1 Introduction

Higher education curricula in Science and Engineering are structured to provide students with a blend of theoretical and practical education. Theoretical components emphasize overall knowledge while practical education emphasizes skill training applicable to the selected field of study. There is growing concern on the shortcomings of current educational practices (Michael et al. 2010).

Sciences require understanding and the practice of reflection enhances visualization of relationships resulting in improved retention of ideas or concepts. The relevance and impact of reflective thinking have been recorded across many studies with diverse audiences (Bubnys and Žydžiūnaitė 2010; Colomer et al. 2013). To understand what assisted reflective thinking, Van Mannen (Manen 1977) & Mezirow (Mezirow 1997), Jacob & Murrays (Jacobs and Murray 2010) indicated several levels of hierarchy that include critical thinking, understanding and habitual reflection, reasoning etc. A study had (Peltier 2006) summarized reflection as learning meanings from experiences.

Multiple studies have investigated how computer simulations have often been used to enhance traditional instruction for Science, Technology, Engineering and Mathematics (STEM) education (Colomer et al. 2013). Using simulations to augment visualization as well as interactivity, and engaging a student’s attention to concepts with exercises connected to learning have shown significant benefits in the learning outcomes (Rushton and Lahlafi 2013). It has been suggested that institutions could allow better access to virtual and online laboratories as collaborative learning environments (Bourne et al. 2005). To quantify augmented learning outcomes, measures through test scores, student engagement with reading material, sense of community among learners and reduced drop-out or failure rates have been employed. (Nguyen 2015).

In many developing countries, that have large and growing populations of students pursuing higher education and a limited number of educational institutions with required facilities, resources to match the needs are often inadequate. Educational resources include availability of trained teachers, laboratory infrastructure to support a large variety of experiments and the facilities required to run them on a frequent basis. Even in well-equipped institutions in India, the standard practice in undergraduate and often postgraduate laboratory education is to have groups of students perform a single experiment in the lab.

Although, collaborative problem-solving with a team enhances learning, often in a laboratory environment, grouping students to perform experiments results in incomplete training of individuals due to shared resources-time for experimentation. This disparity in learning comes from the inherent nature of groups that have both active and passive participants. Certain students often actively do the experimental portions while others passively participate. The goal of laboratory education, i.e., to give practical hands-on training to enhance conceptual understanding to every student remains compromised.

The purpose of this study was to investigate changing learning modalities of laboratory environments with focus on interventions to improve reflective thinking in skill education. The study aims at assessing reflective tendencies and encouraging them by integrating laboratory practical education with Virtual Laboratories (VLs).

2 Background

There are several theories characterizing the stages and processes involved in deep or transformative learning. Dewey’s model (Dewey 1938) of interconnecting educative experiences to reflection, Kolb’s model (Kolb 1984) of experiential learning, Schon’s model of (Schön 1983) reflection as a practice, Gibb’s model of cross-enrichment of theory and practice to enhance reflection (Gibbs 1988), Moon’s model (Moon 2013) of provoking interpretations to complex ideas look into sequential set of events and learning. Another study (Bransford et al. 1999) categorizes teaching strategies as lecture-based, skill-based, inquiry-based, individual vs. group based and technology-enhanced learning. That study also listed three components contributing to reflective learning amongst students namely, metacognition, analogical reasoning and transfer of knowledge.

John Flavell (Flavell 1976) defined metacognition as “one’s knowledge concerning one’s own cognitive processes and products or anything related to them". Eggen & Kauchak (Eggen and Kauchak 1988) suggested the development of metacognitive skills was key to creating self-regulated learners. When considering student communities, metacognition was expected to impact many cognitive activities such as comprehension, communication, attention, memory and problem solving (Howard 2004).

Success of an individualistic learning process may be assessed by the extent of application in practical scenarios, often referred to as transfer of knowledge. Nonaka (Nonaka and Takeuchi 1995) suggested that knowledge creation and transfer happened through social or collaborative methods, as well as the individual’s cognitive processes such as reflection. Transfer of knowledge assisted with engagement in tasks based on the knowledge gained from learning (Jun et al. 2010), (Chiou et al. 2010). Transfer of knowledge was assessed by analogical reasoning, a process of mapping a known source or analog to an unknown target. This mapping includes the process of identifying, matching and transferring the structural information (Vosniadou and Ortony 1989).

