Evaluating the Effectiveness of Video-Based Hybrid Learning in Architecture and Engineering: A Mixed-Methods Approach

Abstract

The use of hybrid classes, where face-to-face classroom and asynchronous activities are combined in an online environment, helps save time and provides students with resources to study and review the materials. Although numerous empirical studies have analyzed the effectiveness of this teaching approach in university degrees in different areas of knowledge, conclusive results regarding academic performance and technical skill acquisition have not yet been provided in architecture and building engineering degrees. These disciplines require specific investigation due to their high visual and practical complexity. In this context, this study aims to evaluate the effectiveness of a video-based hybrid model to improve student performance. Using a mixed-methods design, hybrid teaching was implemented in the construction and installation subject (N = 119) during the 2022/2023 academic year. The results obtained were then analyzed with a holistic approach, including students’ performance, behavior, feelings, and opinions. The results have shown how using the hybrid classroom led to an improvement in student performance rate compared to the previous academic year with traditional teaching methodologies. These findings suggest that hybrid models are a viable solution to reduce high failure rates in technical degrees.

1. Introduction

In recent decades, university institutions have made progress in their development of teaching methodologies and resources that facilitate and make the understanding of content in their degree programs more flexible (Gilbert, 2015). New techniques, resources, and tools have been implemented using technology and communication to interact within the university (Fiadotau et al., 2019). This aspect increased further with the COVID-19 lockdown (Mukhopadhyay et al., 2020), and, as a result of the government-imposed restrictions, teachers made the effort to adapt their way of teaching (Lorente & Pulido-Montes, 2022), thereby consolidating the use of hybrid classroom environments as a standard pedagogical alternative (Lei & Lei, 2019).

In this context, hybrid learning is defined as a design that allows face-to-face time to be allocated to more active approaches, redefining the way knowledge is understood (Louten & Daws, 2022). Student performance has also seen an upward trend with this type of approach. Here, the student can view online content at any time and can interpret the information received (Pisarenko, 2017). This usually generates satisfaction for the student (Valenti et al., 2019), although the perception and level of acceptance may vary between students (Louten & Daws, 2022). The hybrid approach in university degrees can also help them to understand content as a complement to course content. It can also be used to compensate for classes that the student cannot attend, to understand notes outside the classroom, or to review before an exam (Nagy, 2018), thereby adapting to the habits of Generation Z, characterized by a demand for immediacy and flexibility (Princes et al., 2024).

Within this hybrid framework, one of the resources used most for hybrid courses is video-based online lectures (Lancellotti et al., 2016). Using these as an educational tool is an aspect that has been gaining greater prominence in university degrees, although its use dates back to the 1930s (Sablić et al., 2021). Extensive technological developments in the last 50 years have led to videos being used as just another resource in everyday work. In this sense, nowadays it is easy to access a variety of audiovisual content through platforms such as YouTube on a smartphone or a computer (Giannakos, 2013). Hence, there is a wide range of options within the range of audiovisual materials, from Internet resources, films, or videos developed by teachers (Nicolaou, 2021), while the use of videos facilitates asynchronous teaching programs and can adapt to different needs and objectives (Mayer et al., 2020). This has already been proven with massive open online courses (MOOCs) with mainly video-based content (Birgili & Demir, 2022). As such, videos have a great versatility that can help students understand content, giving them greater autonomy (Noetel et al., 2021), and student performance with online lectures can be just as effective as with traditional teaching models (Kinash et al., 2015).

These advantages also extend to the teachers themselves. A high-quality hybrid course design allows them to control the information the student receives and eliminate vague information that is not relevant. Although this may involve an additional workload for the teacher, the results can be satisfactory due to a better understanding of the knowledge by the student (Noetel et al., 2021). Likewise, the hybrid course can favor not just better comprehension of content but also help eliminate communication barriers between teacher and student (Nicolaou, 2021), for example, facilitating communication for people with hearing problems. In this regard, Pappas et al. (2018) confirmed that deaf students who took part in e-learning sessions understood content better than in face-to-face teaching. Communication can also be improved for students who have some other cognitive disability (Tahat et al., 2022). Finally, using videos can allow teachers to adopt other dynamics in the classroom that they would not have time for under normal circumstances, such as gamification (Goedert & Rokooei, 2016). However, there are also drawbacks, such as the time required to design courses and the teacher’s workload compared to face-to-face courses (Kegley et al., 2016).

It is essential to design videos that are attractive and instructive. One of the aspects that generates the greatest discrepancy in the state-of-the-art is the appropriate length for videos, as the students’ attention span in front of the screen can be short (Sturman et al., 2018). In this sense, research has proposed different maximum lengths, such as 6, 10, or 20 min (Louten & Daws, 2022).

