Content area
Background
Deep learning approach plays a pivotal role in nursing education, equipping students with the critical thinking skills and knowledge necessary to address complex clinical challenges. However, nursing students exhibit diverse approaches to deep learning, affected by individual characteristics, academic environments and teaching methods.
ObjectiveThis study aims to identify latent profiles of deep learning approach among undergraduate nursing students and analyze the factors influencing these profiles and their association with learning outcomes.
DesignA descriptive cross-sectional survey.
MethodsA total of 891 undergraduate nursing students from two medical universities in China participated in this study between May and July 2024. Data were collected using the Deep Learning Scale and the Learning Outcomes Scale. Latent profile analysis was employed to identify deep learning profiles. One-way analysis of variance and multinomial logistic regression were used to explore influencing factors of different profiles. The Bolck-Croon-Hagenaars (BCH) method was applied to examine differences in learning outcomes across profiles.
ResultsFour latent profiles of deep learning were identified: "Comprehensive Deep Learners" (27.0 %), "Ability-Oriented Learners" (25.4 %), "Attitude-Driven Learners" (21.7 %) and "Surface Coping Learners" (25.8 %). Gender, grade, preference for the nursing major and participation in flipped classrooms were significant factors influencing profile membership ( p < 0.05). "Comprehensive Deep Learners" had the highest learning outcome scores, while "Surface Coping Learners" scored the lowest.
ConclusionsSignificant heterogeneity exists in deep learning approach among undergraduate nursing students. "Comprehensive Deep Learners" achieved the highest learning outcomes. Nursing education should adopt tailored interventions based on the characteristics of different deep learning profiles to improve students’ learning outcomes and comprehensive competencies.
In recent years, the rapid advancement of medical technology and the increasingly diverse needs of patients have presented new challenges and opportunities for nursing education ( Baker et al., 2021; Nes et al., 2021). The World Health Organization's 2030 Global Nursing and Midwifery Strategy underscores the critical importance of training high-quality nursing personnel ( Rosa et al., 2021). It calls on countries to increase investment in nursing education and develop innovative educational models to address the complex and evolving global healthcare demands. In this context, nursing education should not only equip students with solid theoretical knowledge and practical skills but also cultivate higher-order abilities to enhance the quality and efficiency of nursing services. Deep learning approach, which emphasizes higher-order thinking, has emerged as a key goal in higher education, particularly in nursing( Wang et al., 2024). It not only facilitates a deeper understanding of professional knowledge and skills but also significantly enhances clinical judgment, problem-solving, innovation transfer and critical thinking ( F. MARTON, SÄALJÖ, R., 1976; Walker et al., 2021). These competencies are essential for preparing nursing undergraduates to meet the challenges of an ever-changing healthcare environment. A meta-analysis based on 12 cohort studies revealed that nursing students who engaged in deep learning approach exhibited superior cognitive skills and academic achievement compare with those who adopted surface learning strategies ( Takase and Yoshida, 2021). Nonetheless, fostering a deep learning approach remains a challenge for nursing educators, as the diverse learning strategies and outcomes among students necessitate tailored interventions to meet individual needs ( Joshua and Ingram, 2020; Takase et al., 2020). Learning outcomes, a crucial metric for evaluating student learning, include not only knowledge mastery but also skill development, problem-solving ability and shifts in thinking patterns ( Edwards et al., 2023; Park, 2022). Therefore, further research into the potential characteristics of deep learning approach for nursing undergraduates and its relationship with learning outcomes is necessary to provide evidence-based guidance for personalized teaching approaches.
2 BackgroundWith the advancement of the nursing discipline and the increasing demand for medical services, the training of undergraduate nursing students' competencies has become a critical issue in global nursing education ( Lee et al., 2020; Torres-Alzate et al., 2020). The American Association of Colleges of Nursing ( AACN, 2021) advocates for a competency-based education model aimed at comprehensively enhancing nursing students' overall quality ( American Association of College of Nursing, 2021). In China, the Outline of Nursing Career Development Plan (2021–2025) emphasizes reforming nursing education centered on competence, promoting curriculum optimization, strengthening clinical practice and fostering multidisciplinary collaboration skills to meet modern healthcare demands ( NHC, 2022).