Computer-mediated learning tools have been used in STEM education to assist students with visualization of complex phenomena (Achuthan et al. 2015). Most knowledge imparted in classrooms was theory-oriented, while that gained in laboratories was skill and concept-oriented. Maximizing learning by enhancing understanding that results in better grasp of knowledge is quite challenging (Naghdipour and Emeagwali 2013). In this paper, we also consider the impact of virtual labs training on reflective learning in University classrooms.

2.1 Virtual laboratories

Similar to MOOCs as learning environments (Laplante 2013), emphasis on simultaneous development of laboratory oriented educational material has led to the development of virtual laboratories, allowing learners to understand and practice hands-on skills or protocols outside of the physical laboratory (Diwakar et al. 2012). ICT-enabled VLs allow self-controlled individualistic learning environments for students, which has been known to complement standard laboratory practices. The virtual labs used here were developed as an initiative of Government of India’s Ministry of Human Resource Department under National Mission on Education through ICT, and were designed and developed to provide laboratory learning time and experience to university students, who do not have access to adequate laboratory facilities or equipment. In this paper, we used one of the 43 Virtual Laboratories (350 online experiments) in several disciplines (Raman et al. 2014) that were developed (see Fig. 1). These VLs were posed also as supplementary teaching tools with varied interactive multimedia content to emulate real laboratory experiences. With more than 230,000 registered online users, this platform has served to partly fill the large gap seen in laboratory education by providing learners with 24 × 7 access to a) simulation-based, b) animation-based and c) remotely triggerable experimental interfaces to scientific concepts and instrumentation. In a physical lab, students work in groups, while in a virtual lab, students interacted with specific context-sensitive graphical elements developed to aid individual concept learning.

Fig. 1
figure 1

ICT-based Virtual Laboratories used to test reflective learning and to augment laboratory skill training. A. Online experiments included theory, procedure, self-assessments, assignment, feedback in addition to simulation, animation and remote access tabs. Each virtual lab experiment was meant to be self-explanatory although they have been employed as interactive teaching material as well. B. Total usage of Virtual Laboratories. Online and blended usage numbers (log-scale on y-axis) across 2013–15 of Virtual Laboratory experiments. More than 350 experiments are hosted at http://vlab.amrita.edu

Full size image

The VL platform has grown rapidly with millions of users around the world using them today. The virtual labs have been reported to have more than 230,000 registered users as of January 2, 2017, with a steady increase of new users since 2013 (see Fig. 1b ). We saw two reasons for the adoption and usage of the Virtual Laboratories namely, availability of a single platform and usability as an Open Educational Resource (Raman et al. 2014). VLs have been perceived also as supplementary teaching tools with interactive multimedia content allowing an instructional preview to physical laboratory demos (Diwakar et al., 2014a, b, 2016).

3 Methods

Participants in this study were first year undergraduate engineering students, N = 145 (52 male, 93 female). About 74% of students were from computer science discipline and 26% of students were from the biosciences stream. These students were introduced to two types of experimental environments i.e. physics laboratory as well as the same technique via VLs. Students are categorized in to two groups: CPV (N = 73) students went through class room teaching (CT) then Physical Laboratory (PL) and finally performing Virtual Laboratory (VL); CVP (N = 72) students went through CT, VL then PL. A new assessment (see section 3.1) was conducted between the sessions (Fig. 2). Assessment data was collected after physical lab training as well as virtual lab training. Although data from 162 students was collected only 145 were considered for the assessment (some students were absent in one of the sessions and hence those data points were removed).

