Understanding STEM and non-STEM female freshmen in the Middle East: a post-pandemic case study

Abstract

After a disruptive event, such as the pandemic, it is reasonable to re-assess the status of past educational disparities. Re-assessment is particularly important for female college students from a traditionally patriarchal society attempting to promote gender equity in education. During the pre-pandemic era, such students preferred non-STEM programs over STEM programs at disproportionate rates. The present study examined indices of performance during the freshman year before and after the pandemic to determine whether choice and performance differences exist between female students enrolled in STEM and non-STEM programs. Comparisons involved the number of courses completed and the grades obtained in mandatory general education courses devoted to communication, computational, and professional competencies. In the pre-pandemic period, STEM students outperformed non-STEM students in all three competencies. In the post-pandemic period, STEM students outperformed non-STEM students only in communication competencies and professional competencies. Yet, in both student groups, post-pandemic performance was lower, even though STEM students completed fewer general education courses. These findings suggest that female students’ re-adjustment to on-campus instruction was challenging, particularly in STEM majors. Although female STEM learners remained a minority and faced performance challenges in the post-pandemic environment, their numbers increased. Because the pursuit of gender equity in education and employment rests on the academic success of such students, these results along with qualitative data obtained from on-campus interviews demand a retooling of academic support systems. A discussion of the available options for early interventions is put forth.

Keywords:

• Female college students
• academic performance
• general education

Reviewing Editor:

• Sammy King Fai Hui, Curriculum & Instruction, The Education University of Hong Kong, Hong Kong

Subjects:

• Higher Education
• International & Comparative Education
• Educational Psychology

According to Event System Theory (Morgeson et al., 2015; Roulet & Bothello, 2023), a disruptive event changes people’s behavior by altering the functioning of a social system. The event is an instance of environmental discontinuity that prevents routine activities from being performed the way they were before the event occurred. By all accounts, the COVID-19 pandemic qualifies as a disruptive event. That is, it is an event that has drastically changed learners’ everyday routines for a non-negligible time interval (Li, 2022), thereby demanding substantial adaptation from all affected parties (Uitdewilligen et al., 2021). In higher education, adaptation may refer to adjustments learners and institutions had to make either to moderate potential damages to learning and teaching or to benefit from the opportunities the disruptive event accidentally created. Of course, damages or gains principally refer to measurable academic performance, such as course grades or grade point average (GPA) scores. In the extant literature, comparisons of academic performance before and during the pandemic have spanned the entire globe, including institutions in the Global North as well as those in the Global South. Regardless of the breadth of the coverage, an ambiguous picture of the specific consequences of such a disruptive event on college students has emerged. Namely, findings have been mixed, including performance declines (Andersen et al., 2022; Whitelaw et al., 2022), improvements (Binrayes et al., 2022; Chang et al., 2022; Iglesias-Pradas et al., 2021; Pilotti et al., 2022); Zhao et al., 2022), or no change (AbdelSalam et al., 2021; Al‐Zohbi et al., 2023, El Said, 2021).

The ambiguity of such findings has raised demands for clarity at the local level. Albeit demands are usually linked to the particular context where faculty and administrators operate, the desire for clarity has now moved to the post-pandemic learning environment. Of interest is the extent to which the post-pandemic learning environment requires a simple readjustment to the pre-pandemic on-campus instruction (assimilation) or a drastic readjustment aching to a reinvention (accommodation). That is, can the post-pandemic educational environment be defined as a return to the ‘old normal’ or as an encounter with a ‘new normal’ whose opportunities and challenges are uncertain? In particular, do post-pandemic challenges need to be interpreted within the existing instructional frameworks (e.g. student-centered instruction), or require a reconfiguration of such instructional frameworks? The latter implies that a new ecosystem has been created by the pandemic and its aftermath that demands new ways to conceptualize and deliver instruction to ensure learning (e.g. Ferns et al., 2021; Fullan, 2020; Valiga, 2021).

The conundrum of the extent to which readjustment is needed in the post-pandemic environment at each institution of higher education necessitates a data-driven approach (El-Moussa et al., 2021). The first step in such an approach is the collection of reliable information about students’ performance and academic choices (e.g. selection of a major), which is then used to guide decision-making processes regarding the extent to which changes are required.

In this manuscript, we describe the data-driven approach championed by the general education faculty of a higher education institution located in the Middle East (Saudi Arabia) that has adopted a curriculum largely imported from the United States as well as a student-centered pedagogy. The institution is embedded in a society currently in transition from a patriarchal social system to one that fosters gender equity in education and professional pursuits within a neoliberal economic plan (Singh et al., 2022). The plan, named Vision 2030, is intended to re-engineer the economy from one largely reliant on fossil fuels and its byproducts to a diversified one where knowledge and services take center stage (Moshashai et al., 2020; Nurunnabi, 2017). In this plan, conceived from the top, a diversified knowledge-based economy depends on young women of college age who are expected to be active contributors along with young men (Mitchell & Alfuraih, 2018). As a result of this plan, in recent years, women have been given access to all educational programs and professional opportunities previously restricted to men (Alabdulkarem et al., 2021; Allmnakrah & Evers, 2020). The plan recognizes that a diversified knowledge-based economy requires an educated and skilled workforce in science, technology, engineering, and math (STEM) disciplines (Hanafi et al., 2021). It also recognizes that the success of the reengineering plan heavily relies on females’ enrollment in STEM disciplines. Non-negligible challenges, however, exist (Narasimhan, 2021; Pilotti, 2021). For instance, students’ standardized measures of achievement in STEM fields, which reflect their ability to meet international standards, have been less than desirable, despite educational reforms and the availability of substantial resources (Kayan-Fadlelmula et al., 2022).