Despite these general benefits, a significant gap exists in the literature regarding its application in technical disciplines. While the use of hybrid classrooms has demonstrated satisfactory results in areas such as medicine (Mao et al., 2022), languages (Yang, 2021), science experiments (Gorghiu et al., 2010), basic physics (Halim et al., 2021), teacher training programs (Wiens et al., 2013), ethics (Todd et al., 2017), finance (Tiahrt & Porter, 2016), epidemiology (Shiau et al., 2018), as well as biology and communication (Louten & Daws, 2022). However, research is inconclusive about the effectiveness of using this teaching approach in all areas of knowledge, and there is no consensus on the preference of students regarding this approach (Masland & Gizdarska, 2018). This can lead to limitations of the approach depending on the area of knowledge. Hence, there is a need to evaluate the effectiveness of its implementation in all areas of knowledge (Pattier, 2021).

In university building engineering and architecture degrees, there is a lack of studies that evaluate the effectiveness of using hybrid classroom strategies. These degree programs are characterized by a wide variety of different subjects that usually have a lower student performance compared to other areas of knowledge (Ministry of Universities (Spain), 2022).

In these areas, passive participation through videos does not guarantee quality learning of content (Seo et al., 2021). Likewise, shortcomings in students’ organizational skills can also limit the effectiveness of videos as a teaching tool (Broadbent, 2017).

To address this specific gap, this study adopts a holistic framework grounded in four key dimensions:

(i).

Accessibility: Assesses technical barriers, given that difficulties in this area can compromise the effectiveness of the approach (El Mansour & Mupinga, 2007; Peceño-Capilla et al., 2022).

(ii).

Knowledge: Analyzes whether video instruction enhances the comprehension of complex concepts, thereby mitigating information overload (Merkt et al., 2011).

(iii).

Perspective: Measures student interest in the methodology, drawing upon previous studies from other fields such as chemistry (Jihad et al., 2018).

(iv).

Feeling: Evaluates student emotional satisfaction beyond examination performance (Bekteshi et al., 2023).

Consequently, the main objective of this study is to evaluate the effectiveness of implementing hybrid approaches in Architecture and Building Engineering degrees. In contrast to previous studies that address performance in isolation, this research analyzes performance, behavior, feelings, and opinions to determine whether the approach is reproducible and scalable across engineering sub-disciplines. Finally, this study also addresses the length of online lectures based on the discrepancies there are in previous studies of other areas of knowledge.

2. Materials and Methods

2.1. Case Study and Research Design

The study adopts a sequential explanatory mixed-methods design. The implementation of hybrid teaching was done in construction and installation subjects during the 2022/2023 academic year at the University of Granada. It was applied in 2 subjects of the Bachelor’s Degree in Architecture and 1 subject of the Bachelor’s Degree in Building Engineering (

Table 1. Subjects analyzed and students who took part in the study.

Subject	Bachelor’s Degree	Course	Semester	Type	Number of Students
Installations	Building Engineering	2	2	Compulsory	21
Installations	Architecture	3	1	Compulsory	65
Construction	Architecture	3	2	Compulsory	33

). A total of 119 students took part in the study, and participation was voluntary. The choice of these subjects was given the high dropout and failure rate that construction and installation areas usually have. The logic for selecting these specific cases lies in their historical academic metrics: they represent a “bottleneck” in the curriculum, characterized by high failure rates (approx. 40%) and dropout rates (approx. 20%) (Ministry of Universities (Spain), 2022; Rubio-Bellido et al., 2021). The aspects behind this situation are diverse, mainly highlighting the difficulty of the subject content and the need to adopt new teaching approaches (Bienvenido Huertas et al., 2023). Hence, these are subjects where it is necessary to implement new teaching dynamics that improve the learning of concepts. This negative trend was detected in the subjects chosen for the study and was seen both in performance in the subjects’ final exams and in the subsequent application of the knowledge acquired in other subjects. The problems were associated with the understanding of basic notions that are fundamental for the professional work of architects and building engineers.

2.2. Pedagogical Intervention: The Video-Based Hybrid Model

A large part of the problem detected was associated with the teaching model of the 3 subjects. To address this issue, a hybrid learning framework was implemented where asynchronous video-based instruction served as the primary intervention mechanism.

Previously, the model was based entirely on a lecture format (4 weekly hours of theoretical classes and problem-solving). These contents have to be taught to correctly understand the subject, but they do not allow adopting other, more practical solutions where the student has a more active role. For this reason, for the 2022/2023 academic year, it was decided to adopt a hybrid class approach in all 3 subjects. In the new hybrid model, part of the face-to-face teaching time was replaced by self-directed online learning (Lei & Lei, 2019). This shift allowed the face-to-face sessions to be redesigned in a hands-on way, avoiding a teaching approach based exclusively on lectures.

