Digital technology in science classrooms

When the COVID-19 pandemic began, teachers in European countries (and not only them) had to quickly adapt their courses from face-to-face to online ones due to massive lockdowns. It was a great challenge for them but the transition to technologies in education was driven by need and affected all teachers non-selectively [11]. Because they had to react nearly on a day-to-day basis, teachers tended to focus on gaining the general technological skills specific for online instruction followed by content (re)transformation and paid less attention to the integration of technology with pedagogy and content [12]. This integration is important for effective teaching and, in general, similarly to other teaching settings, covers and combines cognitive, affective, behavioral, and social aspects through communication and collaboration to engage the learner [13]. The theoretical ground for educational technology and consideration of its interaction with other described components in the form of the concept of technological pedagogical content knowledge (TPACK), [14] was available before COVID-19. This framework has been used in many of the studies investigating teachers’ use of technology and is probably the most commonly-used theoretical framework in the context of both research on technology in schools and the professional development of science teachers [15]. Its weaknesses are fairly well known in the context of science and biology and are summarized, for example, by [16]. We assume that they can also be applied to other disciplines, among whose lessons, for example, Walan found no difference [17]. A very important aspect when studying technologies using the TPACK framework is to take into account the dynamic nature of the domain–many technologies (pencil, book) are used routinely, hence sensu Cox and Graham [18] we understand them as transparent technologies. The extension of the TPACK framework over the extension of Shulman’s work [19] on pedagogical content knowledge (PCK) lies only in emerging technologies, those that are being investigated or introduced into a learning environment. However, in our view, the distinction between transparent and emerging technology has shifted significantly since the establishment of the research field. Most of the technologies whose uses and characteristics researchers observe are difficult to consider as emerging technologies, be they email, online resources, or collaborative cloud space.

These technologies are a routine part of the administrative side of a teacher’s job, and to view them as emerging in teaching (let alone after the COVID-19 pandemic during which technology was a de facto core part of teaching) seems to us to be a freezing of discourse (e.g., the use of TPACK as a theoretical framework for analyzing the use of PhET [20] or adding verbs typical for the digital environment (like programming, animating, podcasting, tagging, testing, advanced searching) to Bloom’s taxonomy [21]). There has been a fundamental transformation whereby the use of technology in school is no longer "driven more by the appropriateness of technology rather than the demand of pedagogy and teaching of subject matter" [22], but attitudes toward technology are predictors of how technology is used, not whether it is used. Thus, we are in a situation where the use of technology is an everyday practice and separating research into situations with/without (modern) technology gradually loses its widespread importance. Even for the technologies that are closest to be called emerging, i.e., probably AR/VR, it can be documented that they have been used in school settings for more than a decade [22] and the crucial point ceases to be the technology itself. Therefore, what is important for today’s schools is not whether they use technology and which ones, but, in the words of [23] "convert artifacts [e.g. digital resources, that is, an entity or product of human creation, with embedded knowledge] into epistemic tools". The cited work provides a fundamental theoretical framework for the design of digital resources culminating in three guiding principles that relate to both the resource itself (guiding principle 1 –"A digital resource needs to have other aggregated epistemic artifact(s) to be used for educational purposes"–that includes some kind of "exploration guide", methodological hints for teachers), to the work of the teacher (guiding principle 3 –"A digital resource must be inserted in an orchestrated chain of artifacts, used in a setting of learning in the context of epistemic practices") and to both at the same time (guiding principle 2 –la "A digital resource becomes a tool if used effectively to solve a task/problem in a setting of learning context"). Other similar frameworks of thought, which in our opinion are much better suited to the needs of studying technology in learning than, e.g., the TPACK framework, have been developed based on the experience of pandemic capping. In our work, we have used a framework formulated by [24] focused on online arrangement. We consider this framework to be close in idea to our subject of interest, online science practicals. It describes the foundation components of online pedagogy, namely five pillars that grow out of the principles of learner-centeredness [25], constructivism [26], and situated learning [27]. According to [24], these five pillars include i) cultivating relationships understood as interactions that build relationships both within school (among students and between students and teachers) and outside school (within the family, between students, and nature, students and society); ii) active learning e.g., choice in the path to the learning goal, independent solving; iii) learner autonomy and supporting creativity where learners are seen as co-creators who seek information and build connections; iv) personalization which values, respects, and accommodates learner differences by providing learning materials in a variety of formats and modalities and allows independent exploration, time, and place flexibility; v) mastery learning, which is described as a customized curriculum based on assessment data that allows learners to see the progress they are making and gives them constructive and consistent feedback.

