appear more likely to resist internalized stereotypes, although not entirely free from their influence [51,52].

However, the results also highlight a paradox. Despite rejecting stereotypes about intellectual capacity, a considerable proportion of STEMM students (33%) still perceive gender expectations and rigid household roles as limiting, surpassing the 26–29% reported by non-STEMM students. This suggests that while academic exposure to STEM can foster critical resistance to stereotypes, lived experiences of gender roles remain deeply entrenched. Similar observations have been made in previous studies, which show that stereotype threat generates anxiety and reduces self-efficacy in women aspiring to STEM careers, undermining their academic performance and professional aspirations [52,53].

The findings further align with prior research indicating that stereotypes not only undermine women’s sense of belonging in STEM environments but also influence their long-term career aspirations [41,54]. Almost half of non-STEMM students (46%) agreed that women have fewer “natural abilities” for science, while more than two-thirds of STEMM students (68%) strongly rejected this notion. Yet, both groups largely endorsed the complementary stereotype that women are “better at other things,” such as humanities and languages (67% of non-STEMM; 66% of STEMM). This persistence of complementary stereotypes reinforces occupational segregation and reflects cultural discourses that continue to assign intellectual domains by gender [55].

In addition, the data confirm that many non-STEMM students view success in science as dependent on adopting masculine behaviors, with 53% believing that women must “act like men” to succeed, compared to only 41% of STEMM students. Such perceptions resonate with international findings that traditional gender norms are internalized by students and shape their self-efficacy beliefs and aspirations [54]. The rejection of this idea by most STEMM students (59%) indicates a potential cultural shift, although one that is not yet widespread.

The strong consensus on the societal value of science (94% of STEMM and 88% of non-STEMM students) underscores an important area of convergence. Nonetheless, STEMM students express this conviction with greater intensity, with 78% strongly agreeing that science will generate opportunities for future generations. This reflects not only the relevance of STEMM education but also the need to harness such positive orientations within inclusive educational environments. Evidence suggests that interventions such as mentorship programs, visibility of female role models, and gender-inclusive curricula can effectively mitigate stereotype threat and foster higher self-efficacy among women [36,56,57].

4.3. Motivational Factors and Career Satisfaction

The logistic regression analysis demonstrates the decisive role of motivational factors in shaping students’ entry into STEMM careers. The strongest predictor was prior interest in STEM (OR = 7.76), which increased the probability of choosing a STEMM field nearly eightfold. This result is consistent with international evidence emphasizing the centrality of early exposure and intrinsic interest in sustaining motivation and retention within STEM disciplines [58]. Programs that promote engagement with science and technology from early stages, therefore, represent a strategic approach to increasing enrolment and improving persistence throughout higher education.

Motivation associated with opportunities (OR = 2.24) was another relevant predictor, more than doubling the probability of choosing STEMM and reflecting expectations regarding professional and economic prospects. This finding is consistent with prior studies showing that economic stability and access to employment opportunities act as key drivers of persistence in STEM pathways [59]. It also corresponds with the observations of [60] on the balance of intrinsic and extrinsic rewards across technical and health-related fields, reinforcing the need for curricula that integrate personal aspirations with market demands.

In contrast, the revised robust model showed that altruism (OR = 0.74) was not a statistically significant predictor (p = 0.2759). While previous studies associate altruistic motivations with higher engagement in health-related fields such as medicine [61,62], our final model suggests that, after controlling for other factors, this motivation does not significantly distinguish STEMM from non-STEMM choosers in this sample. This divergence highlights disciplinary differences in the way altruistic and economic motivations interact with career choices.

4.4. Limitations of Sociodemographic and Contextual Factors

The final regression model indicated no statistically significant effects for age or area of residence. However, a key finding of the revised analysis, which employed a penalized (Firth) regression to manage sparse data, was the emergence of Ethnicity as a highly significant predictor. The instability noted in the initial unpenalized model was resolved, revealing that students identifying as ‘Quechua’ (OR = 0.19) or ‘Other(s)’ (OR = 0.16) have a significantly lower probability of enrolling in STEMM compared to the reference group.

The non-significance of age and area of residence should not be interpreted as evidence of irrelevance. However, the significant finding for ethnicity aligns strongly with international literature that documents robust associations between socioeconomic background, parental education, and ethnic identity and the educational pathways that students follow, including gendered differentials in access, persistence, and outcomes [57,63,64]. Higher socioeconomic status has been associated with increased likelihood of selecting STEM majors among women, while parental educational attainment is consistently related to academic confidence and persistence.