Cultivating deep learning approach among nursing students has become a primary focus in educational practice ( Baloyi, 2023; Takase and Yoshida, 2021). The National Scientific Research Council's 2012 report "Developing Transferable Knowledge and Skills in the 21st Century" highlights that deep learning approach involves processes enabling learners to transfer knowledge and skills across contexts ( NRC, 2012). Deep learning approach refers to a learning strategy where students actively select and use a variety of learning methods to explore and integrate the relationships among different pieces of knowledge. This approach enables them to flexibly apply and transfer their understanding to solve challenging and complex problems in practical situations, leading to a comprehensive grasp of the learning material ( F. MARTON, SÄLJÖ, R., 1976). Deep learning approach emphasizes developing higher-order cognitive abilities, with its core objective being to enable learners to innovatively transfer and apply knowledge based on a thorough understanding and mastery of the material ( Jhala and Mathur, 2019; Liu et al., 2024). Rickles et al. emphasize that deep learning approach goes beyond surface-level symbolic knowledge acquisition, enhancing students' understanding and retention of knowledge, promoting active knowledge construction and playing a crucial role in solving complex problems, fostering innovation transfer and developing critical thinking skills ( RICKLES J, 2019).
For nursing students, deep learning approach not only aids in better developing nursing knowledge but also significantly improves their clinical practice abilities and enhances their problem-solving capabilities in multidisciplinary collaborations and complex scenarios ( Xiaoyan, 2021). Research indicates that nursing students' deep learning approach is closely linked to their learning outcomes( Takase and Yoshida, 2021; Wang Shilei, 2018). Learning outcomes refer to the level of knowledge, skills and values achieved by students after completing a series of courses or training programs, serving as a key indicator of nursing undergraduates' learning development ( Coronado-Vázquez et al., 2023).
However, nursing students exhibit varying levels of deep learning approach, influenced by different learning backgrounds, stages and environments( Xiaoyan, 2021). For instance, studies have shown that nursing students' deep learning levels are generally moderate to low, with lower-grade students more inclined toward surface learning, while upper-grade students demonstrate better performance in critical thinking and knowledge transfer ( Wang et al., 2024; Zhang et al., 2022). Additionally, nursing students' deep learning approach is affected by multiple internal and external factors such as learning motivation, metacognitive ability, teaching design and learning environment ( Ding Liandi, 2022; Zhang Jiahui, 2022).
Deep learning approach emphasizes the cultivation of higher-order thinking and complex problem-solving skills, which are essential for advanced nursing education and the practical needs of nursing education reform. Consequently, scholars have increasingly focused on deep learning research ( Wang Zhuo, 2023). However, most existing studies analyze group average levels, overlooking individual differences in students' deep learning performance. Latent Profile Analysis (LPA), a method for exploring individual differences and group segmentation, classifies students into distinct types based on their performance across various dimensions of deep learning approach ( Hickendorff, 2018). This technique enables the identification of diverse profiles of nursing students' learning characteristics, which may be associated with variations in learning outcomes. Therefore, this study aims to identify potential categories of undergraduate nursing students in deep learning approach through latent profile analysis and explore the factors contributing to differences in learning profiles, as well as the relationship between these categories and learning outcomes. Through this analysis, it is hoped to provide detailed theoretical guidance for nursing education and data support for formulating individualized educational intervention strategies.
3 Methods3.1 Study design
This study employed a cross-sectional design to investigate the latent profile of deep learning approach among undergraduate nursing students and analyze their association with learning outcomes.
3.2 Settings and ParticipantsThe study employed a convenience sampling method to recruit full-time undergraduate nursing students from Heilongjiang and Zhejiang provinces in China between May and July 2024. According to previous study, the recommended sample size for latent profile analysis (LPA) is more than 500 participants ( Nylund, 2007). This study included a sample of 891 undergraduate nursing students, exceeding the recommended range.
The inclusion criteria were as follows: (1) full-time undergraduate nursing students; (2) possessing good comprehension and communication skills; and (3) written or online consent and voluntary participation in the study.
3.3 Ethics considerationThe research was conducted following the guidelines of the Declaration of Helsinki and received approval from the ethics committee at [removed for peer review]. The study adhered to principles of voluntary participation and minimal risk for participants. To ensure privacy, the questionnaire was made anonymous and all participants provided informed consent. All procedures were carried out in compliance with relevant regulations and guidelines.
3.4 Instruments3.4.1 The demographic questionnaire
The questionnaire included the gender, grade, residence, whether only child family, preference for the nursing major and frequent participation in flipped classroom activities.
3.4.2 Deep Learning ScaleIn this study, the deep learning questionnaire developed by Liu was used to evaluate the deep learning situation of undergraduate nursing students ( Liu, 2016). The questionnaire comprised 20 items distributed across five dimensions: practical reflection (5 items); information integration (3 items); learning attitude (5 items); learning values (4 items); and comprehension exercises (3 items). It uses a 5-point Likert scale (1 = "strongly disagree" to 5 = "strongly agree"), with total scores ranging from 20 to 100. A higher total score indicates stronger deep learning level. Example item: "I like to explore the differences and connections between learning content". The questionnaire has demonstrated good reliability and validity among college students. In this study, it was applied to undergraduate nursing students, achieving Cronbach's α coefficients between 0.764 and 0.891 for each dimension.
