167Acta Chim. Slov. 2022, 69, 167–186 Dolničar et al.: The Students’ Perceptions Using 3DChemMol ... DOI: 10.17344/acsi.2021.7181 Scientific paper The Students’ Perceptions Using 3DChemMol Molecular Editor for Construction and Editing of Molecular Models Danica Dolničar,1,* Bojana Boh Podgornik1 and Vesna Ferk Savec2 1 Faculty of Natural Sciences and Engineering, University of Ljubljana, Snežniška 5, 1000 Ljubljana, Slovenia 2 Faculty of Education, University of Ljubljana, Kardeljeva ploščad 16, 1000 Ljubljana, Slovenia * Corresponding author: E-mail: danica.dolnicar@ntf.uni-lj.si Received: 09-30-2021 Abstract The paper presents a study in which 54 university students were introduced to a newly developed, free, web-based 3DChemMol molecular editor with a toolbar, which they then evaluated. The tool aims to increase representational competence related to submicroscopic representations. Students who used the software for the first time, were instructed to create molecular models using the model building/editing tools in three activities with varying levels of difficulty: 1) building a simple model (butanoic acid), 2) converting one model (hexane) into two models, 3) converting from a non-cyclic to a cyclic structure (glucose). It took students from two up to 15 minutes to accomplish each of the activities. Several types of help were available in the 3DChemMol molecular editor toolbar to assist students during their activi- ties, including a video tutorial, button hovering, action status display, and a help menu. Undo/redo and restart options were also available. Students’ completion level, difficulties, and use of the help features were investigated using student self-evaluation questionnaires. The 3DChemMol molecular editor proved to be a useful support for students in complet- ing simple chemistry activities. Students were successful in model building, although they encountered some specific difficulties, especially in steps that involved spatial operations, such as rotating the selected part of molecule around the bond. In students’ perception, the video tutorials were the preferred and most frequently used type of help, and the undo function was considered essential. The results suggest that the 3DChemMol molecular editor can be used effectively in introductory chemistry courses at the tertiary level, whether for direct instruction, self-study, or other forms of support in the pedagogical process. The results and new findings of this study will be used to further optimize the interface in future versions of the evaluated tool. Keywords: Representational competence; submicroscopic representations; learning chemistry; 3D model building; model editing tool 1. Introduction 1. 1. Visualization and Molecular Models in Chemistry Education The concept of visualization can be understood in three ways:1 visualization of objects (physical or graphic representations, static or dynamic, analog or digital, can be accompanied by sensory data), introspective visualization (mental models), and interpretive visualization (making meaning from the previous two forms). Vekiri2 states that graphical representations allow for more efficient process- ing of information compared to verbal representations, which reduces working memory load. The adoption of vis- ualization is not automatic but a function of prior knowl- edge.3 Understanding the core ideas introduced in chemis- try education involves engagement with their representa- tions and the associated phenomena.4 Johnstone5,6 was the first to propose three levels of representation of scientific concepts and processes: (1) macroscopic (e.g., chemistry experiments), (2) submicroscopic (e.g., molecular mod- els) and (3) symbolic (e.g., chemical formulae). The three types of representations relate to phenomena perceived through our senses and support explanations at qualita- tive and quantitative levels.4 Students often struggle with understanding and using the triplet concept. 3D models of molecules represent the submicroscopic representation, 168 Acta Chim. Slov. 2022, 69, 167–186 Dolničar et al.: The Students’ Perceptions Using 3DChemMol ... the use of which is important to bridge the gap between the macroscopic and symbolic levels.7 Kozma and Russell8 defined representational com- petence in science education as a set of five distinct abili- ties of students: to analyze the features of representations, transform between representations, create new representa- tions, explain the usefulness of representations, and ex- plain the advantages of representations. Activities aimed at improving representational competence support spatial thinking9 which is critical for understanding 3D spatial concepts in STEM (science, technology, engineering and mathematics) disciplines.10 The use of physical and virtual molecular models promotes representational competence11–13 and fosters spatial understanding,14 although