Fig. 2
figure 2

Assessment models of MAT

Full size image

The physical laboratory and virtual lab experiment chosen for this study involved a spectrometer, a common experiment that all undergraduate students train on as part of their classroom theory as well as laboratory education. Learning objectives from this experiment involved not only the understanding of the concept of the experiment (access via http://vlab.amrita.edu/?sub=1&brch=281) in addition to the physical operation of the instrumentation (see Fig. 3) involved in deriving the quantitative estimation of experimental measurements (Raman et al. 2011, 2014). All students who had used Virtual Laboratory (VL) and Physical laboratory (PL) learning environments were assessed on their understanding of concepts and its application.

Fig. 3
figure 3

Physical and Virtual Lab training. a Student users in a physical laboratory. b Augmenting learning experience via Virtual labs post physical laboratory experience

Full size image

Students spent approximately 2 h interacting and working on the online experiment. The interactive VLs had animations, simulations and videos, in addition to extensive theory and procedures They also included practice assignments and self-evaluation tests. An assessment was performed after the students attempted VL based learning. A retention study was also conducted by assessing the same students periodically over a 7-week period. We also conducted the retention study for a separate group of students, termed “class room training group” (CT), involving 45 student learners who did not use virtual or physical laboratories and were not a part of students used to test CVP and CPV test groups. Participants in CT group do not have access to physical or virtual laboratory but underwent classroom theory lessons using laboratory manuals as in traditional classroom courses. In our studies, since laboratory time was not included for CT group, they had one class hour less in the weekly routine compared to students in CPV and CVP groups. CT group was used as independent control group and as a reference to classroom teaching content only with lab skill education or practice.

3.1 Development of assessment method

Metacognition

Metacognition has been defined as the process of developing self-awareness along with the ability to self-assess one’s learning (Flavell 1976). Having metacognitive skills presumes that learners can set their learning goals, understand their learning styles and evaluate their own learning. Metacognition has been suggested to be used to assess factors assisting a student in gaining expertise (Sternberg 1998). The work involved the development of a questionnaire that assessed the metacognitive elements in learners to correlate reflective learning ability to the learning styles. In order to measure the metacognitive knowledge of students and effect of virtual lab on students self-awareness related to the laboratory knowledge, a questionnaire was made based on motivated strategies for learning (Pintrich et al. 1993), the metacognitive awareness inventory (Schraw and Dennison 1994) and awareness of independent learning inventory (Meijer et al. 2013). Questions related to metacognition included: whether the student learners had been introduced to a spectrometer earlier, what they felt about handling the equipment used in the experiment, concepts related to the device, theoretical principles of operation etc. It is to be noted that students may have different subject interests, i.e., some may have a larger affinity towards electronics in comparison to optics etc. This may also be attributed to their skill and therefore needed to be identified.

Analogical reasoning

Analogical reasoning has been categorized as a process through which learners adapt their knowledge from a familiar scenario to an unfamiliar case (Vosniadou and Ortony 1989). Several studies on analogical reasoning have been done to measure the impact of innate analogical reasoning skills in learning (Abrantes 1999; Richland and McDonough 2010). The questionnaire assessed learners post laboratory process via computing specific parameters related to the experiment. Questions related to analogical reasoning in the current study included observations of parameters of the spectrometer, such as: calculation of angle of incidence, minimum deviation, refractive indices of materials or reading the Vernier scale etc.

Transfer of knowledge

Transfer of knowledge was defined as when knowledge constructed in a contextual content was used in a different context, after being mobilized or recombined by the learner. This also depended on the extent of training one had to perform the task (Alipour et al. 2011). Transfer of knowledge using computer mediated learning tools provided not only technical challenges, but also many research opportunities (Jun et al. 2010). But, failures in transfer of knowledge have been known to occur when tasks were ‘over-embedded’ within the practice context (Detterman and Sternberg 1993). The transfer of knowledge questionnaire used in this study included application-oriented questions requiring descriptive type responses. Sample questions that related to the transfer of knowledge skills by students included: how easily they could translate what they read in the manuals to actual experimentation or if they could trace the path of light with different prism angles and refractive indices.