3.4.3 Learning Outcomes ScaleThis study used the Learning outcomes Scale, developed by Wang Fang, to evaluate the learning outcomes of nursing students ( Wang, 2014). The questionnaire includes 13 items across three dimensions: cognitive gains (2 items), skill gains (5 items) and emotional, attitudinal and value gains (5 items), with one item assessing overall learning gains. It uses a 6-point Likert scale (1 = "strongly disagree" to 6 = "strongly agree"), with total scores ranging from 13 to 78. A higher total score signifies a greater level of learning gains for the individual. Example item: "Through learning, I have acquired practical operational skills in my major". Previous research has demonstrated that this scale exhibits good reliability and validity among college students. In this study, when applied to undergraduate nursing students, the scale maintained its reliability and validity, with Cronbach's α coefficients ranging from 0.776 to 0.910 across all dimensions.
3.5 Data collectionFirst, the head of the nursing school at each institution was contacted and after obtaining their consent, the researcher and counselor established contact and conducted the survey. Prior to the survey, investigators were recruited and underwent unified training. The trained investigators then distributed questionnaires to students, explaining the purpose and methodology of the survey. After obtaining informed consent from the participants, students completed the questionnaires on-site. To ensure data integrity, investigators conducted detailed on-site inspections, promptly identified any gaps and corrected any omissions in the questionnaire entries. For senior internship nursing students, the internship coordinators contacted students across various hospitals and administered the questionnaire via an online platform (Questionnaire Star). In total, 902 questionnaires were distributed in this study, with 218 collected through remote online methods. After excluding eleven invalid questionnaires (response time<2 minutes), 891 valid questionnaires were obtained, resulting in an effective recovery rate of 98.78 %.
3.6 Data AnalysisIn this study, Mplus 8.3 software was used to conduct latent class analysis (LCA). The number of latent classes was incrementally increased based on the scores of each item from the deep learning scale as an observed indicator. The optimal model was determined by evaluating model fit indices (AIC, BIC, aBIC) and clinical judgment. Lower AIC, BIC and aBIC values indicate better model fit, while entropy 0.8 signifies higher classification accuracy. The Lo-Mendell-Rubin adjusted likelihood ratio test (LMR) and Bootstrap likelihood ratio test (BLRT) were used to compare the fit of different models( Scrucca et al., 2016). A significant p-value indicated that the current model outperformed the previous one. SPSS 26.0 software was employed for statistical analysis ( Corp., 2019). Normally distributed data were presented as meanSDstandard deviation, while categorical data were described using frequency and percentage. After identifying the best model, chi-square tests and rank sum tests were conducted to compare the demographic characteristics of nursing students across different latent classes. Indicators with statistical significance were further analyzed using multinomial logistic regression. Finally, Mplus 8.3 was used to apply the Bolck-Croon-Hagenaars (BCH) method to explore differences in learning outcomes among nursing students in different latent classes ( Asparouhov and Muthén, 2019). Statistical significance was set at p < 0.05.
4 Results4.1 The level of deep learning and learning outcomes of undergraduate nursing students
The total score for the deep learning among undergraduate nursing students was 58.64 (SD 17.32) points (range: 20–100). The subscale scores were as follows: practice reflection 15.79 (SD 5.52) points (range: 5–25), information integration 8.49 SD 3.29 points (range: 3–15), learning attitude 15.28 (SD 6.00) points (range: 5–25), learning values 10.52 (SD 4.33) points (range: 4–20) and comprehension exercises 8.57 (SD 3.36) points (range: 3–15). The total score for the learning outcomes scale was 44.19 (SD 7.49) points (range: 13–78), with subscale scores of overall learning gain 3.47 (SD 0.80) points (range: 1–6), cognitive gain 6.92 (SD 1.60) points (range: 2–125), skill gain 13.92 (SD 5.96) points (range: 5–30) and emotional attitude value gain 19.86 (SD 3.12) points (range: 5–30).
4.2 Deep learning approach latent profiles in undergraduate nursing studentsIn this study, five models were fitted, and the fitting indices are presented in Table 1. The results indicate that AIC, BIC and a BIC values decreased gradually as the number of latent categories increased, while entropy values consistently exceeded 0.8. When five latent classes were specified, the proportion of students in the first category fell to under 10 %. After comprehensive comparison of the fitting indices, Model 4 was selected as the optimal model due to its simplicity and adequate fit. Based on Model 4, the distribution of latent classes features for deep learning is illustrated in Fig. 1. Specifically, the purple curve group, labeled as "Comprehensive Deep Learner," represented the highest proportion of all participants, comprising 240 students (27.0 %) and was characterized by a balanced, high level of deep learning across all dimensions; The yellow curve group exhibited higher scores in practice reflection, information integration and comprehension exercises but lower scores in learning attitude and learning values, labeled as "Ability-Oriented Learner" with 227 participants (25.4 %); The green curve group scored lower in practice reflection, information integration and comprehension exercises but higher in learning attitude and learning values, labeled as "Attitude-Driven Learner" with 194 participants (21.7 %); The blue curve group scored low across all dimensions and was labeled as "Surface Coping Learner" with 230 participants (25.8 %).