Science practicals

As in [28], we understand science practicals as an overarching term including any type of activity linked to science education in which students (individually or in groups) manipulate and/or observe real objects no matter if they use a “cookbook”, inquiry, or other approach. They can be held in laboratories (e.g., flame tests, microscopy), outdoors (e.g., plant identification), or in any other environment. Science practicals have had an important role in science education and have been shown to be a challenging method even in face-to-face settings, as they can easily turn out to be ineffective [29]. Their implementation is often considered to be more demanding even in face-to-face teaching: preparation, safety standards, organization, interpretation of unexpected results, and unpredictability. They are effective in making students manipulate objects according to a recipe [30, 31], typically offering a unique opportunity of working in a setting of smaller class sizes, higher instructor-to-student ratios, and the diverse learning possibilities laboratory classrooms offer. Practicals can be effective and develop conceptual understanding when it is “hands on” and “minds on”, and when teachers explicitly guide students to link these two essential components of practical work [32, 33]. Also, in addition to improving knowledge, practical work has been shown to improve attitudes toward the subject [34, 35], interestingly in all of the following settings: traditional (not using computers), computer-supported laboratory, and computer simulation [34].

The interest of researchers in science practicals settings led to the formulation of the design principles of student resources. These principles were developed regardless of forced online teaching by [36] and include the ones relevant for our study which we highlight: i) safety–training to work in safe settings, in the online environment understood as the possibility to safely study dangerous or inaccessible features/processes; ii) authenticity–being based on real word/research situations; iii) flexibility–provides a variety of topics and difficulty levels.

To map the emerging situation, [37] analyzed different types of adjustments made to laboratory curricula and activity types due to the COVID-19 pandemic (e.g., different types of experiments, simulation, pre-collected data analysis, planning, reviews, etc.) and the reported immediate effect on students. The lack of hands-on laboratory experience was reported to be counterproductive to some types of learning and participation, such as understanding procedures, analyzing nonideal data, evaluating the trustworthiness of the data, troubleshooting, making evidence-based conclusions, proposing experiments, or predicting results [37, 38]. However, some goals could be achieved remotely [37] with the recommended setting, for example, including peers in the learning formulated by [24, 36]. Specific recommendations for science practicals were introduced by [37–39] and are in agreement with the general rules. [40] point out that the fact that most students today can use their smartphones for data recording, taking pictures, and connecting to online apps such as virtual simulations helped to reach the science practicals goals.

Activity types in online science practicals

In addition to a description of the features of a digital resource and its potential to meet the needs of online learning, it is also useful to look at the form in which the resource is implemented in teaching, i.e., in what type of technological enhanced activity it is incorporated. For the typology of activities, we used [41] as a basis, selected those that are typical/characteristic for science practicals, and modified them slightly when confronted with more up-to-date resources. We especially considered that the digital resources used for face-to-face labs had to be altered (or even replaced by a new format) from the basis for distance teaching, and instructions began to emerge shortly after the first COVID-19 forced lockdown. This modification can perhaps be seen as a mere change of modality [42] or as a new type of activity given that it is a paradigmatically new teaching approach [41]. The difference between online and face-to-face teaching is not simply a change of communication channel, but a qualitatively different type of work. The role of the digital elements in the online setting can be different from face-to-face settings and take on new functional roles, where, for example, a quiz is not only an activation and feedback tool but also a quick way of presenting the results of measurements or an assumption of the outcome of the intended observation. Therefore, we are closer to viewing it with respect to the total givenness of the modality as a new type of activity, but consideration of this issue is beyond the scope of this paper.