The approach consisted of generating explanatory videos about the calculation processes in the subjects. The videos for the construction subject consisted of explanations to calculate reinforced concrete structures, and in the case of installations, thermal installation calculation videos. In total, 35 videos were recorded for the construction subject and 76 videos for the installations subject. These videos were between 1 and 12 min in length (with a median duration of 7.7 min) (Figure 1).

Also, at the end of each video, the students had to answer a short questionnaire of 1 or 2 questions. By explaining the calculations through videos, half of the classroom time was available for practical sessions where the students had a leading role. Students had to watch the videos before going to the next class the following week, and the face-to-face sessions were designed in a hands-on way, considering the content of the online component of each subject. This made it possible to avoid a teaching approach based exclusively on lectures.

The videos were recorded and managed using the Kaltura application. Kaltura is an application that allows both video recording and editing, as well as user management and interaction. Its use as a teaching resource in other university areas has been shown in several studies (Paris et al., 2022), highlighting the ease of managing the videos. Different formats were used for the videos, depending on the type of content explained. When it was necessary to explain theoretical content for calculations, a lecture capture format was used with a PowerPoint presentation. This allowed for improving the learning of concepts and reducing the cognitive load compared to other approaches, such as voice-over (Chen & Wu, 2015). For the calculations, videos were recorded using a Wacom One 13” digital whiteboard. In all the videos, the professor explained the video by recording their face in the bottom corner of the video. The online videos of each course were hosted on the PRADO learning management system of the University of Granada.

2.3. Data Collection Instruments

To evaluate the intervention, two main data sources were used. The selection of variables was grounded in the study’s conceptual framework regarding student engagement and demographic factors in technical education.

2.3.1. Learning Analytics (Quantitative)

Student behavior when viewing videos was monitored throughout the course using Kaltura analytics. Five key variables were tracked for the 111 videos:

• Impressions (Count loads). A count of the number of times a student loads the video’s page (they do not have to play the video). Through this indicator, the aim is to evaluate the reach that the video has and the potential number of plays it may have. To evaluate the latter, the results had to be compared with the count plays.
• Engagement (Count plays). A count of the number of times a student plays the video (when they click on the option to play the video).
• Retention (Average completion rate %). Average completion percentage of the video. This value is also evaluated partially, at 25, 50, and 75% of the video viewing time. With this, it can be seen whether the videos were watched completely by the students or only in part.
• Reach (Percentage of unique known users %). The percentage of the number of students over the total who viewed the video. In the number of students, the same student cannot appear two or more times. Through this variable, it can be controlled whether the videos were viewed by most of the students in the cohort.
• Access modality (Device used). The number of times the video was played on a computer or mobile device. This allows evaluating the need to design videos to be watched on one device or another.

2.3.2. Student Satisfaction Survey (Qualitative/Quantitative)

To evaluate student perceptions, a structured questionnaire was designed based on a validated format used in the scientific literature (Divjak et al., 2022). The authors of the study, based on analysis of scientific literature, designed a 20-question survey intended to evaluate the students’ opinions in 4 different dimensions (

Table 2. Form given to students grouping questions by dimensions.

Dimension	Question
D1. Accessibility	Q01	The image quality of the video is adequate.
	Q02	The audio quality of the videos is adequate.
	Q03	The length of the videos is adequate.
	Q04	Using videos in the course material gives me more control and flexibility when studying (i.e., I can watch the videos when I prefer).
	Q05	I have the technical tools needed (computer, mobile phone, internet access) to watch videos.
D2. Knowledge	Q06	By watching videos, I gain more knowledge.
	Q07	I learn better with videos compared to the subject’s PDF documents.
	Q08	Watching videos in lessons allows me to better learn the concepts taught in class.
	Q09	The use of videos in lessons significantly contributed to acquiring relevant knowledge in the specialized topic of the lesson.
D3. Perspective	Q10	In the future, I will continue to use the videos provided on the subject.
	Q11	In the future, I may look for other short videos from different sources related to the course topic to gain more knowledge.
	Q12	In the future, I may look for longer online video conferences on the subject topic (MOOC, YouTube, TEDx)
D4. Feeling	Q13	The videos facilitate reflection, analysis, and critical thinking in the subject area.
	Q14	The videos are useful during the lessons and meet my learning needs.
	Q15	Videos save me time compared to studying PDF material.
	Q16	Using videos in lessons or when studying material is useful in the learning process.
	Q17	Using videos in lessons or when studying material increases my productivity (I learn more).
	Q18	I am satisfied with the learning process using videos for the subject.
	Q19	Watching videos helps me to achieve the subject’s objectives faster.
	Q20	I would recommend other students to take subjects when online videos are used to improve learning.

):

(i).