4.3 Comparison of general characteristics among undergraduate nursing students across different latent profiles of deep learning approachAmong the 891 undergraduate nursing students, 11.3 % were male and 88.7 % female. By grade level: 22.1 % were freshmen, 26.0 % sophomores, 28.6 % juniors and 23.2 % seniors. In terms of residence, 59.5 % were from urban areas and 40.5 % from rural areas. A total of 68.9 % were only children. Regarding attitudes toward their major, 39.5 % liked it, 41.0 % felt neutral and 19.5 % disliked it. As for flipped classroom participation, 24.2 % reported frequent participation and 75.8 % infrequent.
The comparison of general characteristics among the four latent profiles of deep learning showed no statistically significant differences in place of origin or whether the student was an only child ( p > 0.05). However, significant differences were observed in gender, grade, preference for the nursing major and participation in flipped classrooms ( p < 0.05). The detailed results are presented in Table 2.
4.4 Multinomial logistic regression analysis of associated factors for latent profiles of deep learning among undergraduate nursing studentsIndicators with statistically significant differences in univariate analysis were selected as independent variables, while the four latent classes of deep learning were taken as dependent variables. The "Comprehensive Deep Learner" was used as the reference group. Multinomial logistic regression analysis was conducted to investigate the associated factors for each latent class of deep learning.
The factors associated with undergraduate nursing students belonging to the "Ability-Oriented Learners" group (with Comprehensive Deep Learners as the reference group) included gender and major like ( p < 0.05). Males had 130 % higher odds of being classified as Ability-Oriented Learners compare with females. Students who disliked their major had 53 % lower odds of being classified as Ability-Oriented Learners compare with those who liked it.
The factors associated with undergraduate nursing students belonging to the "Attitude-Driven Learners" group (with Comprehensive Deep Learners as the reference group) included gender, grade and frequency of participation in flipped classrooms ( p < 0.05). Males had 100 % higher odds of being classified as Attitude-Driven Learners compare with females. Freshmen had 370 % higher odds of being classified as Attitude-Driven Learners compare with the fourth-year students. Students who participate less frequently in flipped classrooms were 68 % lower odds of being classified as Attitude-Driven Learners compare with those who participate more frequently.
The factors associated with undergraduate nursing students belonging to the "Surface Coping Learners" group (with Comprehensive Deep Learners as the reference group) included grade and frequency of participation in flipped classrooms ( p < 0.05). Males had 100 % higher odds of being classified as Surface Coping Learners compare with females. Freshmen had 222 % higher odds and Sophomore had 83 % higher odds of being classified as Surface Coping Learners compare with the fourth-year students. Students who participate less frequently in flipped classrooms were 38 % lower odds of being classified as Attitude-Driven Learners compare with those who participate more frequently. The detailed results are presented in Table 3.
4.5 Differences in learning outcomes levels among nursing students with different latent classes of deep learningThe results indicated that the total learning outcomes and all dimensions for nursing students in the four latent classes of deep learning were statistically significant ( p < 0.01). Specifically, "Comprehensive Deep Learner" nursing students had the highest total learning outcomes score. "Surface Coping Learner" nursing students had the lowest total learning outcomes score. For overall learning gain, emotional attitude value gain and cognitive gain, "Attitude-Driven Type" nursing students scored higher than "Ability-Oriented Learner" nursing students ( P < 0.01). For skill gain, "Ability-Oriented Learner" nursing students scored higher than "Attitude-Driven Type" nursing students ( P < 0.01). Both "Ability-Oriented Learner" and "Attitude-Driven Type" nursing students scored higher in all dimensions compare with "Surface Coping Learner" nursing students, as shown in Table 4.
5 DiscussionThe current study identified latent profiles of deep learning among undergraduate nursing students and analyzed the factors influencing these profiles, as well as their relationship with learning outcomes. This study provides a foundational understanding of the heterogeneity in deep learning approaches within the nursing students and highlights the need for tailored educational strategies.
5.1 Group Heterogeneity in Deep Learning Among Undergraduate Nursing StudentsAccording to the results of latent profile analysis, there are significant differences among the four potential categories of deep learning of undergraduate nursing students, which are named "Surface Coping Learner", "Ability-Oriented Learner", "Attitude-Driven Learner" and "Comprehensive Deep Learner", accounting for 25.8 %, 25.4 %, 21.7 % and 27.0 % of the total. In this study, the comparison of different types of accounts is balanced, reflecting the diversity and group heterogeneity of undergraduate nursing students' deep learning approach.