Literature review

One of the most glaring challenges since the pre-pandemic era has been young women’s enrollment in STEM disciplines. The sources of this phenomenon are still a matter of contention, although some overlap exists. For instance, according to Kanny et al. (2014), who conducted a systematic review of the literature, five categories of motives explain women’s under-enrollment in STEM fields: (a) individual background characteristics, (b) structural barriers in K–12 education, (c) psychological factors, values, and preferences, (d) family influences and expectations, and (e) perceptions of STEM fields. Instead, Wang and Degol (2017) reported (a) cognitive ability, (b) relative cognitive strengths, (c) occupational interests or preferences, (d) lifestyle values or work-family balance preferences, (e) field-specific ability beliefs, and (f) gender-related stereotypes and biases.

Most studies have delved deeper into one or more categories in the hope of uncovering the largest contributors. For instance, Delaney and Devereux (2019) reported that female students’ under-enrollment in college may be attributed to subject choices and grades in high school. They also found that females were less likely than males to choose engineering, technology, and mathematics subjects, yielding a difference in pre-college academic preparation. The authors argued that subject choices have a causal effect on college students’ decision-making regarding STEM majors. On the other hand, Weeden et al. (2020) found that occupational plans in high school (i.e. operationally defined as the job one plans to have at age 30) may explain female students’ STEM under-enrollment. Both sources of gender disparity fit the theory of status expectation (Ridgeway, 2014) according to which evaluative cultural beliefs about gender differences in competencies lead to differences in students’ self-assessment of a range of abilities, including STEM abilities. As a result, women assess their STEM abilities less positively than men do even given the same test scores or grades (Thébaud & Charles, 2018). Yet, gender differences in interests vary within STEM fields. Engineering favors men, whereas social sciences and medical services favor women. According to Su and Rounds (2015), these differences are attributable to women being more interested in people-oriented professions, and less interested in things-oriented ones.

Of particular interest here is a curious fact about Saudi Arabian women in the pre-pandemic era. Although in most countries women have been underrepresented in many STEM fields (Hutchinson, 2014; Wang & Degol, 2017; Xie et al., 2015), Saudi Arabia has exhibited a higher percentage of female STEM enrollment and graduation than males in natural sciences, mathematics, and statistics, whereas males have dominated engineering (Islam, 2017). Of course, female under-enrollment in engineering is not unique to Saudi Arabia as it has also been reported in other parts of the world (Su & Rounds, 2015).

Notwithstanding the noteworthy women’s enrollment rates in many STEM fields, Saudi Arabia’s employment statistics in such fields have illustrated that academic pursuits have not translated into career choices (Islam, 2017). In this region of the world, the common link between students’ STEM selection and career choices in STEM is gender stereotypes. Persisting gender stereotypes within the Saudi Arabian culture are said to influence the beliefs that particular fields and careers, such as nursing, are female domains (Abdul Razzak, 2016), and others, such as engineering, are male domains (Hassan, 2000). Cultural norms not only make fields such as engineering male-dominated but also lead female STEM graduates to prioritize family duties, such as raising children, over careers (Alghneimin et al., 2023).

Since before the pandemic, at the institution selected for the present study, as at other higher education institutions in Saudi Arabia, STEM fields have been, by and large, represented by engineering and computer science. In such fields, female students have been underrepresented. At this institution, for instance, Pilotti et al. (2022) reported that before the pandemic there were no gender differences in the selection of STEM or non-STEM majors at the start of students’ academic journey. However, towards the end of the journey, males were more numerous in STEM than non-STEM majors. The opposite pattern was displayed by females due to their withdrawing or transferring to non-STEM majors such as business (see also Corwin et al., 2020). When data from around the world are examined, the evidence concerning pre-pandemic performance is less than clear. For instance, Owston et al. (2020) reported that STEM learners performed better than non-STEM learners in a variety of blended STEM and non-STEM courses. Bene et al. (2021) and Bautista et al. (2021) found the same pattern for GPA. Bene et al. (2021) linked the pattern to STEM students’ self-regulation abilities. Ost (2010) and Rask (2010) found performance in STEM courses to be poor relative to performance in non-STEM courses mostly due to the rigor and higher workload demands of the former. Others recognized STEM disciplines as defined by tougher grading standards (Rask, 2010; Sabot & Wakeman-Linn, 1991). Evidence that grades are impactful was reported by Dynan and Rouse (1997) and Rask and Tiefenthaler (2008) who found that relative performance in introductory courses determines to a large extent undergraduate students’ major choices. Concerning women’s STEM performance in college (e.g. grades), the evidence is mixed (Farrar et al., 2023; Gomez Soler et al., 2020; Sax et al., 2016; Stoet & Geary, 2018). However, if performance is poor, evidence exists that female students more often than male students switch out of male-dominated STEM majors (Kugler et al., 2021).

Our study: questions and hypotheses

After the pandemic, Pilotti et al. (2022) found no gender or major (STEM versus non-STEM) differences in the math performance of Saudi Arabian students, including freshman, sophomore, junior, and senior levels. To examine further this population in the post-pandemic period, the present case study focused on female students during their first year of college. Furthermore, the study examined courses grouped by the main competency they were intended to foster to compare the performance of STEM and non-STEM learners. It included courses that offer instruction and practice in one of three key competencies: communication, computation, and professional skills (i.e. critical thinking, the use of modern technologies, and teamwork). Mandatory general education courses for all undergraduate students were selected. Although the selected courses were mandatory, some flexibility existed as to the timing of enrollment during the first two years of college.