D1—Accessibility, to evaluate the accessibility of videos (5 questions). This dimension was suggested since, as reflected upon by El Mansour and Mupinga (2007) and Peceño-Capilla et al. (2022), technical difficulties in watching videos could decrease the effectiveness of the approach.

(ii).

D2—Knowledge, to evaluate the perception of the level of learning with the videos and the focus of the subject (4 questions). This aspect is considered key because even if students have an online resource that they can view as often as necessary, information overload can be counterproductive (Merkt et al., 2011).

(iii).

D3—Perspective, to evaluate the interest in using this methodology, both for the subject under analysis and in others (3 questions). This dimension is proposed following the approach of online content studies of other areas of knowledge, such as chemistry (Jihad et al., 2018).

(iv).

D4—Feeling, to evaluate the students’ feelings with the focus adopted (8 questions). This allowed evaluating whether the feeling and assessment of the students was positive, beyond performance in exams (Bekteshi et al., 2023).

Each of these questions had answers based on a Likert scale (from 1 ‘strongly disagree’ to 5 ‘strongly agree’). The students were also asked about their age, gender, and the number of times they had enrolled in the subject. The logic for including these specific demographic variables is as follows:

• Gender: This variable was included to analyze potential gender gaps in technical engagement, given the traditional underrepresentation of women in STEM fields.
• Age: This variable was necessary to control the maturity of the student and its correlation with autonomous learning capacity.
• Number of enrollments: This variable was used to distinguish between novice students and those repeating the course, as previous academic failure may influence the acceptance of new teaching models.

Finally, a question on other comments and opinions was available so that students could make additional assessments.

2.4. Sample Description and Participant

The study employed non-probabilistic convenience sampling, inviting all enrolled students to participate voluntarily. A total of 92 responses were obtained. Figure 2 shows the histograms for age, gender, number of enrollments, and subject of the analyzed cohort. Regarding the subjects, to simplify the analysis, all the installation subjects were grouped since the videos, and the focus of both subjects is the same (even if they are from different degree programs). As can be seen, the number of female students (n = 51) was higher than that of male students (n = 39), with a low number of non-binary students (n = 5). The age range was from 18 to 31, although most students were in the 18–23 age range (n = 78). Most students were studying the subject for the first time (81 students). Finally, 30 of the surveys are from the construction subject (90.9% of the students) and 62 from the installations subject (72.1%).

2.5. Data Analysis

The students’ assessments of the approach were analyzed both qualitatively and quantitatively. In this analysis, both the answers to each of the questions of the 4 dimensions and other data, such as gender or age, were used. For the quantitative analysis, the association analysis between questions was made. The association analysis was based on the evaluation of the Chi-Square (Equation (1)) and the correlation between variables. The corrected contingency coefficient for polytomic variables with the same number of categories (5 × 5) (Equation (2)) and the V’ Cramer for polytomic variables with a different number of categories (Equation (3)) were used for the correlation. The association between responses was made considering a significance value of 0.05. Data analysis was performed using the R programming language.

$${\mathsf{\chi }}^2=\sum _{i,j}{\displaystyle \frac{{\left(n_{ij}-\frac{n_i{n}_j}{n}\right)}^2}{\frac{n_i{n}_j}{n}}}$$

(1)

$$CC=\sqrt{\left({\displaystyle \frac{{\mathsf{\chi }}^2}{{\mathsf{\chi }}^2+N}}\right)\left({\displaystyle \frac{k-1}{k}}\right)}$$

(2)

$$V=\sqrt{\left({\displaystyle \frac{\frac{{\mathsf{\chi }}^2}{N}}{\mathrm{m}\mathrm{i}\mathrm{n}(r-1,c-1)}}\right)}$$

(3)

where $n_i$ is the observation with an A value, $n_j$ is the observation with a B value, ${\mathsf{\chi }}^2$ is the Chi-squared coefficient; $N$ is the number of observations; $k$ is the minimum number of rows and columns of the polytomic variables matrix; $r$ is the number of rows; and $c$ is the number of columns.

3. Results and Discussion

The analysis of the results was divided into three phases: (i) student behavior and performance; (ii) student perceptions and survey valuations (qualitative and quantitative analysis); and (iii) the influence of video length and additional design considerations. Each of these results will be analyzed in detail in the following subsections.

3.1. Phase 1: Student Behavior and Performance

First of all, an analysis of student behavior in the subjects under analysis was carried out. As indicated in Section 2, different variables of online material viewing behavior were monitored in the study. Figure 3 shows the histograms of the distributions obtained in these variables. The data on construction and installations were grouped to distinguish by subject area. As can be seen, the number of loads was variable depending on the video and the subject, with the construction videos being the ones that accumulated a greater number of loads. However, this is not the case with the number of views, where installation videos saw a higher number. In any case, the average completion rate in both areas was quite satisfactory, with a high percentage of videos watched completely: (i) 32 of the 35 construction videos were viewed for between 75 and 100% of their duration; and (ii) 54 of the 76 installation videos were viewed for between 75 and 100%. Finally, the percentage of unique users was high in most of the videos, and in the vast majority of the cases, this percentage was higher than 75% of the students participating in each subject. Thus, in general, the results obtained through this first descriptive analysis were satisfactory, as it was possible to verify that the number of playbacks and complete views of the videos was high.