Among the groups identified, "Surface Coping Learner" nursing students accounted for 25.8 %. This group of undergraduate nursing students may predominantly focus on mechanical memorization and exam-taking strategies, rather than engaging in deep processing and reflection on the material. According to Deep Learning Theory ( MARTON, F. and SÄLJÖ, R., 1976), deep learning is characterized by students' efforts to understand the underlying meaning of the material, integrating new knowledge with existing knowledge and applying it in various contexts. In contrast, surface learning is driven by the intention to simply meet the basic requirements of the task, often focusing on rote memorization and passive learning techniques.
In addition, “Ability-Oriented Learner” nursing students accounted for 25.4 %. This category scored higher in practical reflection, information integration and comprehension practice but lower in learning attitude and learning values. These findings suggest that these students focused more on skill acquisition and operational proficiency while giving less consideration to the meaning or value of learning. This may be related to the overemphasis on technical training in traditional nursing education, which often neglects the cultivation of comprehensive abilities and the integration of theory with practice. While this approach can be beneficial in the short term, it may limit the long-term development of nursing students, particularly when dealing with complex clinical problems ( Hurst, 1985).
On the other hand, “Attitude-Driven Learner” nursing students accounted for 21.7 %. These students scored higher in learning attitude and learning values, indicating a higher level of professional identity and intrinsic learning motivation. However, they scored lower on practical reflection, information integration and comprehension exercises. This discrepancy may be due to improper learning strategies or inadequate course design, which failed to translate their motivational strengths into effective deep learning behaviors. Previous studies have shown that when instructional design does not effectively guide learners to connect theory with practice, even highly motivated students may face challenges in achieving optimal learning outcomes ( Khalil and Elkhider, 2016; Li et al., 2021).
It is worth noting that "Comprehensive Deep Learner" nursing students accounted for 27.0 % of the sample, representing the largest proportion. This finding indicates that a significant portion of nursing students demonstrate adopt deep learning approaches. These students actively construct knowledge frameworks, integrate information and reflect on learning content, which allows them to effectively combine theory with practice and exhibit high transferability when solving complex problems. Consistent with this, Wang et al. found that the deep learning scores of undergraduate nursing students are generally at a moderate level ( Wang et al., 2024). Despite the relatively high proportion of "Comprehensive Deep Learner" nursing students, 73.0 % of nursing students are distributed across other profiles, highlighting ongoing challenges in nursing education.This underscores the need for nursing educators to implement individualized teaching approaches and optimize teaching methods to foster deep learning level across all student groups.
For "Surface Coping Learner" nursing students, efforts should focus on stimulating their enthusiasm for learning by clarifying learning objectives, enhancing their interest and motivation and strengthening their professional identity. These strategies can gradually cultivate their deep learning level. For "Ability-Oriented Learner" nursing students, the priority should be guiding them to understand the broader meaning and value of learning. Career education and emotional support can enhance their intrinsic motivation, helping them expand beyond skill acquisition to include knowledge reflection and comprehensive application. Similarly, "Attitude-Driven Learner" nursing students would benefit from targeted guidance on learning strategies to help translate their positive attitudes into actionable deep learning behaviors. Such support could improve their ability to integrate and reflect on knowledge, ultimately fostering deeper learning outcomes.
5.2 Influencing Factors of Latent Profiles of Deep Learning Among Undergraduate Nursing StudentsFurther analysis revealed that gender, academic year, preference for the nursing major and participation in flipped classrooms were significant factors influencing the latent profiles of deep learning among nursing students. The findings indicate that male students and freshmen are more likely to adopt a "Surface Coping Learner" learning approach. Gender differences in deep learning approach remain inconsistent in existing research. Some studies suggest that female students outperform males in deep learning ( Dogham et al., 2024), while others identify males as positive predictors of deep learning ( Zhang et al., 2022). Additionally, some studies report no significant gender differences in deep learning ( Wang et al., 2024). In this study, male nursing students showed a greater tendency toward the "Surface Coping Learner" profile, which may be associated with lower interest in the nursing major and the presence of gender bias in the nursing profession. Younas et al. highlighted that males in the predominantly female nursing profession often experience greater psychological pressure and societal expectations, which may have a negative impact on their learning engagement and strategy choices ( Younas et al., 2022). Nursing education should focus on enhancing male nursing students' professional identity and providing psychological support to help them develop a positive attitude toward their major and mitigate the negative effects of external pressures on their learning behaviors.