The present study adopted a mixed-method approach with the intent to integrate quantitative and qualitative approaches, thereby drawing on the strengths of each (Bernard, 2000). Namely, a quantitative approach was used to ask whether pre-pandemic patterns of choice of academic major and attainment were replicated in the post-pandemic environment defined by a return to entirely on-campus classes. Then, a qualitative approach was used to understand the uncovered patterns of choice of major and performance. To this end, questions regarding students’ views of themselves and their learning environment (qualitative information) were guided by the Expectancy-Value Theory of Eccles (Eccles, 2009; Eccles-Parsons et al., 1982; Eccles & Wigfield, 2020, 2023), according to which expectancy of success and subjective task value are the two determinant motives of choice, performance, and persistence in a given field of study. Self-perceptions of competence define learners’ expectancy of success, which refers to their belief that they can perform challenging tasks well (i.e. self-efficacy). Expectancy of success develops over time from experiences of success and failure that learners have in life. Subjective task values instead refer to the importance attached to tasks that learners expect to undertake (Flake et al., 2015). Both motives are shaped by sociocultural influences (e.g. gender stereotypes; Farrell & McHugh, 2017; Moè et al., 2021) as well as by the behavioral choices made and the outcomes obtained, which inform subsequent choices and their persistence. These motives were examined as potential sources of the shortage of female students in the STEM fields of engineering and computer science at the institution selected for the present study as well as for any performance difference between STEM and non-STEM learners in mandatory general education courses.

The following hypotheses were formulated and tested:

H1: If the post-pandemic learning environment of on-campus classes replicates the pre-pandemic one, female freshman students will be underrepresented in STEM majors relative to non-STEM majors. However, if the gender-equity push of Vision 2030 is seeping into the minds of female learners in the post-pandemic learning environment, female freshman students in STEM fields will exhibit an enrollment increase relative to the pre-pandemic era. Furthermore, no differences in the selection of STEM or non-STEM majors at the start of female students’ academic journey will emerge in the post-pandemic environment.

H2: Concerning post-pandemic performance (as measured by course grades), if students are still re-adjusting to on-campus instruction, their performance in mandatory general education courses will be lower. Alternatively, if the aftermath of the pandemic has preserved effective study habits learned during the pandemic period (e.g. continuous learning; Gonzalez et al., 2020; Sánchez-Mendiola et al., 2023), performance may be even higher.

Pilotti et al. (2023) found evidence of successful adjustment limited to specific courses. Namely, test performance before the pandemic was inferior to test performance after the pandemic in a communication course and a course devoted to computational competencies. Whether performance differences would be detected between STEM and non-STEM students for courses grouped by the competencies they foster was a matter to be investigated as the extant literature yielded little guidance.

H3: If choice and performance differences are found between STEM and non-STEM female students, such differences will be likely to reflect their self-perceptions and views of their current learning environment in the post-pandemic era. Students’ self-perceptions may include their perceived competence, which leads to the expectancy of success, and self-relevance of course activities, which defines subjective task value.

Method

Participants

The participants were a sample of 1339 female students who completed the first year of their undergraduate journey in the post-pandemic period. The sample included 457 students enrolled in STEM programs (i.e. engineering, computer science, or architecture) and 882 students enrolled in non-STEM programs (i.e. law, interior design, graphic design, or business).

The control group included a sample of 1602 female students who completed their undergraduate first-year journey in the pre-pandemic period. In such a period, 346 students were enrolled in STEM programs, and 1256 students were enrolled in non-STEM programs.

All learners in the two samples were full-time students ranging in age from 18 to 25. To be included in either sample, they had to take at least one mandatory general education course devoted to the competencies that the university deemed key (see below). Students were bilingual Arabic-English speakers. Competency in the English language was assessed through standardized tests (IELTS or TOEFL) before enrollment.

Procedure and materials

Students’ grades in courses completed during their first year of college (fall, spring, and summer semesters) were collected. Each letter grade was translated into a percentage: A+ (100–96%) = 98.00%, A (95–90%) = 92.50%; B+ (89–86%) = 87.50%; B (85–80%) = 82.50%; C+ (79–76%) = 77.50%; C (75–70%) = 72.50%; D+ (69–66%) = 67.50%; D (65–60%) = 62.50%; F (>59%) = 0%. At the selected university, a D + grade or higher was required to pass a course. General education courses were organized into three main instructional categories based on the knowledge and skills such courses were primarily intended to impart: communication competencies, computational competencies, and professional competencies. All involved English as the sole mode of instruction. Communication courses (n = 4) offered instruction and practice in written and oral communication within formal and informal settings. Computational courses (n = 4) offered instruction and practice in quantitative skills, including models and methods of analyzing data, needed to be successful in subsequent courses and real-world settings. Professional competency (n = 4) courses entailed instruction and practice in critical thinking skills, the use of modern technologies, and teamwork, all fostering students’ capacity to make informed and responsible decisions in diverse settings through the use of reason.

During the first year, students of different majors mostly enrolled in general education courses. Freshmen had some flexibility in how they completed general education requirements within a span of approximately two years. During the first year, they could also take elective courses, which focused on particular subjects in the social sciences (e.g. history, psychology, geography, etc.) and natural sciences (e.g. chemistry, biology, etc.), or introductory courses related to their major. Thus, these courses contributed to students’ first-year GPA and their course load. Given the variety of choices available to students and the difficulty of aggregating courses that are qualitatively different, comparisons between STEM and non-STEM students were limited to mandatory general education courses.