3.2 MAT instrument

The objective of MAT questionnaire was to assess underlying factors that contributed towards learning in laboratory environment. The questionnaire was developed integrating reflective metacognition and self-regulated learning (Ertmer and Newby 1996). An initial questionnaire taxonomy was developed and further refined into a 48-item list involving consultation with subject experts in physics with over 10 years of teaching experience. The list was administered to a set of students outside of the study and was evaluated for content, clarity and its alignment to the attributes. A reliability analysis was further conducted prior to this study (Creswell 2013). Cronbach’s alpha of the MAT questionnaire that had the three observables was 0.85, making it a reliable instrument for identification of learning skills. The final questionnaire included questions on metacognition, analogical reasoning and transfer of knowledge.

The assessment of the three attributes was performed using a combination of Likert style questions (LS), multiple choice questions (MCQs) and True/False statements as well as descriptive questions. The questionnaire was administered in three formats: I – A questionnaire using a five-point Likert-type scale with anchors from “strongly disagree” to “strongly agree”, II– An objective questionnaire that had a combination of multiple choice questions as well as True/False statements, III -a descriptive questionnaire.

Part I (30 questions) and Part II (10 questions) were completed by students in 30 min per section and were administered sequentially. An hour was given to the participants to solve Part III (8 questions) (also see Table 1). In Part I, about 70% of questions related to metacognition, 20% to analogical reasoning and 10% to transfer of knowledge. In Part II, all questions were structured as multiple choice questions and True/False statements. The MCQ style questions had 20% related to assessing metacognition, 40% analogical reasoning and remaining 40% related to transfer of knowledge. The True/False questions in Part II had 80% of questions related to metacognition and 20% related to analogical reasoning. The scores allotted for MCQ and True/False questions were one mark each. Part III had questions that were either descriptive short answer questions or were mathematical in nature for evaluating their ability to think analogically and transfer this knowledge in a problem-solving environment. The scores allotted for Part III questions varied between 4 and 20 marks. Scores were allotted based on number of concepts in the questions. In this section, 50% of questions related to analogical reasoning and 50% to transfer of knowledge.

Table 1 Distribution of questionnaire for metacognition, Analogical reasoning and Transfer of knowledge
Full size table

Among three student groups CVP, CPV and CT, CVP and CPV were used to compare the learning roles during class room training, virtual laboratory training and physical laboratory training. CT group (45 students) was a classroom-only group without lab experience and was used for the retention study. MAT instrument was used to assess learning attributes (M, A and T) in CPV and CVP. As indicated, the MAT instrument contained three sections namely Likert style (LS), multiple choice (MCQ) and descriptive-type questions. Three assessments, first (FA), second (SA) and third (TA), using the same set of MAT questions were conducted after classroom training, virtual lab training and physical lab training on CVP and CPV groups. First assessment (FA) was conducted after class room-based teaching alone. SA with same questionnaire was performed for CPV after physical lab training and virtual lab training for CVP group was conducted. TA was performed after physical lab training for CVP and virtual lab training for CPV groups using the same questionnaire. Retention study was conducted every week over a 7-week duration using MCQ and descriptive type questionnaire in the MAT instrument. All three groups (CPV, CVP, CT) went through the retention study.

3.3 Designing the questionnaire for the assessment.

On the basis of the previous literature, we have developed domains for evaluating metacognition, analogical reasoning and transfer of knowledge. Each observable (i.e. M, A, T) was defined by four distinct domains and set of questions were prepared to access each of the domains that combine to say the effect of M, A and T. The domains for the metacognition was derived from the past studies of metacognition and its assessments, were categorized as ‘Self-reported questionnaire based on subject domain’ (Desoete et al. 2003; Kramarski and Mevarech 2003; Meijer et al. 2006; Veenman and Spaans 2005), ‘Statements to support student’s verbalized self-instructions’(Brown 1978) (Brown et al. 1981), ‘Statements analyzing student’s observations’ (Jaafar et al. 2014; Pintrich and De Groot 1990; van Hout-Wolters 2000) and ‘Statements to analyze stimulated recall with in the subject domain’ (Afflerbach 2000; Veenman et al. 1993) (see Table 2). Self-reported questionnaire based on subject domain described types of questions to assess student’s thoughts, beliefs and perceived knowledge about the topic.