First-year students were more likely to adopt a "Surface Coping" learning approach due to their limited knowledge base and insufficient adaptation to university learning environments. According to Biggs' theory of learning approaches, students who lack clear learning goals are more inclined to rely on surface-level strategies such as rote memorization ( Biggs, 2011). Additionally, the transition from high school to university can leave freshmen feeling disoriented, further hindering their ability to engage in deep learning( Venezia and Jaeger, 2013). To address this issue, educators should support freshmen in adapting to university learning by providing clear learning objectives, offering supportive feedback and fostering intrinsic motivation. These efforts can gradually guide students toward adopting deeper learning strategies.
Nursing students who frequently participated in flipped classrooms were more likely to belong to the 'Comprehensive Deep Learner' profile. The flipped classroom is a student-centered teaching approach that combines pre-class learning with in-class interaction, enhancing students' autonomous learning approach and fostering deeper engagement ( Zhang, 2021). According to the theory of deep learning ( F. MARTON, SÄALJÖ, R., 1976), deep learning not only focuses on the understanding and memorization of knowledge but also emphasizes students' internal integration of knowledge and critical thinking. By providing a flexible learning environment and interaction opportunities, the flipped classroom encourages students to actively participate in classroom activities, thereby deepening their learning motivation and cognitive depth ( Li et al., 2020; Xu et al., 2019).
Additionally, preference for the nursing major was found to significantly associated with learning approaches. Students who reported liking their major were more likely to adopt the "Comprehensive Deep Learner" profile, likely due to higher intrinsic motivation and greater interest in course content. According to Eccles' expectancy-value theory ( Eccles and Wigfield, 2002), the effort learners invest in a task is influenced by their perceived interest in and the importance they place on the task. In the context of our study, nursing students with a strong preference for their major likely assign higher value to the learning content, which increases their intrinsic motivation and engagement. As a result, they dedicate more time and effort to their studies, ultimately leading to a higher level of deep learning. This aligns with previous research, such as that by Zou et al. (2024), which highlighted the importance of professional identity in promoting learning engagement( Zou et al., 2024). Our findings suggest that nursing students who have a positive attitude toward their major are more likely to engage in deep learning behaviors, further supporting the role of interest and value in learning outcomes.
Therefore, nursing education should aim to develop more "Comprehensive Deep Learners" by incorporating innovative teaching methods such as problem-based learning (PBL), flipped classrooms and case-based teaching to enhance students' abilities in knowledge integration and reflection. Additionally, strengthening professional identity education is crucial to increase students' interest in and commitment to the nursing profession, while emotional support and goal-oriented guidance can help stimulate intrinsic motivation for learning. Furthermore, the assessment system should be optimized by reducing an overreliance on rote memorization exams and shifting toward comprehensive evaluation methods that emphasize practical reflection and knowledge transfer. These strategies can ultimately facilitate a greater number of students transitioning to deep learning approaches, thereby laying a solid foundation for cultivating high-quality nursing professionals.
5.3 Differences in Learning Outcomes Among Nursing Students Across Latent Profiles of Deep LearningThe results indicated significant differences in learning outcomes among nursing students across the latent profiles of deep learning. Specifically, "Comprehensive Deep Learners" exhibited the highest total learning outcome scores. This group demonstrated a holistic approach to learning, characterized by higher levels of intrinsic motivation, active exploration, integration and reflection on learning content. These attributes fostered a deeper understanding of knowledge and enhanced the ability to transfer learning to new contexts, ultimately contributing to superior learning outcomes. This finding is consistent with prior research indicating the positive relationship between intrinsic motivation and deep learning strategies ( Li, 2019; Zheng, 2019). In contrast, "Surface Coping Learners" had the lowest learning outcome scores. This profile was marked by a reliance on rote memorization and passive learning strategies, with an emphasis on short-term goals and limited motivation. Such learning approaches have been associated with poorer outcomes, aligning with earlier studies that highlight the negative impact of surface-level learning on academic performance ( Dolmans et al., 2016; Hua et al., 2023). The lower scores of "Ability-Oriented Learner" compare with "Attitude-Driven Learner" may be attributed to the former's focus on acquiring technical skills, often at the expense of developing professional attitudes and emotional engagement. On the other hand, "Attitude-Driven Learners" tended to demonstrate greater commitment to professional identity and intrinsic motivation, which contributed to higher overall learning outcomes. These findings highlight that focusing solely on skill acquisition may not suffice to achieve comprehensive learning outcomes. Instead, fostering professional attitudes and intrinsic motivation plays a more critical role in the learning process, underscoring the importance of holistic educational approaches.
6 LimitationsThis study has several limitations that should be acknowledged. First, the participants were recruited from a limited number of universities, which may restrict the generalizability of the findings to other regions or educational settings. Future studies should consider conducting multi-center educational research to increase sample diversity and explore deep learning patterns among nursing students from institutions of varying levels. Secondly, while latent profile analysis (LPA) is effective in classifying groups based on shared characteristics, it cannot fully uncover the dynamic causal mechanisms underlying the relationships between latent profiles. This limitation constrains the depth of interpretation and the applicability of the study's conclusions. Further research could employ longitudinal designs or experimental methods to explore causal pathways more comprehensively. Finally, this study relied on self-reported questionnaires, which may be subject to social desirability bias and recall bias, potentially affecting the objectivity and accuracy of the results. Future studies should integrate behavioral observations, interviews and other multidimensional assessment tools to obtain more comprehensive and precise findings.