In the post-pandemic period, the development of a research center devoted to teaching and learning on the campus of the selected university gave general education faculty associated with the center a comfortable space devoted to student-faculty interactions. In the center, faculty developed protocols for assisting a wide range of students from those at risk of academic failure to those with adequate academic performance merely expressing course-relating queries. Interactions were systematically guided by inquiries on the concerns that students in general education mandatory courses had regarding their learning environment and themselves. Besides inquiries about students’ majors, open-ended questions were intentionally general and comprised either performance-based (e.g. How are you doing in your general education classes?), or attitudinal (e.g. How do you find on-campus classes?) topics. Previous pilot work suggested that general prompts were key to students’ willingness to provide copious comments on what was on their minds. Running records of student-faculty interactions, which were created during each post-pandemic semester of data collection, had been carefully anonymized and then categorized by two independent analysts. Students’ comments, labeled as STEM or non-STEM, were transcribed immediately after interactions had taken place. Data were anonymized so that they could not be traced to any particular individual. Interactions approximately ranged in duration from 10 to 30 minutes. Qualitative data included 28 students deemed at risk of academic failure and 32 students who did not exhibit performance challenges in any of the courses selected for the present study. The classification of ‘at-risk student’ was a self-reported one. Anonymized performance data were obtained from the Office of the Registrar of the selected university. The study was approved by the Deanship of Research of the university as conforming to the standards for educational research of the Office for Human Research Protections of the US Department of Health and Human Services.

Design

The present study examined the extent to which two non-manipulated variables (Rosenberg & Daly, 1993) might differentiate behavior (i.e. enrollment and performance): academic major (STEM versus non-STEM), which is a personal descriptor (i.e. a selected variable), and timeframe (before and after the pandemic), which is defined as the participants’ exposure to an unplanned event (i.e. a natural treatment variable). As such, the study was correlational in nature. This label applied to the behavior examined, irrespective of whether it could be quantified (such as grades or enrollment rates) or was measured qualitatively (students’ comments).

Data analyses

Quantitative information, including enrollment in compulsory general education courses and performance in such courses, was submitted to Independent-Samples Mann–Whitney U-Tests. Analyses driven by inferential statistics aimed to uncover group differences (a) between STEM and non-STEM students either before or after the pandemic, and (b) between before and after the pandemic within either the STEM or non-STEM clusters (overtime changes).

Qualitative analyses of students’ comments to open-ended questions were used to explain the outcomes of inferential statistics (Bernard, 2000). Thematic analyses summarized the transcribed participants’ comments (Robinson, 2022). In such analyses, coding was approached as a descriptive method (Maxwell, 1992) to ensure that the content of participants’ responses was accurately reported. Two independent analysts examined the students’ records, including becoming familiar with the content of students’ responses, developing conceptual sets, and coding records according to conceptual sets. For each question, the ocular scan method (Bernard, 2000) was used to determine whether students’ responses fell into performance-based or attitudinal conceptual sets, including expressions of expectancy of success or self-relevance of course activities. Within each set, the ocular scan method was utilized to determine whether similarities or differences existed between STEM and non-STEM students’ responses. Inter-analyst agreement (Robinson, 2022) was used to create subsets of homogenous comments within each conceptual set. The inter-analyst agreement obtained by two independent analysts who coded students’ responses into conceptual sets and subsets was 88%. The agreement of the independent analysts was necessary for comments to be considered as reflecting the participants’ views and behaviors and thus be included in the qualitative results of the study.

Results and discussion

All results of inferential statistics were considered significant at the 0.05 level (Cohen, 2001). If data were submitted to multiple comparisons, the Bonferroni correction was applied. Statistical analyses were organized by the questions they answered.

The pre-pandemic period: choices and performance

In the pre-pandemic period, female STEM learners were 21.60% of the first-year female students. Our focus was on compulsory general education courses that belonged to three key instructional categories: communication, computation, and professional skills. Table 1 displays the descriptive statistics (mean and standard error of the mean) of the number of mandatory general education courses in which students enrolled in their first academic year, whereas Table 2 illustrates the descriptive statistics (mean and standard error of the mean) of the grades obtained.

Table 1. Pre-pandemic period: descriptive statistics of the number of courses completed by STEM and non-STEM majors in each category.

Course category	STEM		Non-STEM
	M	SEM	M	SEM
Communication comp.	2.98	(0.01)	2.98	(0.01)
Profession comp.	2.99	(0.01)	2.99	(0.01)
Computational comp*	3.70	(0.03)	0.90	(0.01)

Note: *Significant group differences.

Table 2. Pre-pandemic period: descriptive statistics of the grades obtained by STEM and non-STEM majors in each category.

Course category	STEM		Non-STEM
	M	SEM	M	SEM
Communication comp*	90.68%	(0.31)	87.61%	(0.20)
Profession comp*	90.66%	(0.31)	87.81%	(0.19)
Computational comp*	82.97%	(0.59)	80.56%	(0.35)
First-year GPA*	3.49	(0.02)	3.31	(0.01)

Note: *Significant group differences.

An Independent-Samples Mann-Whitney U-Test was first used to compare the ranks of 346 STEM students and 1256 non-STEM students. Overall, STEM and non-STEM students took the same number of courses in communication and professional competencies classes during their freshman year [Us ≤ 219146.00, ns]. However, non-STEM took fewer mathematical competency courses [U = 433408.00, with the mean rank for STEM equal to 1426.12, and for non-STEM equal to 629.43].