We have identified these relatively specific activity types for online science practicals: I) Investigation, experiments, inquiry, e.g., in kitchen settings which can be combined with online tools, for biology possibly including individual nature visits, e.g., sample collections outdoors like making a (digital) herbarium, includes remote experiments [37, 41, 43]. II) Virtual reality, which provides realistic interactions with 3D computer generated learning environments, or digital lab environments which are entirely in a virtual interface (e.g., Labster, Beyond Labz). Virtual laboratories can provide a safe learning environment accessible online, helping to better understand the experiment as a whole. The recommended setting is to use virtual laboratories before the real ones as a preparation for face-to-face laboratory experiments, as they lower stress level and anxiety and improve learning [44–47]. And this is where the paradigm shift became apparent, where their role was seen as an adjunct to the face-to-face setting, but during forced online learning it became a separate form. During distant teaching, virtual laboratories were often substitutes for wet labs. III) Computer simulations [41] have been shown to improve conceptual understanding of chemistry [48] and physics [49]. The second study shows an example where students using computer simulations explicitly modeling electron flow understood the electricity concepts better and also could assemble the real circuit and describe how it worked better than students working in a real laboratory. IV) Data analysis and computational science (using computers to solve science problems) [41] demonstrate the current approach in all science disciplines–solving, for example, molecule structure, molecular genetics [50] or enzyme kinetics [51]. V) Video-based learning can be in many styles (“images” according to [41]), from simple pre-recorded video experiments to real-time lab recordings [52]. VI) knowledge revision, like answering questions, taking a test or a quiz [41] have been shown to significantly improve knowledge retention [53, 54]; VII) games including a story line and rewards when students have to reach a goal either individually or in a team [55]; VIII) modeling and construction when students learn by doing, e.g., construction a measurement tool with Arduino followed by use of the mobile app “Arduino Science Journal” [56] or building a model [41]; and IX) virtual field trip when the “visitors” do not go to the actual site but “visit” it through a virtual tour [41, 57, 58].

New digital resources are continuously emerging and derived from new activity types, worthy of the interest of the general teaching public. Forced online learning has accelerated the “coevolution” of digital resource developers and their users, clarifying the requirements for their functions and teachers’ demands for the inclusion of aspects (e.g., communication) that might have been neglected before. Therefore, we consider it important to be able to define their qualitative characteristics and to know which ones are decisive when teachers decide whether to adopt them in their future (face-to-face) practice. We start from the principles formulated by [24, 36] together with activities [37, 41] and combine them to collect data about digital resources used during online science practicals.

Aims

Our aim was to identify digital resources that were adopted by science teachers during lockdown and incorporated into online science practicals; and to propose an interpretation of their shared characteristics which made them useful to teachers when planning effective online science education.

Specifically, we wanted to find answers to the following research questions that describe the key features of digital resources used during online science practicals:

1. What were the most important characteristics of the digital resources that led teachers to use them?
2. What types of activities were enhanced by using technologies?
3. What characteristics of the resources make the difference in the appropriateness of the individual activity type?

Sample and sampling

Digital resources were collected in 5 EU countries: Czechia, Slovakia, Slovenia, France, and Spain from in-service science teachers using an online survey (GoogleForm). The teachers were contacted based on their contacts and possibly also previous cooperation, and asked to share only activities they knew, had used, and considered useful for online science teaching. As has been shown [59] even similar low stakes, e.g., via email provides relevant results. As we aimed at describing the digital resources which teachers used, we omitted the idea of asking for the resources they had tried and rejected. We decided so also because one can hardly assume that teachers, when searching for resources, will remember which resources they have not accepted, let alone attach relevant assessments to them retrospectively [60].

Instrument

Teachers provided the name of the resource, hyperlink, overall rating of the resource (scale 1–5, the more points, the better rating), information about modality (synchronous, asynchronous, both) and the subject (biology, chemistry, physics, geology, geography). The appropriate educational levels ISCED 1–3 were recorded together with information about availability (if it was free or needed payment).

The pedagogical characteristics of the resources were sampled in the following sections: 1) type of activity (experiment, video, knowledge revision, augmented reality, simulations, etc.) with classification based on literature, see Table 1; 2) pedagogical suitability; 3) interactions; and 4) specific added value which are described below.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Activity types used in the survey.

https://doi.org/10.1371/journal.pone.0283054.t001

We provide a short definition, and the literature from which we draw.