Table 2 Sample questionnaire based on the four domains assessing metacognition
Full size table

Analogical reasoning was known to take place between two systems that belong to different domains, or between two systems that belong to the same or similar domains (Gentner and Toupin 1986). Inductive reasoning was considered to be an important element of thinking and learning that is particularly problematic for those with moderate and severe learning difficulties (Büchel et al. 1997). For assessing the analogical reasoning, the domains are ‘remote conceptual domain’ (Gentner and Toupin 1986), ‘identifying surface similarity’ (Vosniadou 1989), ‘discovery of new knowledge’(Chen et al. 1997; Gentner and Toupin 1986; Goswami 2001; Hesse 1966; Waltz et al. 2000)., ‘analogical mapping of domain’ (Gentner 1983, 1986; Gentner and Landers 1985) (Table 3). Remote conceptual domain represented the analogy from near relation of the subject domain, and could connect the concept to the learner. For example, − “Refractive index of the prism is always greater than the refractive index of the surrounding”. To answer this question, students need to resolve the solution by thinking analogically connecting the path of light travelling through the prism and the bending occurs due to the transmission of light from the prism to the surrounding. But this ‘domain’ does not have any direct relation to the aim of the experiment. The next ‘domain’ called ‘identifying surface similarity’ described analogical reasoning by comparing a schematic figure along with the questionnaire. ‘Discovery of new knowledge’ was a type of questionnaire which help to resolve the analogical reasoning question along with understanding of new perspective of learning. For example – “The edges of the prism have equal priority in the spectrometer experiment”. With the help of a figure, student can identify the analogy of the two figure and understand new knowledge. The fourth domain called ‘analogical mapping of domain’ helped to assess the reasoning ability of the students with the help of the statement or a situation. This type of questionnaire was developed to motivate the students to think about the analogy in the procedure or situations and to create the connection with given statements.

Table 3 Sample questions based on the domain for assessing analogical reasoning
Full size table

Transfer of knowledge was a construct of enormous importance for both education and practice (Lauder et al. 1999). An individual perspective on knowledge and learning enables us to explore differences both in what and how people learn and in how they interpret what they learn. (Daniels et al. 2009). This transfer of knowledge was accomplished by mapping or matching processes which consist of finding the correspondences between the two systems (Gentner and Toupin 1986). Transfer of knowledge and skills has been described as the main goal of education (Prawat 1989), and as the ultimate goal of education for lifelong learning (Fogarty 1995). For the assessment of the transfer of knowledge we derived the domains from the basis of the previous literature, called ‘access the ability to access one’s intellectual resources’ (Prawat 1989) (Renkl et al. 1996), ‘mapping process to find the correspondences between two systems’(Lauder et al. 1999), ‘access the ability to use certain procedures’ (Lauder et al. 1999). An individual perspective on knowledge and learning enables us to explore differences both in what and how people learn and in how they interpret what they learn (Daniels et al. 2009) between the two systems (Gentner and Toupin 1986). Transfer of knowledge and skills has been described as the main goal of education (Prawat 1989), and as the goal of education in lifelong learning (Fogarty 1995). For the assessment of the transfer of knowledge, we derived the domains based on literature, called ‘access the ability to access one’s intellectual resources’ (Prawat 1989), (Renkl et al. 1996), ‘mapping process to find the correspondences between two systems’ (Lauder et al. 1999), ‘access the ability to use certain procedures’ (Lauder et al. 1999). An individual perspective on knowledge and learning enables us to explore differences both in what and how people learn and in how they interpret what they learn. (Daniels et al. 2009) (Gentner and Toupin 1986). Transfer of knowledge and skills has been described as the main goal of education (Prawat 1989), and as the objective of education for lifelong learning (Fogarty 1995). For the assessment of the transfer of knowledge, literature based assessment domains included, ‘ability to access one’s intellectual resources’, ‘mapping process to find the correspondences between two systems’, ‘access the ability to use certain procedures’, ‘subject knowledge and metacognitive domain’ (Table 4). First domain describes the types of questions that can access the intellectual knowledge of the learner and apply to a scenario. The second domain assessed the ability to apply the known knowledge to identify the relation between two systems. Third domain interpreted one’s ability to apply the learned procedure in a correct order. Fourth domain categorized to identify the subject knowledge and metacognitive ability by applying the knowledge to resolve subject related problems.