However, this did not allow for assessing the complete effectiveness of the hybrid class model design proposed in the study. To do this, it was necessary to evaluate the relationships detected between the analysis variables and to look more closely at the behavior of the plays. This was done by evaluating the relationship between the number of loads and the number of plays. This aspect was considered essential as loading the video is the first step to play it and watch it completely. A correct design of the subject contents on the virtual teaching platform should ensure that the number of loads is not high and that the student is clear about what content they want to view. Figure 4 shows the point clouds obtained for the two subjects. As can be seen, the two subjects showed increasing trends: as the number of loads increased, so did the number of plays. This reflects that the students viewed the video they loaded most of the time. In addition, a similarity was detected in the behavior of the students in the two subjects: the correlation coefficient was 0.98 in installations and 0.74 in construction.

In any case, at the average completion rate level, it was possible to verify how a greater number of plays meant a decrease in the completion rate (Figure 5). This downward trend was detected in the two subjects, although with different correlation values: −0.59 in construction and −0.07 in installations. Thus, in the case of construction, the trend is clearer, while in installations, the behavior of the students was more varied. This may be due to the increased number of videos in Installations. Therefore, it is to be assumed that many students will make a first complete view of each video and, subsequently, the second play will be a review of certain aspects. In this sense, Figure 6 shows the day-to-day behavior of students in viewing the Installation videos. As can be seen, the students were watching the videos throughout the semester, with views every day. However, in the days approaching the subject’s final exam, the number of views increased significantly. This increase in views in many of the videos was related to lower play percentages. In any case, Figure 5 also shows how the greater number of plays has a clear correlation with the number of minutes watched: 0.69 in construction and 0.76 in installations. Thus, as the number of plays increased, the number of minutes watched also increased. This meant that the students viewed a significant percentage of the videos.

Finally, this first phase of analysis was completed by comparing the pass percentage of the 3 subjects in the course that applied the hybrid classroom approach (2022/2023) with the immediately preceding course (2021/2022).

Table 3. Percentage of pass rates of the 2021/2022 cohort (without hybrid teaching) and the 2022/2023 cohort (with hybrid teaching).

Subject	Bachelor’s Degree	Pass Percentage (2021/2022)	Pass Percentage (2022/2023)
Construction	Architecture	70.3%	81.8%
Installations	Architecture	44.4%	69.2%
Installations	Building Engineering	15.4%	90.5%

shows the pass rate percentage in the 3 subjects under analysis. As can be seen, the use of the hybrid classroom approach led to an increase in the pass percentage of all subjects, with the installation subject of the Bachelor’s Degree in Building Engineering being especially significant (rising from 15.4 to 90.5%). These results suggest that the hybrid classroom approach allowed students to have a more detailed learning process of content, by having virtual materials that could be viewed as many times they need and with complete freedom. However, although there is a strong positive association between the introduction of the hybrid teaching model and the improved pass percentages, a direct cause-and-effect relationship cannot be definitively isolated. The improvement in academic performance should be interpreted with caution, as mediating factors beyond the teaching method may have also contributed to these results. Variables such as intrinsic differences between student groups, variations in baseline motivation, or changes in the general academic workload throughout the semester could have an influence. Nevertheless, the consistent upward trend in performance, coupled with the students’ active viewing behavior, strongly supports the premise that this video-based hybrid approach acted as a primary facilitator for better knowledge acquisition in university-level architecture and building engineering education.

3.2. Phase 2: Student Perceptions and Survey Valuations (Qualitative and Quantitative Analysis)

Once the behavior of the students was evaluated, their assessment was analyzed through the survey they took. In this survey, four dimensions were evaluated as indicated in Section 2. The histogram with the answers given by the students is included in Figure 7. Histogram with the students’ answers, with responses on a Likert scale, with A1 being “Strongly disagree”, and A5 being “Strongly agree”.

3.2.1. Dimension 1: Accessibility (D1)

Regarding the D1 dimension (accessibility), the results showed the following distribution: 1.1% for A1, 2.4% for A2, 7.4% for A3, 23.3% for A4, and 65.9% for A5. Thus, the majority percentage was “Agree” and “Strongly agree”. This dimension covers questions related to the image and audio quality of the videos, their length, and the availability of time and means to view the contents. These contents were recorded using a lecture capture format with a PowerPoint presentation explaining theoretical contents and solving calculations using a Wacom One digital whiteboard. In both cases, Kaltura was used to record and edit the videos. Thus, this setup is considered appropriate by students, and it would not be necessary to make modifications in a more extensive application of the hybrid classroom model. In addition, one of the main concerns when adopting the approach is the technological gap that the hybrid approach could have. The results have shown that the students’ high level of digitalization has not posed problems for adopting this approach in the subjects analyzed.