In the next section of the survey, teachers reported the pedagogical suitability of the resource as the level of their agreement with statements describing the ease of use (“The structure is clear”, “The visual is appealing”), suitability of the online tool for teaching science (“This resource helps to teach the subject”) and teaching ICT (“It helps to teach an important ICT skill’) and other skills (“It helps to teach something else important (except subject topic and ICT)”), and willingness to use the activity based on this resource as is or after modification during online teaching or standard teaching (“I would use this resource as it is, even in standard teaching mode”, “I would use this resource as inspiration and modify it for standard teaching mode”) on a 5-point Likert scale. Teachers rated the support for interactions (cultivating relationships, [24]) among students, students and teachers, students and society, students and families, students and nature, to assess the support in building relationships and community (using a 5-point Likert scale). Practicals are appreciated for specific added values as shown by the research; the extracted categories used for this study are summarized in Table 2. Except for the categories based on principles formulated mostly by [24, 36], we reported two more: providing worksheets and connecting different organizational levels. Providing worksheets to students is often part of science practicals [64], and ready-to-use worksheets are typically appreciated by teachers. Connecting different organizational levels is necessary to understand science and is described as an issue especially in biology [65, 66].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Resources added value overview.

https://doi.org/10.1371/journal.pone.0283054.t002

The authors also additionally recorded whether the resource was aimed at higher levels of cognitive processes of the revised Bloom taxonomy in factual and/or conceptual knowledge dimensions [21, 67] or at “remembering and reproducing”. The simultaneous assignment of the resource to the knowledge revision and Bloom’s “remembering and reproducing” categories was taken as a coding check.

Data cleaning

The data was cleaned in Excel. The activity types which were not mentioned or were mentioned only once (A_virtual trip, A_modelling) were not included in the analysis. If the question allowed multiple answers, the format was transformed to dummy variables. All the answers were translated into English. Inconsistent terminology used by teachers was unified.

Statistical analysis

R v.4.1.2 [68] was used for further analyses of data collected. For the description of pedagogical suitability, we proposed two items–I_17+I_18 (I_17 “I would use the resource as it is, even in standard teaching mode”, I_18 “I would use this activity as inspiration and modify it for standard teaching mode’) which we considered as complementary, i.e., negative correlation. We expected a split between resources that are ready-to-use and resources that need more or less modification to be accepted by teachers as suitable. We tested this assumption with the Spearman correlation coefficient.

Principal component analysis (PCA) computed on the correlation matrix was used to describe the variability of the pedagogical suitability component. Data were not transformed as all items describing pedagogical suitability were collected in the same manner as a 5-Point Likert scale [69]. The decision on principal component count to be analyzed was based on a scree plot of the explained variance of each of the principal components compared to the amount of variance each variable would contribute if all contributed the same amount.

Interaction support provided by the resource was analyzed using a divisive hierarchical clustering. For clustering, we used the cluster package [70] with the similarity coefficient metric [71], which is suitable for categorical data [72]; evaluation of the number of interpretable clusters was based on a scree plot of internal cluster variability and a silhouette index comparison [73].

A linear regression model to predict the overall score from individual subscales was fitted. We use a backward elimination to choose the final model.

To investigate the effect of subcategories from the added value group, we used the non-parametric Kruskal-Wallis test with Bonferroni correction on multiple testing.

All statistical tests were performed at the chosen significance level of 95%. Effect sizes were labeled following recommendations [74]. 95% Confidence Intervals (CIs) and p-values were computed using a Wald t-distribution approximation.

To analyze the relationship between features of the digital resource and activity type (in which the resource was used by the teachers), we used multi-label classification methods by package mlr using conditional random forests as a learner. The training set consisted of 50 randomly selected resources, while the remaining 39 served as the test set. All of the descriptive features that we had available were used as features, in addition to the overall teacher ratings. The metric of performance used was Hamming Loss, that is, a proportion of activity types that are predicted incorrectly, following the definition by [75].