Table 4 Sample questions for assessing transfer of knowledge
Full size table

4 Results

An assessment of metacognition, analogical reasoning and transfer of knowledge of all students post classroom teaching was performed. A significant number of student scores were below 50%. Approximately 62% of students scored less than 50% in metacognition, while 69% scored less than 50% in analogical reasoning and 54% scored less than 50% in transfer of knowledge (see Fig. 4, Tables 5 and 6).

Fig. 4
figure 4

Distribution of score of three observables (M-Metacognition, A- Analogical Reasoning, T- Transfer of Knowledge) in the first assessment (FA)

Full size image
Table 5 T-test of CPV and CVP on first assessment (FA)
Full size table
Table 6 Score distribution in MAT assessment
Full size table

4.1 Impact of training on virtual and real laboratories

The students were split into two equal groups i.e. CPV and CVP (Figs. 2 and 3a, b). Three assessments of the three attributes were done to both groups. The first was after the classroom teaching, the second was done after the virtual labs for the C″V″P group and after physical labs for the C″P″V group. The third assessment was done after the physical lab for the CV”P″ group and after the virtual lab for the CP”V″ group (see Fig. 2). Post classroom training, the groups had similar average score and standard deviations (see Table 5). The combined mean scores of M, A and T were 47.71 and 48.95 for CPV and CVP groups. T-test shows t(144) = −0.69, p > 0.05 which represents there is no significant difference in groups based on the assessment of the attributes.

The students in CVP group underwent VL followed by real Labs (PL) while those in CPV were trained first on PL followed by VL. All three attributes were again assessed after both VL and PL in both groups.

4.2 Data analysis

Variation of score showed the effect of learning from physical laboratory and virtual laboratory. From the metacognition assessment (Tables 7, 10, 11), the gain in scores at the end of physical labs for the CPV group was 25% (Mean = 62.5) while at the end of virtual lab sessions for the CVP group were around 55% (Mean = 77.5). When both sets of groups underwent both physical and virtual labs, the overall increase in scores ranged between 70% (Mean = 84.7) to 83.3% (Mean = 92.2) for the CPV and CVP groups respectively. A sample t-test was conducted to study the variation in the score in-between each assessment for CPV and CVP groups (Table 7). Analysis showed a significant difference in the scores when students underwent virtual labs training (t (71) = −22.7, p < 0.05) before handling physical labs (t (72) = −12.0, p < 0.05). The change was further perceived in the follow up assessments of CPV group that underwent one more round of learning from virtual labs after performing experiments in the physical lab. The mean score increased from 62.5 to 84.7 or a 35% increase in the CPV group in comparison to an incremental 19%. Mean score improved for students who did virtual labs before physical labs (Fig. 6). Another indication of the improved performance from virtual lab-based training was also seen as the higher slope observed for the CVP group in comparison to the CPV.

Table 7 Comparing metacognition scores for first, second and third assessments
Full size table

At the end of first assessment (after classroom teaching), the scores were similar with less than 3% difference between the CPV and CVP groups for analogical reasoning (see Fig. 5, Tables 8 9). However, the change between the groups in the second assessment when students were trained also on virtual labs was significantly higher with an increase 55% (mean score increases from 45.7 to 71.0) for CVP batch of students (see Table 8). The increase due to influence of training with physical labs in the CPV group after classroom teaching was approximately 22.57%, mean score changes from 44.3 to 54.3(see Tables 8 and 9). At the end of third assessment i.e. training on physical labs after virtual labs in the CVP group and training on virtual labs after physical labs in the CPV group, the scores converged to 84.6 (see Table 8). The results from p-value and t-test indicated significant changes in the second assessment after classroom teaching and with the third assessment (see Table 10). Also, CVP shows more significant differences (t(71) = −23.31, p < 0.05) in comparison to the CPV group (t(72) = −17.4, p < 0.05). Like metacognition (Table 7), in the case of analogical reasoning assessments (Table 8), the slope reflecting the increase in scores across the three assessments indicating a steeper increase for the CVP group in comparison to CPV.