the impact of spatial ability on success is influenced by learning strategy and task demands.15 Students who used models were more likely to implement new concepts, transform from 2D to 3D representations, and answer visual-spatial tasks. In the past, physical modeling kits with balls and sticks or mag- nets were used to construct 3D analog models of chem- ical compounds.16–18 Later, molecular modeling software brought chemical visualizations into the digital virtual realm.19–21 Since then, numerous stand-alone and web- based applications for viewing and manipulating chemi- cal structures have become available, such as, ArgusLab, Avogadro, BALLView, Biovia discovery studio visualizer, Chime, Chimera, JME molecular editor, Jmol/JSmol, Os- cail X, Pymol, RasMol, Spartan, SwissPDB Viewer, Tinker, Chemis 3D Molecular Viewer Applet, VMD, Yasara, and others.20,22–26 Some reported course activities and research in- volved the construction or use of physical models by stu- dents.27–32 Thayban et al.33 found that virtual models were more effective than physical model in learning symmetry. On the other hand, the use of physical or virtual molecular models was found to assist students in solving chemistry problems that require spatial thinking.34 Studies at all levels of chemistry education indicate that in order to construct correct mental models of chem- ical compounds, students should be engaged in construct- ing and manipulating three-dimensional (3D) visualiza- tions.35,36 The construction of submicroscopic models is part of representational competencies. Kelly and Akaygun37 suggested that visualizations are too often used only as a method of direct instruction. Instead of being passive observers students should become interactive participants and critical thinkers. In a survey38 that was part of the workshop for molecular visualization in science education researchers, educators, and software developers discussed the role of molecular modeling in college chemistry and were asked about the features of molecular representation and the types of interactions with molecular visualization that most help students. The responses suggested that students should be able to create their own visualizations and interact with existing ones. In some reported course activities, students were us- ing molecular modeling software. Some of the advantages over physical modeling are flexibility in model building, switching between different representations, and accura- cy of structural representations.39 According to Kozma,8 the construction, calculations and manipulation of mo- lecular models support the laboratory practice of synthe- sis by looking at reaction sites and speculating on reac- tion mechanisms. Clauss and Nelsen40 used WebMo and Gaussian to teach students the fundamentals of computa- tional chemistry by performing ab initio and DFT (density functional theory) calculations in an undergraduate lab- oratory course with the goal of gaining a deeper under- standing of their experimental work. Linenberger et al.41 conducted a guided experiment using the student version of Spartan to discover the relationship between structure and molecular properties, e.g., through measurements, calculating dipole moments, and studying electron density potential maps and molecular shapes. Raiyn and Rayan42 reported on the impact of a workshop using ChemDraw in a college chemistry course that significantly improved students’ understanding of 3D structure and polarity, boiling point, and isomerism. Rothe & Zygmunt43 used Gauss View 5 and Gaussian in an undergraduate chemical reaction engineering course to promote understanding of the relationship between molecular properties and mac- roscopic concepts such as internal energy, enthalpy, rate constants, and activation energies. In a web-based chem- istry course, Dori et al.44 gave first-year students the task of using Weblab and IsisDraw to create molecular models, calculate molecular weight, and construct the hybridiza- tion and electric charge distribution of carbon atoms. On the posttest, which required higher-order thinking skills, the experimental students showed better reasoning skills and a better ability to transfer between levels of representa- tion than the control group. Ealy45 introduced molecular modeling using Spartan Pro to a general chemistry labora- tory. Students performed measurements and investigated properties such as symmetry, electrostatic potential, and dipole. The experimental group performed significantly better than the control group, and the test results at the end of the semester also showed that a transfer of knowl- edge had occurred. In an ethnographic study by Kozma,46 students who first conducted experiments in the laborato- ry and then constructed molecular models using Spartan. When using the computer modeling software, students referred to chemical concepts (e.g., atoms, bonds, elec- tronegativity, dipole