Activity types

The teachers were asked to state an activity type for which they use the resource. Most frequently, it was A_revision, followed by A_simulation (e.g., population dynamics; input data, factors, etc., see results) (Table 3). More than one activity could be chosen and 3 was the most frequently used number of them in their answers. The highest number of activities stated for one resource was 6.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. List of activity types with the frequency among resources.

https://doi.org/10.1371/journal.pone.0283054.t003

In the situation where multiple answers were chosen at the same time, the combinations of A_revision + A_simulation or A_revision + A_investigation were the most common (N = 24); an overview of the combinations that occurred in more than 15 cases is given in Table 4.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. Most common pairs of activity types reported together for a single web resource.

https://doi.org/10.1371/journal.pone.0283054.t004

Pedagogical suitability

The highest sampled score of pedagogical suitability was 35 (out of 35 possible), and the lowest pedagogical suitability score was 18 with the median value 27. The three most highly valued resources were all targeting the PhET project (DR_25, DR_70, DR_71).

The two survey items–I_17+I_18 (I_17 “I would use the resource as it is, even in standard teaching mode”, I_18 “I would use this resource as inspiration and modify it for standard teaching mode”)–were designed to be complementary but turned out to be independent. The Spearman’s rank correlation rho between I_17 and I_18 is positive, statistically not significant, and very small (rho = 0.09, S = 0, p = 0.398).

The analysis of the primary components of pedagogical suitability showed the primary component PC1 explaining 37.8% of variability, the following axes 15.6% and 14.6% of variability, respectively, see Figs 1 and 2. Therefore, we interpret only the first component, graphically shown in Fig 2. PC1 is loaded with items I_17, I_19, I_20, I_21. That is "use as is", "teach subject", "structure clear", "important ICT skill", see Table 5 for PCA loadings and item wording.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Explained variance percentage of primary components of the DR pedagogical suitability items.

https://doi.org/10.1371/journal.pone.0283054.g001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Visualization of the first and second primary components of the DR pedagogical suitability items.

https://doi.org/10.1371/journal.pone.0283054.g002

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 5. PCA loadings in the first principal component of the DR pedagogical suitability items.

https://doi.org/10.1371/journal.pone.0283054.t005

Thus, the features forming PC1 are the main source of variability in the teachers’ evaluation of resources (as overall ranking) in a linear model. The model’s intercept, corresponding to PC1 = 0, is at 4.02 (95% CI [3.85, 4.19], t(87) = 47.54, p < .001). Within this model, the effect of PC1 is statistically significant and positive (Std. beta = 0.61, 95% CI [0.44, 0.78], t(87) = 7.15, p < .001).

The first primary component explains 38% of variance. The red horizontal line indicates the amount of variance each variable would contribute if all contributed the same amount (14.3%).

The length of the arrows corresponds to the loadings of each item, the colored ellipses highlight the distribution of the resources with respect to their overall teacher rating. The trend to increase the overall rating of the resources with respect to the positive direction of PC1 is clearly visible.

Interaction support

The highest sampled score of the interaction support was 25 (out of 25 possible), and reached by three resources (DR_42 Química en context, DR_70 PhET and DR_71 PhET), The lowest pedagogical suitability score was 5, the median value was 17.

The interaction support category formed two distinctive clusters. The chosen number of clusters is supported by the Silhouette index, showing that two clusters are most dissimilar in between. The Elbow plot does not indicate a clear cut-off and therefore we interpret two clusters (Fig 3). Group A was characterized by interactions that focused inside the school, student-teacher (I_24) and student-student (I_25) interactions which were supported by resources like DR_88 Mentimetr, DR_29 Kahoot, DR_85 Padlet with which activities are typically used synchronously. Group B did not have the school interaction and contained the student’s interaction with society (I_26), nature (I_27), and/or family (I_28). Examples can be resources DR_42 “Química en context”, or DR_17 “Climate facts” which collects data on climate and climate change provided by scientific institutions (e.g., NASA, Eurostat, national data about weather, and others) and compiles them into graphs and infographics for further use.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Interaction support of resources.