Fig. 5
figure 5

Variation of score of CPV and CVP in three assessments of metacognition, analogical reasoning and transfer of knowledge. A. Metacognition for CPV vs CVP. The slope of the average score of CPV, m1 = 17.35 while the average score of CVP, m2 = 20.94. B. Analogical reasoning for CPV vs CVP. The slope of the average score of CPV, m1 = 15.54 while that of CVP, m2 = 20.39. C. Transfer of knowledge for CPV vs CVP. The slope of the average score of CPV, m1 = 16.33 while that of CVP, m2 = 20.55

Full size image
Table 8 Comparison of analogical reasoning scores for first, second and third assessments
Full size table
Table 9 Transfer of knowledge scores comparison for first, second and third assessments
Full size table
Table 10 Comparison of average scores for MAT assessments among groups
Full size table

After classroom training, both groups CPV and CVP as in the case of prior attributes did not show significant differences for transfer of knowledge (see Fig. 5a, Table 11). In the CVP group with virtual labs after the classroom training, at the end of the second assessment a recognizable improvement in the scores of 58.67% (see Table 11), with mean scores changing from 51.3 to 81.4, was observed (see Fig. 5b, Table 9). In case of CPV group, at the end of second assessment the change in scores was around 23.50% (with mean scores increasing from 50.3 to 62.0). In the third assessment, there was a 33.70% change in the CPV group that had virtual labs training after physical labs. In the CVP group, however, the change in scores was about 13.14% (i.e. from 81.4 to 92.1). The statistical tests i.e. t-test and p-test showed that the increase in scores in the second assessment in the CVP group were higher (t(71) = −23.7, p < 0.05) in comparison to CPV group (t(72) = −19.1, p < 0.05). Final scores for both CPV and CVP groups varied between 82.9 to 92.1.

Table 11 Percentage of score after second and third assessment
Full size table

A common observation was that the significant variation to mean scores happened only when students underwent virtual lab training in both groups (see Table 10). In the CPV group, the steepest increase in mean scores were observed in the third assessment i.e. mean score increased from 58.8 to 80.9. The third assessment (see Fig. 5c) in CPV group was done after virtual lab session. In case of CVP group, the second assessment showed the highest increase in scores i.e. mean scores increased from 49.1 to 76.7.

One-way MANOVA revealed a significant multivariate main effect for region, Wilks’ λ = 0.261, F (9, 135) = 42.55, p < 0.001, partial eta squared = 0.74. Power to detect the effect was 1 confirming significant differences between student groups, CVP and CPV. Percental univariate main effects for CPV and CVP groups were obtained for FA to TA (Table 10). First assessment (FA) indicated no significant change between two student groups CVP and CPV. Second and third assessments showed relative variations, F value indicated significant impact of virtual labs on student learners through a score improvement during second assessment (SA) and during third assessment (TA) after virtual and physical laboratory exercises (see Table 12).

Table 12 Tests of Between-CPV and CVP
Full size table

4.3 Improvement in answering of questions

Independent analysis was conducted to estimate the change in the number of responses as a function in each assessment and types of questions. The types of questions included a combination of Likert style, multiple choice questions (MCQs) and True/False statements as well as descriptive questions show progression of the percentage of questions attempted by students in the first, second and third assessments by CPV and CVP groups respectively. For the CPV group, in the first assessment, approximately 23%, 42% and 60% attempted the descriptive questions, True/False and MCQ questions respectively (see Fig. 6a). In the second, the attempts increased between 60% - 70% at the end of physical lab while at the end of third assessments these numbers are between 93% - 100% (Fig. 6b). For the CVP group, in the first assessment, the questions attempted are 25%, 35% and 62% for descriptive, MCQ and True/False questions. In the second assessment, a raise to 79% - 89% was observed during evaluation after virtual lab practice and 100% was perceived at the end of third assessment (Fig. 6c).