The significant negative association found for Quechua and Other(s) ethnic groups underscores the analytical importance of modelling intersectional dynamics affecting women and minoritized groups in STEM. Our finding confirms that these structural barriers are not just methodological ‘noise’ but are statistically significant disadvantages. Evidence points to marked differences in motivations and expectations across ethnic groups, with consequences for engagement and retention [65]. In parallel, the function of “counterspaces” that provide support and belonging for women of color has been emphasized as a buffer against isolation and marginalization within STEM environments [53].

These structural and cultural dynamics carry significant psychological implications. Students from disadvantaged or underrepresented backgrounds frequently confront systemic barriers that erode self-confidence and academic performance [66,67]. Broader societal expectations around race and gender intersect with economic pressures, generating additional obstacles for women and minorities considering STEM trajectories [68]. The statistical significance of ethnicity in our model provides direct evidence for this phenomenon in the Peruvian context, underscoring factors that remain central to understanding gender disparities in STEMM.

4.5. Implications for Public Policy and Higher Education

The results indicate the urgency of implementing comprehensive gender-equality policies and equity initiatives within STEMM education. The pronounced underrepresentation of women (12.7% overall), together with marked horizontal segregation across engineering disciplines, points to structural barriers that extend beyond individual motivation or interest. Addressing these disparities necessitates interventions that begin early and continue throughout the educational pipeline.

Strengthening early vocational orientation is a central implication. Evidence shows that career-guidance initiatives targeting young women mitigate gender stereotypes and biases that dissuade engagement with STEM pathways [69]. In the Peruvian and broader Latin American context, sustained exposure to female role models in engineering and medicine is particularly strategic, as such visibility counters cultural narratives that frame STEM as a male domain [70,71].

Mentorship programs constitute another critical strategy. The literature indicates that mentoring relationships enhance confidence, resilience, and persistence among women in male-dominated fields [72]. Tailored mentoring, especially in engineering programs where female participation is below 10%, is likely to reduce attrition and promote inclusion [73]. Complementary initiatives, including staff development on equity and implicit bias, are essential to embed these efforts within an inclusive institutional culture [74].

Policy design should further incorporate financial mechanisms to alleviate socioeconomic barriers. Targeted scholarships and wraparound support for underrepresented groups are vital in Latin America, where family background and economic constraints substantially shape academic trajectories [70]. Ensuring access to financial aid broadens participation and supports long-term retention in STEMM careers. Creating inclusive and bias-free university environments is fundamental to consolidating progress. Research shows that environments fostering diversity and equity are associated with increased female participation in STEM disciplines [75,76]. This requires systematic institutional reforms, including diversity-sensitive curricula, inclusive pedagogical practices, and visible organizational commitments to gender equity.

5. Conclusions

The evidence indicates a persistent underrepresentation of women in STEMM fields within public Peruvian universities, with an overall participation of 12.7%. Marked horizontal segregation is observed: female enrollment is critically low in several engineering majors (often <5% in Civil, Mechanical, and Computer Engineering), whereas comparatively higher representation appears in Medicine and Sanitary Engineering (27–38%). These patterns are consistent with structural and cultural barriers that channel women away from male-coded disciplines. Across attitudinal measures, a pronounced divergence emerges between STEMM and non-STEMM students. STEMM students show stronger rejection of gender-essentialist beliefs about scientific interest and ability, yet a substantial share still perceives limiting gender expectations and rigid household roles. Consensus exists on the social value of science, although STEMM students express this conviction with greater intensity. This combination—normative rejection of stereotypes alongside perceived social constraints—suggests that academic exposure may weaken stereotypes without fully dislodging entrenched gendered norms.

Determinant analysis underscores the centrality of both motivational and sociodemographic factors. Prior interest in STEM (OR = 7.76) is the strongest predictor of entering STEMM, and opportunity-driven motivation (OR = 2.24) also exhibited a strong positive association. In contrast to preliminary models, Motivation: Altruism did not show a significant association in the final robust analysis. Crucially, after addressing methodological issues of quasi-separation using a penalized regression, Ethnicity emerged as a significant structural barrier. Identifying as ‘Quechua’ (OR = 0.19) or ‘Other(s)’ (OR = 0.16) significantly decreases the probability of enrolling in STEMM compared to the reference group. Other sociodemographic predictors (age, area of residence) and family in STEM were not statistically significant. These patterns point to the importance of strengthening early interest formation and addressing the intersectional barriers faced by ethnic minority students. Policy and institutional implications follow directly: integrated interventions that combine early, gender-responsive career guidance, visible female role models and mentoring, targeted financial support, and inclusive curricula and pedagogy are indicated. Such measures address both motivational levers (interest, perceived opportunities) and structural barriers (including those related to ethnicity), thereby improving access, persistence, and progression for women in STEMM.