First row: Dendrogram of interaction support of resources. The left cluster (marked A) represents interactions in the school context; the right cluster (marked B) represents interactions in the out-of-school context. Second row: Analysis of the elbow scree plot and the silhouette index to select the number of interpreted clusters. The plot on the left represents the internal variability of clusters without sharp breaks. The right plot represents the Silhouette index, the greater the dissimilarity of the clusters, the more dissimilar. The highest dissimilarity occurs only at two clusters.

https://doi.org/10.1371/journal.pone.0283054.g003

The peer interaction (I_25) was the most frequently mentioned (57 resources, median = 4) followed by the student-teacher interaction (I_24, 47 resources, median = 4), student-society interaction (I_26, 41 resources, median = 3) together with the student-nature interaction (I_27, 40 resources, median = 3). The interaction mentioned least frequently was the student-family one (I_28, 17 resources, median = 3).

Added value

The categories of resource added value occurred with comparable frequency–I_35 (providing worksheets) was the least represented at 19 resources (21%), and I_29 (supporting independent solving) and I_32 (safety) were the most frequently assigned–these were listed for 45 resources (50.6%), with a median of 37.5.

More than a third, 32 resources (36%), had no added value assigned. All categories were assigned to 11 resources (12.4%). The remaining 46 resources (51.7%) had between one and seven categories assigned (Fig 4).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Number of added value categories listed together.

More than a third of the resources had no added value assigned; see the first bar; otherwise there was considerable variance in the number of added value categories listed together.

https://doi.org/10.1371/journal.pone.0283054.g004

Bloom’s taxonomy

Most (48, 53.7%) of the resources were rated as “remembering” and aimed at a low cognitive level.

Overall rating

As part of resource reporting, the teachers gave an overall rating, a “number of points” to each resource (the higher the better). The rating was mostly high (maximum 5 points out of 5, median 4 points), as the teachers were asked only for examples of good practice, the resources they wanted to recommend to others.

Relationship between overall rating of resource and individual subscales

We fitted a linear model to predict the overall rating of the resource. As explanatory variables, we used pedagogical suitability, interaction support, added value, and Bloom’s taxonomy, selecting the model by a backward selection. The accepted model formula is overall rating ∼ pedagogical suitability + interaction support. This model explains a statistically significant and substantial proportion of variance (R² = 0.38, F(2, 86) = 25.87, p < .001). Within this model:

• The effect of pedagogical suitability is statistically significant and positive (Std. beta = 0.44, 95% CI [0.22, 0.66], t(86) = 4.02, p < .001)
• The effect of interaction support is statistically significant and positive (Std. beta = 0.23, 95% CI [0.02, 0.45], t(86) = 2.14, p = 0.035)

Relationship between features of digital resource and activity type

The multi-label classification model reached the performance measured by HammingLoss = 0.225, that is, more than 75% of resources were assigned to the correct activity type. This is a good success rate for the model in a situation where we assume considerable variability in the data. The success rate of the single activity classification varied. Virtual activities and augmented reality were the easiest to predict, whereas knowledge revision or simulation was the most difficult, see Table 6. Among the entire set of available features, the most important features used by the classifier were (sorted descending) I_36 (connecting different organizational levels) = 31, I_37 = 37 (Bloom’s taxonomy), I_X35 = 30 (providing worksheets), I_30 = 25 (supporting creativity), I_25 = 10 (student-student interaction), I_24 = 10 (teacher-student interaction).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 6. Multilabel classification performance measure per activity type.

https://doi.org/10.1371/journal.pone.0283054.t006

Description of the most frequently mentioned resources

PhET (phet.colorado.edu/) offers free interactive simulations for science and math that can be used in a variety of settings, from demonstration experiments to individual exploration. Students test their hypotheses and answer questions by changing the setting of different variables and observing the results, and gain a deeper understanding of basic concepts. The simulations convey an understanding of very small, dimensional, or fast processes that cannot be easily observed directly. The topics are, for example, neuron function, gene expression, color vision, natural selection, energy forms and changes, pH scale, diffusion, molecules and their polarity or shapes, atomic interactions or balancing chemical equations, the solar system, electrical circuits, the force of gravity, density, vector addition, Coulomb’s law, curve fitting, fractions, graphing quadratics, projectile motion, and many more. Registered users have access to teaching materials. The design of the simulations has been tested and proven effective [76–79]. The resource was mentioned 16 times in our survey, typically gained an overall rating of 5 and a positive rating in pedagogical suitability. The evaluation of interaction support varied among the individual simulations evaluated as different teachers described different simulations or the resource as a whole.