3.2.2. Dimension 2: Knowledge (D2)

Regarding the knowledge dimension (D2), a very similar distribution to that of accessibility was obtained: 1.1% for A1, 2.7% for A2, 8.2% for A3, 25% for A4, and 62.2% for A5. Therefore, the use of the hybrid classroom approach was valued positively in content learning by the students, leading them to value that they learned more with the videos than reading a PDF syllabus, and that the virtual sessions allowed them to learn the concepts presented in class better.

3.2.3. Dimension 3: Perspective (D3)

In the case of D3 (Perspective), slight variations were detected in the percentages of each Likert scale, with an increase in the neutral rating percentage (17.4%). To understand this variation, it is advisable to make a detailed study of the 3 questions included within D3 (Q10, Q11, and Q12). Among these questions, the one that had a clear variation compared to the rest was Q12, a question on the future interest in viewing TEDx or MOOC online videoconferences related to the subject. In this case, it was seen that “Agree” and “Strongly agree” scored 55.4%, varying compared to the totals seen in Q10 and Q11 (76.1%). Thus, the students positively valued the future use of videos related to the course, but they were not so willing to look for online lectures. This may be because they associate online videoconferences with a long video that may be difficult to watch, while Q10 and Q11 are more associated with viewing videos with the format used in the course.

3.2.4. Dimension 4: Feeling (D4)

Finally, D4 (feeling) showed a high percentage of responses in A4 and A5 (83.7% of the total responses). Therefore, students positively value using the hybrid classroom approach to increase their productivity and to have more critical thinking with the content, among other aspects. They also recommend its use in future courses. Hence, from this qualitative analysis, the students’ positive assessment of using the hybrid classroom model in the Bachelor’s Degree in Architecture and Bachelor’s Degree in Building Engineering is confirmed. This aspect, together with the behavior and performance analysis of Section 3.1, reflects the great potential of using this type of approach.

3.2.5. Dependence Analysis of Evaluated Variables

To complete the analysis of the students’ assessments, it was decided to initially perform the dependence analysis of the analyzed variables. This was done to evaluate whether there are associations between the responses of the different dimensions, looking to detect behavioral patterns of the assessments. To do this, the contingency coefficient between the polytomic variables was determined. The Chi-square values are shown in

Table 4. Chi-square values between the questions on the student assessment form. Chi-square values are shown without decimals.

Question	Chi-Square
	Q02	Q03	Q04	Q05	Q06	Q07	Q08	Q09	Q10	Q11	Q12	Q13	Q14	Q15	Q16	Q17	Q18	Q19	Q20
Q01	144	131	154	152	85	169	140	100	67	52	41	150	70	51	132	113	116	125	120
Q02		116	128	142	74	109	123	80	60	40	26	100	54	51	139	107	120	106	104
Q03			132	113	86	90	120	71	73	52	31	97	72	46	118	109	121	135	128
Q04				169	119	147	174	134	100	83	39	103	71	95	164	142	150	179	122
Q05					94	121	147	107	75	58	35	91	87	60	147	141	127	128	120
Q06						127	138	115	85	66	26	88	194	87	121	131	127	123	104
Q07							115	91	58	46	31	128	114	66	96	92	86	90	77
Q08								123	98	65	49	116	118	99	207	186	189	156	139
Q09									84	98	24	146	146	113	155	140	141	148	82
Q10										64	60	64	82	65	132	114	115	104	125
Q11											36	91	64	50	61	58	53	60	66
Q12												34	34	46	52	65	45	30	54
Q13													142	79	104	100	93	88	92
Q14														108	139	132	147	82	95
Q15															98	101	131	87	59
Q16																197	214	141	158
Q17																	225	172	157
Q18																		153	150
Q19																			130

, and the contingency coefficients are shown in

Table 5. Adjusted contingency coefficient between the questions on the student assessment form.