7 ConclusionThrough latent profile analysis, this study categorizes the deep learning levels of undergraduate nursing students into four types: "Surface Coping Learner", "Ability-Oriented Learner", "Attitude-Driven Learner" and "Comprehensive Deep Learner". Factors influencing deep learning categories include gender, grade, major preference and frequency of participation in flipped classrooms. There are significant differences in the learning outcomes levels among undergraduate nursing students with different potential categories of deep learning. Therefore, nursing education should emphasize the key factors that associated with deep learning and bolster students' deep learning approaches through a variety of teaching methods as well as personalized intervention strategies.
Ethics approval and consent to participantThis study was conducted in accordance with the Declaration of Helsinki, and the protocol was approved by the Harbin Medical University- Daqing campus(HMUDQ-20240309011). Informed consent was obtained from all individual participants included in the study. Participants were provided with detailed information regarding the study's purpose, procedures, potential risks and benefits, and confidentiality measures. Written consent forms were signed by all participants prior to their involvement in the research.
FundingThis study was supported by the Heilongjiang Province Higher Education Teaching Reform Research General Project [number SJGY2024268] and Heilongjiang Province planning Office Key Topic [number GJB1424159] and Heilongjiang Provincial Philosophy and Social Science Research Planning Project ( 21SH214 and 21RKC212) and Fundamental Research Funds for the Provincial Universities ( JFYQPY202302).
CRediT authorship contribution statementYang Li: Formal analysis, Data curation. Xiangzi Ji: Writing – review & editing, Writing – original draft, Methodology, Investigation, Conceptualization. Jiayuan Zhang: Writing – review & editing, Writing – original draft, Funding acquisition, Formal analysis, Data curation, Conceptualization. Hui Zhang: Formal analysis, Data curation. Lina Meng: Writing – review & editing, Writing – original draft, Supervision, Funding acquisition, Formal analysis, Conceptualization.
Declaration of Competing InterestNone.
AcknowledgmentsWe thank all the participants for their corporation.
| Model | AIC | BIC | aBIC | Entropy | P(LMRT) | P(BLRT) | Probabilities of Classes |
| 1 | 25797.863 | 25845.786 | 25814.028 | - | - | - | 1 |
| 2 | 23086.926 | 23163.603 | 23112.790 | 0.998 | 0.0000 | 0.0000 | 0.47587/0.52413 |
| 3 | 21927.918 | 22033.350 | 21963.482 | 0.993 | 0.0000 | 0.0000 | 0.47475/0.25814/0.26712 |
| 4 | 20908.713 | 21042.898 | 20953.976 | 0.990 | 0.0000 | 0.0000 | 0.25477/0.25814/0.21773/0.26936 |
| 5 | 20630.982 | 20793.922 | 20685.944 | 0.971 | 0.0000 | 0.0000 | 0.08866/0.20875/0.17957/0.26936/
0.25365 |
| Item | Total | Surface Coping Learners (n = 230) | Ability-Oriented Learners (n = 227) | Attitude-Driven Learners (n = 194) | Comprehensive Deep Learners (n = 240) | X 2 | | |
| Gender | Male | 101 | 27 (26.74 %) | 36 (35.64 %) | 20(19.80 %) | 18(17.82 %) | 8.38 | 0.04 |
| Female | 790 | 203 (25.70 %) | 191(24.17 %) | 174(22.03 %) | 222(28.10 %) | |||