The U-Test was also applied to performance data to determine whether the choice of a major corresponded to performance differences between STEM and non-STEM students. Notwithstanding their being a small constituency, STEM learners outperformed non-STEM learners in communication competencies [U = 274544.50, with the mean rank for STEM equal to 966.98, and for non-STEM equal to 755.91], professional competencies [U = 273195.50, with the mean rank for STEM equal to 963.08, and for non-STEM equal to 756.99], and computational competencies [U = 222074.00, with the mean rank for STEM equal to 815.33, and for non-STEM equal to 694.61]. Their first-year GPA was also higher [U = 262649.50, with the mean rank for STEM equal to 932.60, and for non-STEM equal to 765.38].

Have female students changed their choices and performance in the post-pandemic era?

Female STEM learners were still a minority (34.13% of all female learners). However, a Chi-Square Test of Independence confirmed that their presence increased from the pre-pandemic era [ꭓ² (1, n = 2790) = 36.44]. To determine whether performance differed between STEM and no-STEM students, we again considered general education courses taken during the first year, including communication competencies, computational competencies, and professional competencies. Table 3 displays the descriptive statistics of the number of mandatory general education courses in which students enrolled in their first academic year, whereas Table 4 illustrates the descriptive statistics of the grades obtained.

Table 3. Post-pandemic period: descriptive statistics of the number of courses completed by STEM and non-STEM majors in each category.

Course category	STEM		Non-STEM
	M	SEM	M	SEM
Communication comp	2.54	(0.04)	2.59	(0.03)
Profession comp*	2.54	(0.03)	2.59	(0.03)
Computational comp*	2.32	(0.05)	2.08	(0.04)

Note: *Significant group differences.

Table 4. Post-pandemic period: descriptive statistics of the grades obtained by STEM and non-STEM majors in each category.

Course category	STEM			Non-STEM
	M	SEM	n	M	SEM	N
Communication comp*	85.84%	(0.57)	457	83.40%	(0.42)	882
Profession comp*	86.50%	(0.59)	454	84.72%	(0.39)	860
Computational comp.	74.88%	(0.89)	444	73.41%	(0.71)	829
First-year GPA	3.36	(0.02)	457	3.35	(0.02)	882

Note: *Significant group differences.

The U-Test was used to compare the ranks of 457 STEM students and 882 non-STEM students. Overall, non-STEM students took more courses devoted to professional competencies [U = 188294.00, with the mean rank for STEM equal to 641.02, and for non-STEM equal to 685.01], whereas STEM took more computational competency courses [U = 227715.00, with the mean rank for STEM equal to 727.28, and for non-STEM equal to 640.32]. There were no group differences in courses devoted to communication [U = 194464.00, ns].

The U-Test was also used to compare the performance (as measured by grades) of STEM and non-STEM students. Contrary to the pre-pandemic data, the analysis of each instructional category involved different numbers of students, illustrating that post-pandemic enrollment in general education courses was much more variable. During their first year, STEM students obtained higher grades than non-STEM students in courses devoted to communication competencies [U = 234109.00, with the mean rank for STEM equal to 741.27, and for non-STEM equal to 633.07] and professional competencies [U = 226530.50, with the mean rank for STEM equal to 726.47, and for non-STEM equal to 621.09]. Performance did not differ in computational competency courses [U = 191897.00, ns]. First-year GPA also did not differ [U = 203623.00, ns].

In sum, students’ behavior in the post-pandemic era was not isomorphic to the pre-pandemic one. Although the post-pandemic era was characterized by more female learners in STEM fields (H1), evidence suggested that they were facing challenges (H2). For instance, concerning students’ choices of general education courses in the post-pandemic environment, even though STEM learners continued to complete more computational competencies courses than non-STEM learners, they completed fewer courses devoted to professional competencies. Students’ performance showed considerable change from the pre-pandemic era. Before the pandemic, STEM learners outperformed non-STEM in all the selected instructional categories. Not surprisingly, their first-year GPA was also higher. After the pandemic, STEM learners outperformed non-STEM learners in courses devoted to communication and professional competencies, but not in courses devoted to computational competencies.

When a comparison was carried out on the grades obtained before and after the pandemic by each student group, a broad decline in performance was observed in both STEM learners and non-STEM learners. Specifically, for STEM learners, grades declined in courses devoted to communication competencies [U = 56264.50, with the mean rank for the pre-pandemic period equal to 467.89, and for the post-pandemic period equal to 352.12], professional competencies [U = 62958.00, with the mean rank for the pre-pandemic period equal to 445.54, and for the post-pandemic period equal to 366.17], and computational competencies [U = 54246.50, with the mean rank for the pre-pandemic period equal to 460.72, and for the post-pandemic period equal to 344.68]. A similar pattern of performance declines was displayed by non-STEM students. Specifically, for non-STEM learners, grades declined in courses devoted to communication competencies [U = 447341.00, with the mean rank for the pre-pandemic period equal to 1154.34, and for the post-pandemic period equal to 948.69], professional competencies [U = 470470.00, with the mean rank for the pre-pandemic period equal to 1113.92, and for the post-pandemic period equal to 977.56], and computational competencies [U = 375538.00, with the mean rank for the pre-pandemic period equal to 1038.10, and for the post-pandemic period equal to 868.00].