Question	Adjusted Contingency Coefficient
	Q02	Q03	Q04	Q05	Q06	Q07	Q08	Q09	Q10	Q11	Q12	Q13	Q14	Q15	Q16	Q17	Q18	Q19	Q20
Q01	0.87	0.86	0.88	0.88	0.77	0.90	0.87	0.81	0.72	0.67	0.62	0.88	0.73	0.66	0.86	0.83	0.83	0.85	0.84
Q02		0.83	0.85	0.87	0.75	0.82	0.84	0.76	0.70	0.62	0.53	0.81	0.68	0.66	0.87	0.82	0.84	0.82	0.81
Q03			0.86	0.83	0.78	0.78	0.84	0.74	0.74	0.67	0.56	0.80	0.74	0.64	0.84	0.82	0.84	0.86	0.85
Q04				0.90	0.84	0.88	0.90	0.86	0.81	0.77	0.61	0.81	0.74	0.80	0.89	0.87	0.88	0.91	0.84
Q05					0.79	0.84	0.88	0.82	0.75	0.69	0.58	0.79	0.78	0.70	0.88	0.87	0.85	0.85	0.84
Q06						0.85	0.87	0.83	0.77	0.72	0.52	0.78	0.92	0.78	0.84	0.86	0.85	0.84	0.81
Q07							0.83	0.79	0.69	0.64	0.56	0.85	0.83	0.72	0.80	0.79	0.78	0.79	0.75
Q08								0.85	0.80	0.72	0.65	0.83	0.84	0.80	0.93	0.91	0.92	0.89	0.87
Q09									0.77	0.80	0.51	0.87	0.88	0.83	0.89	0.87	0.87	0.88	0.77
Q10										0.71	0.70	0.71	0.77	0.72	0.86	0.83	0.83	0.81	0.85
Q11											0.59	0.79	0.71	0.66	0.70	0.69	0.67	0.70	0.72
Q12												0.58	0.58	0.64	0.67	0.72	0.64	0.55	0.68
Q13													0.87	0.76	0.81	0.81	0.79	0.78	0.79
Q14														0.82	0.87	0.86	0.88	0.77	0.80
Q15															0.80	0.81	0.86	0.78	0.70
Q16																0.92	0.93	0.87	0.89
Q17																	0.94	0.90	0.89
Q18																		0.88	0.88
Q19																			0.85

. As can be seen, the Chi-square values were in the vast majority of cases higher than the critical value of 26.3 (obtained for p = 0.05 and 16 degrees of freedom). This assumes an association between the variables when the null hypothesis is rejected. The contingency coefficients also obtained high values, with a maximum value of 0.91. This high association is mainly due to the high concentration of “Agree” and “Strongly agree” responses. In any case, dissociations were also obtained between the responses. All the cases were with Q12. This question already stood out in the histogram as its response trend was different from the rest of the questions of D3. The difference is extensible to the other dimensions. Thus, it was detected that the Q12 responses were independent of Q2 (D1—Accessibility) and Q6 and Q9 (D2—Knowledge). In addition, it was possible to verify how the Chi-square with the rest of the questions was low and close to the critical value of 26.3. Therefore, it was detected that there is a discrepancy among students about watching long video conferences related to the subject in the future. This aspect differs from the viewing of short videos, where there is a clear association with the other dimensions. Regarding the rest of the questions/dimensions, the high association of the answers could be verified. Thus, positive assessments were given in all 4 dimensions by the vast majority of respondents.

Once the relationship between the four dimensions was analyzed, it was decided to extend the association analysis with the additional variables that were available on the form dataset (age, gender, number of enrollments, and area of knowledge (construction/installations)). This was done to know whether there are associations in each variable with some of the dimensions/questions (

Table 6. Chi-square values and adjusted contingency between the survey variables and age, gender, number of repetitions, and type of subject (construction or installations).

Question	Age		Gender		Number of Enrollments		Subject
	Chi-Square	C	Chi-Square	C	Chi-Square	C	Chi-Square	C
Q01	25.93	0.52	7.22	0.40	2.09	0.21	13.93	0.55
Q02	25.38	0.52	2.12	0.21	6.02	0.36	9.14	0.45
Q03	15.66	0.43	10.99	0.49	2.29	0.22	3.65	0.28
Q04	21.83	0.49	5.40	0.34	11.07	0.49	5.46	0.34
Q05	99.03	0.80	4.87	0.33	15.43	0.50	8.31	0.43
Q06	32.51	0.57	5.44	0.34	8.89	0.44	1.59	0.19
Q07	25.24	0.52	4.44	0.31	9.44	0.45	5.58	0.35
Q08	18.78	0.46	3.66	0.28	3.11	0.26	3.39	0.27
Q09	24.33	0.51	3.25	0.27	3.99	0.29	3.23	0.26
Q10	14.37	0.41	9.95	0.46	7.36	0.40	1.66	0.19
Q11	33.21	0.58	4.21	0.30	11.69	0.50	2.62	0.24
Q12	10.83	0.36	10.81	0.48	6.39	0.37	1.91	0.20
Q13	23.17	0.50	3.43	0.27	9.21	0.45	4.64	0.32
Q14	38.04	0.60	5.78	0.35	3.99	0.29	2.36	0.23
Q15	20.31	0.48	5.02	0.33	10.99	0.49	1.43	0.18
Q16	25.40	0.52	4.29	0.31	6.37	0.37	3.32	0.27
Q17	21.63	0.49	5.45	0.34	6.79	0.38	6.65	0.38
Q18	26.95	0.53	3.32	0.27	11.41	0.50	5.81	0.36
Q19	16.67	0.44	5.68	0.35	2.47	0.23	3.16	0.26
Q20	20.29	0.48	9.15	0.45	2.06	0.21	4.02	0.30