| Grade | Freshmen | 197 | 76 (38.58 %) | 41(20.81 %) | 48(24.36 %) | 32(16.25 %) | 35.15 | < 0.01 |
| Sophomore | 232 | 51 (21.98 %) | 63(27.16 %) | 56(24.14 %) | 62(26.72 %) | |||
| Junior | 255 | 62 (24.31 %) | 63(24.71 %) | 53(20.78 %) | 77(30.20 %) | |||
| Senior | 207 | 41(19.81 %) | 60(28.99 %) | 37(17.87 %) | 69(33.33 %) | |||
| Region | Urban | 530 | 127 (23.96 %) | 140(26.42 %) | 120(22.64 %) | 143(26.98 %) | 2.64 | 0.45 |
| Rural | 361 | 103 (28.53 %) | 87(24.10 %) | 74(20.50 %) | 97(26.87 %) | |||
| Single child | Yes | 614 | 152 (24.76 %) | 156(25.41 %) | 137(22.31 %) | 169(27.52 %) | 1.38 | 0.71 |
| No | 277 | 78 (28.16 %) | 71(25.63 %) | 57(20.58 %) | 71(25.63 %) | |||
| Major Attitude | Like | 352 | 77 (21.88 %) | 68(19.32 %) | 91(25.85 %) | 116(32.95 %) | 25.08 | < 0.01 |
| General | 365 | 100 (27.40 %) | 110(30.14 %) | 70(19.18 %) | 85(23.28 %) | |||
| Dislike | 174 | 53 (30.46 %) | 49(28.16 %) | 33(18.97 %) | 39(22.41 %) | |||
| Participation in Flipped classroom | Always | 216 | 27 (12.50 %) | 75(34.72 %) | 44(20.37 %) | 70(32.41 %) | 32.57 | < 0.01 |
| Not always | 675 | 203 (30.07 %) | 152(22.52 %) | 150(22.22 %) | 170(25.19 %) |
| Ability-Oriented Learners | Attitude-Driven Learners | Surface Coping Learners | |||||||||||||
| β | SE | Wald X 2 | | OR(95 %CI) | β | SE | Wald X 2 | | OR(95 %CI) | β | SE | Wald X 2 | | OR(95 %CI) | |
| Gender | 0.83 | 0.33 | 6.44 | 0.01 | 2.30
(1.21,4.37) | 0.69 | 0.35 | 3.95 | 0.04 | 2.00
(1.01,3.96) | 0.63 | 0.36 | 3.02 | 0.08 | 1.88
(0.92,3.83) |
| Male a | |||||||||||||||
| Grade | |||||||||||||||
| Freshman b | 0.54 | 0.30 | 3.15 | 0.07 | 1.71
(0.94,3.11) | 1.55 | 0.30 | 26.37 | < 0.01 | 4.70
(2.60,8.48) | 1.17 | 0.31 | 13.91 | < 0.01 | 3.22
(1.74,5.95) |
| Sophomore b | 0.30 | 0.26 | 1.28 | 0.25 | 1.35
(0.80,2.28) | 0.43 | 0.29 | 2.24 | 0.13 | 1.54
(0.88,2.70) | 0.61 | 0.28 | 4.50 | 0.03 | 1.83
(1.04,3.20) |
| Junior b | 0.12 | 0.27 | 0.19 | 0.66 | 1.12
(0.67,1.90) | 0.34 | 0.28 | 1.43 | 0.23 | 1.40
(0.80,2.43) | 0.26 | 0.29 | 0.85 | 0.36 | 1.30
(0.74,2.28) |
| Major Like | |||||||||||||||
| Dislike c | −0.74 | 0.30 | 6.12 | 0.01 | 0.47
(0.26,0.85) | −0.08 | 0.30 | 0.06 | 0.80 | 0.93
(0.52,1.65) | 0.28 | 0.31 | 0.83 | 0.36 | 1.33
(0.72,2.44) |
| General c | 0.14 | 0.28 | 0.25 | 0.62 | 1.15
(0.65,2.02) | 0.22 | 0.29 | 0.59 | 0.44 | 1.25
(0.71,2.12) | 1.23
(0.66,2.27) | ||||
| Participation in Flipped classroom | 0.42 | 0.22 | 3.65 | 0.06 | 1.52
(0.98,2.34) | −1.15 | 0.27 | 18.32 | < 0.01 | 0.32
(0.18,0.53) | −0.48 | 0.24 | 3.97 | 0.04 | 0.62
(0.38,0.99) |
| Not Always d | |||||||||||||||
| Variables | Surface Coping Learners | Ability-Oriented Learners | Attitude-Driven Learners | Comprehensive Deep Learners | Overall test | |||||||
| X 2 | | |||||||||||
| Overall learning gain | 2.99 ± 0.04 | 3.23 ± 0.03 | 3.64 ± 0.06 | 4.03 ± 0.05 | 275.68 | < 0.001 | ||||||
| Emotional, attitudinal, and value gains | 18.44 ± 0.21 | 19.06 ± 0.16 | 20.58 ± 0.19 | 21.42 ± 0.20 | 142.38 | < 0.001 | ||||||
| Cognitive gains | 6.157 ± 0.10 | 6.72 ± 0.08 | 7.11 ± 0.09 | 7.70 ± 0.12 | 108.87 | < 0.001 | ||||||
| Skill gains | 16.31 ± 0.22 | 17.99 ± 0.15 | 17.33 ± 0.20 | 19.94 ± 0.19 | 168.49 | < 0.001 | ||||||
| Total Score | 43.90 ± 0.53 | 47.01 ± 0.35 | 48.65 ± 0.46 | 53.09 ± 0.48 | 181.58 | < 0.001 | ||||||
©2025. Elsevier Ltd