Performance declines in the post-pandemic period corresponded to declines in the number of general education courses completed by STEM students in all competencies, including communication [U = 47904.00, with the mean rank for the pre-pandemic period equal to 492.05, and for the post-pandemic period equal to 333.82], professional [U = 49723.00, with the mean rank for the pre-pandemic period equal to 486.79, and for the post-pandemic period equal to 337.80], and computational categories [U = 20885.00, with the mean rank for the pre-pandemic period equal to 570.14, and for the post-pandemic period equal to 274.70]. For non-STEM students, performance declines corresponded to fewer communication courses [U = 363898.50, with the mean rank for the pre-pandemic period equal to 1220.77, and for the post-pandemic period equal to 854.08] and professional competency courses completed [U = 395295.00, with the mean rank for the pre-pandemic period equal to 1195.77, and for the post-pandemic period equal to 889.68]. However, compared to the pre-pandemic period, non-STEM students completed more computation competency courses [U = 925383.50, with the mean rank for the pre-pandemic period equal to 773.73, and for the post-pandemic period equal to 1490.69].

Declines were observed in the first-year GPA of STEM students [U = 68539.00, with the mean rank for the pre-pandemic period equal to 432.41, and for the post-pandemic period equal to 378.98]. Instead, the first-year GPA of non-STEM students improved [U = 587231.00, with the mean rank for the pre-pandemic period equal to 1042.96, and for the post-pandemic period equal to 1107.29]. Improvement might be attributed to major-related and elective courses that non-STEM students took during their first year.

What has changed for female students in the post-pandemic era?

Of course, overall patterns of performance suggested that post-pandemic learning for key competencies had declined for all students. First-year GPA data indicated that non-STEM students compensated for their declining performance by taking courses other than mandatory general education courses. Instead, STEM students did not appear to be able to do so. To develop a successful intervention, however, the voices of the students needed to be heard.

Thematic analysis yielded categories reflecting the concerns that students had raised regarding their learning environment and themselves. Included were students at risk of failure in general education mandatory courses and those expressing satisfactory performance in the same courses. The classification of ‘at-risk student’ was a self-reported one, which included mentions of academic counselors who had suggested contact with teaching faculty for advice and guidance. Although the content of comments about the post-pandemic learning environment did not differ between STEM and non-STEM majors, STEM students’ comments were more frequent, overt, and unprompted. The following are the learning-environment categories ordered by the frequency with which they were mentioned by self-described at-risk students

Unutilized time

Students traveled to and from the university campus via bus or car. Commuting and waiting time between classes were described as a major source of unutilized time. Most at-risk students reported studying at home, but not while on campus or traveling by bus.

Individualized support needs

Demands were made for greater availability of tutoring services for writing-intensive courses and math courses. Demands were also made for a greater faculty presence on campus besides official office hours to enrich student-faculty communication and address difficulties that students experienced. Under this category, educators’ active listening and empathy were reported as helpful.

Course content and activities

At-risk students reported being burdened by the length and quantity of the take-home assignments and tests in general education courses. Probed further, they mentioned that the key issue was the time they had available to complete assignments and prepare for tests related to the multiple classes in which they were enrolled. The choice to stay home to catch up with work often resulted in their falling behind even more. Another widespread comment was about the difficulties encountered in group assignments (see also Brown & Baume, 2023) mostly attributed to the challenges of finding suitable team members, coordinating schedules, and meeting deadlines.

The self-perception categories involved expectancy of success and subjective task values, as per the Expectancy-Value Theory of Eccles (Eccles, 2009; Eccles-Parsons et al., 1982). These are the categories that displayed a difference between STEM and non-STEM students not merely in their frequency, but also their content. According to this theory, expectancy of success and subjective task value are the two key motives of students’ choice, performance, and persistence in a given field of study. Expectancy of success refers to students’ belief that they can perform challenging tasks well (i.e. self-efficacy). It develops over time from the accumulation of experiences of success and failure. Subjective task values instead refer to the importance attached to activities that learners expect to undertake (Flake et al., 2015), such as those related to tests and assignments to be completed. Both motives are shaped by the broader sociocultural context (e.g. gender stereotypes; Farrell & McHugh, 2017; Moè et al., 2021) as well as by past behavioral choices and obtained outcomes. Important to note is that the theory is moot to the distinction between STEM and non-STEM learners, which was a relevant variable in our qualitative data.

Expectancy of success

At-risk STEM students much more than non-STEM students tended to exhibit high expectations of academic success at the beginning of the semester, which were easily crashed as soon as minor difficulties or challenges were experienced. Their expectations rested on largely untested beliefs of competency as indexed by grades on assignments or tests. Minor deviations from a full mark often led them to quickly consider course withdrawal. Yet, at-risk STEM students were those who sought corrective feedback more often, viewed feedback as fostering confidence in their abilities to overcome challenges, and conveyed a desire to learn the skills and knowledge that their chosen profession would require.

Subjective task values

At-risk STEM students much more than non-STEM students did not see the knowledge and skills acquired in general education courses as transferring to courses in their chosen field. They saw such courses as a necessary rite of passage to move to upper-level courses where the particular knowledge and skills of their envisioned profession would be learned. Their attributing little value to general education courses made their academic difficulties or challenges particularly hard on them, leading to emotional reactions ranging from skepticism to irritation.

It is important to note that a running record was also kept for comments made by students without academic challenges (n = 32). A similar pattern of self-perceptions was displayed by STEM and non-STEM students without academic difficulties, albeit at a much more subdued level. Thus, H3 was only partially supported as differences between STEM and non-STEM students, irrespective of whether they were or not at risk of academic failure, were limited to self-perceptions. Taken together, these findings not only support the predictions of the Expectancy-Value Theory but also qualify them by highlighting that different fields of study and performance in such fields may shape the degree to which expectancy of success and subjective task values are expressed.