Kahoot (https://kahoot.com/) is suitable for those who want to create their own knowledge tests in a fun way. Upon creating an account, the user can be a teacher, a student, or groups of students. Kahoot can be used as a class competition when everyone gets the link at once and also as a challenge for home repetition of the subject content. In this case, Kahoot is open for a limited time (one week) for everybody who gets the link. The students see their own position in the competition at the end of the game. The teacher has an overview of the difficulty of each question and also about the placement of each player. There are more types of questions in the paid version than in the free version. The resource was mentioned 4 times in our survey, typically gained an overall rating of 5 or 4, and would be used even in standard teaching mode; the visual was rated as appealing. The support of the interaction within school was mostly rated as primarily positive contrary to the outside interactions. Kahoot represents an example of a web resource which can have any content, not only a scientific one.

Mozaweb (https://www.mozaweb.com) is another paid application designed for science and humanities subjects. The registered teacher can use the digital teacher books, presentations, e-learning, 3D animations, videos, and tasks to test the knowledge of the students. The resource was mentioned 3 times in our survey (overall ratings 5, 5, 1), and would be used even in standard teaching mode; the visual was rated as appealing. The interaction support was rated as neutral.

“Chemistry in context” (Química en context, on Google sites) incorporates competency work in high school to support meaningful learning and motivate students toward chemistry through connections with everyday life. Among the activities, there are many inquiry-experimental ones, some of which take advantage of the potential of data recording and analysis equipment. In general, the activities focus on modeling, inquiry, and argumentation [80]. The resource was mentioned 3 times in our survey (overall ratings 5, 5, 4), and would be used even in standard teaching mode; the structure was rated as clear. The interaction support within the school was rated as neutral and outside school (with nature, family and society) as positive.

“We know the facts” (https://www.umimefakta.cz) is a tool suitable to repeat or test knowledge from almost all subjects. The school has to pay for the application to have the pupils in a class together and to allow the teacher to give them tasks and see the result. At the same time, individual students can use these activities individually for free; there is only a limit of 77 questions (approximately 30 minutes of revision) per day without registration. The tasks are done in a funny way at different difficulty levels. If teachers want to use the free version of the application, they can ask the students to send them a screenshot of the results. The resource was mentioned 2 times in our survey (overall ratings 4 and 3), and would be used even in standard teaching mode. The interaction support within the school was rated as neutral or nonexistent.

The following two apps were mentioned only once, but we consider them worth sharing.

The geocaching app (https://www.geocaching.com/sites/adventure-lab/en/) leads users outside. With the help of an application map, everyone sees the places that are near the position where they are standing. Students (without the teacher) follow the path given by the teacher and collect the specified information about the places they visit. There are many prepared places with content about their history or some interesting facts related to them, even in the free version. In the paid version, teachers can prepare their own places with the content they need. The resource can be used as it is, or modified for standard teaching mode. The interaction support among peers and with nature was rated as positive.

The virtual microscope app (https://www.ncbionetwork.org/iet/microscope/) is designed to familiarize students with the microscope in advance to using one. Everybody with the link gets into the application, for free, to see the structure of the microscope and can explore selected plant, animal, human, and bacteria microscopic slides. The slides are also described, and the user can change the magnification, adjust the light, and adjust both coarse and fine focus. The application points out that the magnification needs to be gradually changed or shows a broken slide if such a situation would occur in reality. There is also a test on the care and use of the microscope, calculation of magnification, and terminology. The pedagogical suitability was rated positively and the interaction support was rated neutral.

Limitations of the study

As in other similar studies, only teachers willing to participate in the survey shared their ideas with us. Not all science teachers in the countries were asked to participate, rather each institution used its own communication means to contact science teachers, which creates a bias. We only collected the web resources the science teachers would recommend; consequently, we do not know which web resources they have tried and possibly rejected. Therefore, we cannot distinguish which ones they do not want to use and which ones they do not know. As mentioned above, a resource is taken as a base entity in our survey, and therefore, it is impossible to distinguish reported resource features from the specific digital element context.