). With regard to age, associations with questions from the 4 dimensions were obtained. Thus, the younger students tended to concentrate their answers on “Agree” and “Strongly agree” in some questions. This may imply a predisposition to adopt this type of approach among the youngest students of architecture and building engineering. With regard to gender, the number of times the course had been taken, and the area of knowledge, no associations were detected between the variables and the questions of each dimension. Thus, the Chi-square values were lower than the limit value (15.507 for gender and enrollments, and 9.49 for area of knowledge). Therefore, the assessments provided by the students were independent of gender, the number of repetitions of the subject, and the area of knowledge. The only aspect where a certain association was detected was with age. This aspect may be interesting, since it would mean that, in future editions, with an increasingly digitized student body, the reception of this type of approach will be even more positive.

3.3. Phase 3: The Influence of Video Length and Additional Design Considerations

The analysis of the results concluded with the assessment of the influence of the video length and the type of device used by users to view the videos. For the video length, this analysis was done to respond to the discrepancies there are in the scientific literature about the length of videos in hybrid classroom approaches. Figure 8 shows the relationships between video length and the monitored variables. Although there is a certain similar trend by variable in the two areas, the correlations were low. Thus, it was detected that the number of plays and the percentage of unique users had an upward trend as the length of the videos increased, although the correlations were between 0.19 and 0.288. Similarly, a downward trend was detected between the average completion rate and video length (the longer the videos, the lower the average completion rate). However, the correlation coefficients were again low in both areas: −0.53 for construction and −0.16 for installations.

Although there is a certain upward/downward trend depending on the variable, the correlations were low, and a clear association with video length cannot be highlighted. Thus, long videos obtained high play numbers and average completion rates. This lack of correlation may be due to the premises considered for the length of the videos. In the design of these videos, it was intended that their length should not be long, with a maximum limit of 12 min (Figure 1). Hence, the length set for the videos is considered appropriate since it is not an influential factor on student behavior.

Finally, the type of device used by the students was evaluated. With this aspect, it was intended to evaluate whether videos should be designed depending on the device that the students use. In this sense, the videos were originally designed for computer viewing, although in view tests made by the teachers, it was found that they could be viewed correctly. The student behavior is reflected in Figure 9. It can be seen how the boxplots show a clear tendency to use the computer over mobile phones. In this sense, the percentage of time spent playing on the computer was more than 90% in most videos. Therefore, it is not necessary to focus on the design of videos for mobile devices.

4. Conclusions

This study analyzed the potential of using hybrid classes in university architecture and building engineering degrees.

The results have shown how the use of the hybrid classroom approach led to an improvement in student performance rate compared to the results obtained in the previous academic year. Likewise, it was detected that user behavior obtained high play and average completion rate values. On these days, students made partial views, mainly to review for the exam. Thus, the student behavior had a positive reception depending on the number of plays and the completion rate.

Regarding the students’ assessments, positive evaluations of the hybrid classroom approach were obtained in the four analyzed dimensions. Thus, the design of the material was accessible, helped reinforce knowledge, and showed a critical attitude toward the subject, among other aspects. Therefore, students do not find viewing long videos related to the subjects interesting and prefer a shorter video approach as used in the design of this study.

Regarding video length, it could be seen that this was not a determining factor in user behavior. The usefulness of this material to learn concepts and pass the subject has more weight than other factors, such as the video length. This is related to the lack of outstanding correlations between video length and the monitored variables. Therefore, the design of videos with a maximum duration of 12 min may be suitable for the design of hybrid courses.

To conclude, the results of this study show the potential of adopting hybrid dynamics in architecture and building engineering education. Through this approach, it is possible to avoid excessive use of lectures and allocate part of classroom time to other, more practical and active approaches. In addition, viewing online material as many times as the students deem necessary favors better learning of the theoretical content. This has made it possible to address the existing student performance problem of these degrees. Despite the positive results, some aspects should be addressed in future work. First of all, it would be a good idea to perform a more extensive hybrid course approach. In the current format, the course design has mainly been based on the design and monitoring of video questionnaires lasting between 1 and 12 min. The design of additional material, such as serious games, could improve the use of this type of approach. Secondly, this study has not considered the limitations of the hybrid classroom model with students from other countries. Since the entire study cohort was local students, it has not been possible to assess aspects such as the language barrier, which should be addressed in future work.