General discussion

In the present study, Event System Theory situates the study in the post-pandemic era while the pre-pandemic setting serves as the reference point. Event System theory justifies the collection of evidence of enrollment and academic performance within these two timeframes. Expectancy-Value Theory serves a different purpose. Its goal is to inform the selection of two key students’ motives in the post-pandemic world: expectancy of success and subjective task value. These motives are then examined to understand whether they can account for the shortage of female students in STEM fields at the institution selected for the present study as well as for performance differences between STEM and non-STEM learners in mandatory general education courses.

The larger societal context of Saudi Arabia where the study takes place gives significance to its findings (as presented above). Namely, the economy of a country that is in the process of restructuring its very foundations needs the contribution of young women and men of college age. In this context, gender equity becomes more a necessity than a noteworthy aspiration. The particular ambition of developing a sustainable knowledge- and service-based economy rests in large part on women’s success in STEM fields. Thus, the number of women choosing to enroll in STEM programs, graduate from such programs, and pursue a STEM career becomes an important matter for Saudi Arabia and the Middle East at large.

The pandemic has disrupted established learning routines and altered the lives of learners across the globe, conveying a mixture of distressing realities, uncertainties, and opportunities for change (Brodie et al., 2022; Many et al., 2022). In the post-pandemic era, a key question is whether the enrollment and academic attainment of women in STEM and non-STEM fields are satisfactory. The present case study, although limited to one institution, has shown that the enrollment of female students in STEM programs has improved relative to the pre-pandemic period. However, female STEM freshmen, much more than female non-STEM freshmen, are experiencing challenges when key competencies rather than individual courses or particular course activities are used to measure their performance. For all female freshmen who are at risk of academic failure, remedies may entail changes in both their learning environment and self-perceptions. For STEM female freshmen at risk of academic failure, changes in self-perceptions are especially critical to their success.

The present study has limitations to be addressed in future research. One of such limitations is its focus on one Middle Eastern institution whose curriculum is of Western import and English is the language of instruction. Other institutions in the Middle East may yield unique patterns of outcomes if their economy is structured differently. Another limitation stems from the choice made by the institution selected for the present study to return to entirely on-campus classes after the end of the pandemic. The aftermath of the pandemic may be less noticeable to other institutions that have switched to blended learning or have preserved some of the pandemic-related online learning (Cheung et al., 2023). Yet, our study and its findings suggest that analyses of academic performance by competencies may offer a more reliable window into the current status of students than analyses of performance in individual courses. Of course, the latter may still be valuable tools for faculty members to make curriculum and instruction decisions in specific domains.

Although the quantitative and qualitative data of the present study suggest that STEM female freshmen are experiencing challenges mostly with self-perceptions, the data do not offer specific information as to the extent to which the post-pandemic learning environment is different from the pre-pandemic one, thereby requiring either adaptation or re-invention. A much more in-depth and detailed examination of the post-pandemic learning environment is to be carried out before a conclusion can be made on this issue. One thing is clear though. Higher education institutions do not exist in a vacuum. The ecosystem in which an institution exists, including its socio-cultural and economic makeup, is impactful. Learners understand and interpret their performance within a particular situation, which is embedded in the larger sociocultural context that shapes their expectancy and value hierarchies (Eccles & Wigfield, 2020). As time goes by, the post-pandemic data of today may be quite different tomorrow if the ecosystem changes. Notwithstanding its limitations, the present study offers a simple roadmap for educators and administrators who desire to understand (a) whether the post-pandemic environment is accompanied by changes in their students’ academic attainment, and (b) the extent to which changes can be traced to particular features of the post-pandemic learning environment and/or students’ self-perceptions.

Implications

The results of our study have informed changes that include teaching and the broader institutional setting in which teaching occurs. Specifically, remedies have targeted changes in the learning environment and self-perceptions in the post-pandemic era. For instance, in cases in which the commute time is extensive, plans have been made for retraining students’ study habits to ensure that work would be done on campus. Study areas at the library have been advertised and special workspaces (i.e. capsule office pods) have been created in the main buildings that host classrooms to allow students to work alone or with others. Study areas have been also created at the research center but restricted to students working on research projects. In specific cases, suggestions have been made for relocation along with the offering of accommodations at nearby housing facilities. Also, information has been disseminated on the challenges of estimating correctly the time an activity may require to counteract planning fallacies (Francis‐Smythe & Robertson, 1999).

For concerns regarding course content and activities, the general education program has initiated a series of self-examinations to determine suitable changes. Demands for individualized support have been answered through an increased presence of counselors and students hired as teaching assistants as well as through extended office hours to address students’ most pressing needs. More flexibility in deadlines for assignments has also been introduced.

Practical remedies concerning the learning environment have been easier to conceive and implement than those involving students’ expectancy of success and subjective task values, although the two domains are interrelated. Changing students’ self-perceptions has been viewed as a challenging long-term project. As a starting point, talks and workshops have been organized to tackle counterproductive self-perceptions and their related manifestations (e.g. stress) through information sharing. Opportunities for individualized counseling have been also made available. Individual faculty members have stressed the need to inform students of the impact of factors such as expectancy of success and subjective task values on academic performance. Thus, in several courses, faculty members have discussed such factors with their students. Recognition exists that learners enter an academic course with diverse aptitudes and past experiences, which affect their initial sense of ability to learn in that particular course. During course activities, students may re-assess their abilities by utilizing cues made salient by instructional materials and personnel, which convey additional information about their abilities to acquire knowledge and skills. Thus, feedback (especially, affective feedback; Cabestrero et al., 2018) and the mode of its delivery have been also revised to foster students’ self-regulation skills. In this context, feedback has been conceptualized not as a mere provision of information to students about their work but rather as a process through which students make sense of information about their performance to further their learning (Winstone et al., 2022). As evidence suggests that early research experiences can benefit both STEM and non-STEM learners (Stanford et al., 2017), students have been encouraged to pursue research activities under the supervision of faculty members. Role models have been provided through an undergraduate research society that enlists successful student researchers.