Fig. 6
figure 6

Extent of questions answered by learners during CPV and CVP. a, d are scores during first assessment. b, e are scores during second assessment. c, f are scores during third assessment. See noticeable changes soon after second assessment with CVP where virtual lab training came after classroom training and before training with physical labs

Full size image

4.4 Retention of gained knowledge

To verify if retention of gained skill knowledge was enhanced, all students were subjected to another set of periodic assessments. The assessments were conducted every week over a 7-week duration, using MCQ and descriptive type questionnaire. Students with VL training showed reliable retention of knowledge after several weeks of learning, demonstrated by negligible change in the scores over the period of assessment (Fig. 7). To compare the effect of VL, we used an independent group of students (not related to CVP and CPV) who had only undergone class room training (CT). We observed a rapid decline in scores among those that had classroom training only compared to CVP and CPV groups.

Fig. 7
figure 7

Assessment of retention among CPV, CVP and class room teaching (CT) groups

Full size image

5 Discussion

Through usage and feedback analysis from other studies, virtual labs have already been observed to complement laboratory skill education by providing learners with 24 × 7 access to a) simulation-based, b) animation-based and c) remotely triggerable experimental interfaces to scientific concepts and instrumentation (Diwakar et al. 2016; S. Diwakar et al. 2014a, b; Radhamani et al. 2014; Sasidharakurup et al. 2015). In applying virtual labs as an educational tool, learner’s proclivity to adapt the learning components was as relevant as transfer of knowledge, thereby allowing metacognition, analogical reasoning as components of the new assessments.

With student learners as CVP and CPV groups allowed us to understand roles of including virtual laboratories as pre or post-laboratory training exercise, where student groups who had virtual laboratory training showed more than 100% perceived understanding of concepts compared to those who only attended classroom lectures. The changes in the mean scores and slopes during MAT assessment also suggested that reflective learning was effective when virtual labs followed classroom training. A common observation with the MAT scoring was that the significant differences in scores happened when students had training with virtual labs in both groups. The use of virtual labs as a pre-laboratory or post-laboratory education platform may be used to help or identify learners with difficulties in understanding of concepts and applying the knowledge gained in newer scenarios.

With student’s metacognitive level significantly increased after performing virtual laboratories, this study suggests the self-knowledge about a subject was enhanced. Teacher’s metacognition on the topic also affected the overall performance of the students. In a classroom environment without ICT tools, teaching analogy in two different situations is not easy and time-consuming. Perhaps, this reflected on some students in their ability to interpret and solving the problems. Through animations and simulations, virtual lab helped students to augment analogical reasoning. Significant difference among groups in transfer of knowledge indicated, students with virtual lab training demonstrated better understanding about the subject and can use their training and perceived knowledge to solve problems. Virtual lab practice aided learners to improve the ability to apply knowledge in addition to providing basic concepts.

It may be essential to study student’s proclivity in larger populations of subjects with different backgrounds as well across several subjects of study, to provide a more generalized understanding of how reflective learning was enhanced by using computerized emulations. We also noted that information retention in students improved with virtual laboratory training alongside physical laboratories during the 7-week tests. With the MAT domains as a reference, it may be proposed that VLs are effective and complementary ICT-based tools to support laboratory skill-training.

6 Conclusion

The focus of this study involving use of ICT-enabled virtual laboratories has been two-fold. An assessment based on three attributes, i.e., metacognition, analogical reasoning and transfer of knowledge (MAT) was developed to measure conceptual understanding of theoretical and practical concepts amongst students. The MAT assessment for reflective thinking was found effective in evaluating a significant sample of participants in a blended teaching environment. The second objective of the study was to enhance the reflective learning in classrooms using computer-mediated tools. Providing access to virtual labs before physical labs showed enhanced learning via MAT assessments. Retentivity studies may need more testing across longer time-scales but our initial study reveals virtual labs augment skill training and concept understanding complementing traditional classroom.