Faculty members also have requested another way to assess whether performance differences exist between STEM and non-STEM learners. Instead of organizing courses by competencies, they have suggested examining performance in particular courses that are deemed critical to the success of all undergraduate students irrespective of their major. To this end, faculty members have mentioned a general education course that covers research methodologies and scientific writing skills and an introductory statistics course. Faculty in the general education program have deemed these courses as offering the foundations for the success of both STEM and non-STEM students in their respective majors. However, the same faculty have recognized that the outcomes of examinations of individual courses might lead to results that do not apply to competencies defined by clusters of courses (see Pilotti et al., 2022).

More broadly, our findings have policy implications for academic institutions that are searching for ways to adapt to the aftermath of an unplanned and disruptive event (i.e. a natural treatment variable), such as the pandemic (Sutton & Jorge, 2020). The aftermath of such an event offers opportunities to rethink educational contents and deliveries, which may lead to building materials and implementing instructional modes that reduce inequalities and enable equal access to quality learning for all students. Not surprisingly, the study described here and its findings are an appeal to action research, which is praxis entailing acting upon the conditions one faces to induce change (Mertler, 2019). In action research, educators enter a real-world setting aiming not only to acquire knowledge but also to induce desirable change. Thus, the roles of educator and researcher coalesce into the unifying role of reflective practitioner. A key responsibility of such a practitioner is to first assess current conditions, evaluate the need for change, implement change, and then go back to the assessment of current conditions, in a recursive loop of actions intended to foster students’ academic success (Lufungulo et al., 2021; Messikh, 2020; Pilotti et al., 2023).

Conclusion

Research reporting on the effects and experiences of the pandemic in educational settings belongs to three broad categories: (a) prescriptive research that aggregates existing instructional knowledge to ensure learning amid the disruptive event (emergency remote teaching); (b) theoretical analyses reflecting and framing challenges and opportunities; and (c) empirical studies examining changes in teaching, and learning. In the latter category, the most basic approach has been either to ask students, educators, and parents about their experiences or to compare performance data before and during the pandemic. Although studies have asked diverse questions, all inquiries have recognized that the pandemic has represented a non-negligible environmental discontinuity. That is, as defined by Event System Theory (Morgeson et al., 2015), the pandemic has been viewed as severely disrupting routine activities and habits. Inquiries have coalesced into the far-reaching issue of the extent to which adaptation, accommodation, or modification (McGlynn & Kelly, 2019) is required for some or all parties involved (i.e. learners, educators, and administrators). Our study adopts a forward-looking, data-driven approach (El-Moussa et al., 2021), thereby comparing academic performance and enrollment before and after the pandemic. It focuses on assessment, collecting data that can be useful to the decision-making process of determining the extent to which the post-pandemic environment is different from the pre-pandemic one, and if so, if it requires adaptation, accommodation, or modification. Nevertheless, it merely represents the first steps of an action research project whose recursive nature (Lufungulo et al., 2021; Messikh, 2020) is yet to be fully realized.

Ethical approval

The data collection process for the present research was approved by the Deanship of Research to comply with the guidelines for educational research set by the Office for Human Research Protections of the U.S. Department of Health and Human Services (§46.104 exempt research) and with the American Psychological Association’s ethical standards in the treatment of research participants. Identifiers were deleted upon data entry. Random numbers were used on data sheets to differentiate participants.

Author contributions

All authors contributed equally to the research and related manuscript.

Acknowledgment

A special thank-you note goes to the members of the Undergraduate Research Society. The authors did not receive support from any organization for the submitted work.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

The datasets generated during and/or analyzed during the current study are available from the corresponding author upon reasonable request.

Additional information

Notes on contributors

Maura A. E. Pilotti

Maura A. E. Pilotti is a cognitive psychologist whose research interests include learning and memory processes across the lifespan. Currently, her research focuses on the interrelations of memory, language, emotion, and culture. She received her Ph.D. in Cognitive Psychology at the City University of New York (USA). Corresponding Author’s Email: [email protected].

Khadija El Alaoui

Khadija El Alaoui is a scholar of American culture whose specializations encompass history, peace and justice studies, and higher education. Currently, her research focuses on the history of the Arab and Western worlds, human diversity, and cultural practices. She received her Ph.D. in American Studies from the University of Dresden (Germany). Email: [email protected].

Hanadi M. Abdelsalam

Hanadi Mohamed Abdelsalam is a scholar whose interests encompass Astrophysics and STEM education. Currently, her research explores quantum physics. As a chair of the Department of Sciences and Human Studies at Prince Mohammad bin Fahd University, her research efforts are focused on the educational attainment of undergraduate students. She attained her Ph.D. in Astrophysics from Oxford University (Great Britain). Email: [email protected].

Omar J. El-Moussa

Omar Jaudat Elmoussa is a scholar in the field of education broadly defined. Currently, he serves as the Associate Vice President of Student Affairs at Prince Mohammad bin Fahd University. His work is devoted to predicting as well as promoting the educational attainment and well-being of undergraduate and graduate students. He received his Ph.D. in Education from the University of North Dakota. Email: [email protected].