moment) more frequently than in the laboratory session. Yet, they did not relate the models to the materials they synthesized. Molecular modeling was used by pre-service teachers in combination with class- room materials and mind map tools to learn hydrogen bond.47 Kolar et al.48 suggested the didactic use of com- putational chemistry to create models of amides to illus- trate acid-base properties. Winfield et al.49 have developed activities that incorporate model building in the iSpartan 169Acta Chim. Slov. 2022, 69, 167–186 Dolničar et al.: The Students’ Perceptions Using 3DChemMol ... tool to teach conformations of alkanes. Similarly, Johnson et al.50 reported integrating of iSpartan into the classical organic chemistry laboratory experiment to help students learn about the stability of alkenes. Conformational analy- sis of small molecules using Vega ZZ software was used by Soulère51 in an undergraduate chemistry course. User-friendliness of graphical interfaces to opti- mize small and medium sized molecules has enabled the possibility to introduce computational chemistry tools to the undergraduate level.52 Rodriguez-Becerra et al.53 described the use of educational computational tools on pre-service chemistry teachers, with Avogadro used for model building. Due to the identified deficiency in educational use of molecular modeling in chemistry classes by teachers and/or students,54 molecular modeling was introduced into chemistry education by Aksela et al.,55 developing pedagogical solutions, training mentors, creating teaching materials and investigating their effectiveness. The mode- ling approach was adopted by schools and the experiences were shared in a book.56 The Edumol.fi web application was used.57 1. 2. Tools for Building Molecular Models in Teaching Organic Chemistry At the beginning of this study, we analyzed existing molecular modeling tools for teaching organic chemistry at the university undergraduate level in order to select the most appropriate tool to serve as the basis for the devel- opment of a new tool, 3DChemMol molecular editor.58 Its editing functionality and help tools are described and eval- uated in this article. Some of the external factors influencing the poten- tial for wider adoption of molecular visualization tools for teaching and active learning could be their suitability for a particular level of education (primary/secondary and college), their focus (small molecules, macromolecules, crystal structures), the presence of editing feature (mo- lecular modeling), functionalities (display of properties), and their cost and convenience. The degree of complexi- ty and the usability of the user interface could also play a role. With the advent of web-based technologies (HTML5, CSS, WebGL, canvases, and the use of JavaScript), there has been a shift from standalone applications and web applications requiring plug-ins to readily available web- based tools.59 In terms of availability, molecular modeling tools have been developed that are open source.60 In this study we focus on the software that is suitable for educa- tion, focuses on small molecules, allows molecular mode- ling and is freely available. Some of the tools are compared in Table 1. Due to immediate availability, we limited our choice to web-based applications that do not require in- stallation. These criteria exclude tools such as Spartan20 (proprietary, standalone), Web Doodle Web Components61 (proprietary, web-based), Avogadro62 and Jmol63 (free, standalone), leaving us with mainly web-based tools. We also excluded web tools that are viewers only (e.g. 3dmol. js64) or those that involve creating a 3D model by drawing a 2D structure (e.g. MolView65). The remaining web-based interfaces were based on JSmol,66 a web version of Jmol. They included interfaces for the creation of 3D models: CheMagic,67 MolCalc68 and 3DChemMol.58 The latter was developed by the first author of this study. The original JSmol editing module is menu-based, cumbersome to use, and lacks a functional undo and help function. CheMagic has implemented both, but the functionalities of the tool (as in JSmol) are all visible at once, which can be distract- ing if you are only focused on editing. MolCalc’s editing feature creates the input for the computational software. It is simple and efficient but uses only basic editing func- tions. 3DChemMol was designed to structure the JSmol functionalities into multiple toolbars accessible from the main menu, including editing, with additional interactive functions with toolbars for model exploration (e.g., elec- tronegativity, measurement, symmetry, creating confor- mations and isomers, model comparison, and exercises). It was chosen for our study because the new editing interface is intended to resemble that of familiar 2D editing tools. 1. 3. Motivation and Aims of the Study The aim of this study was to evaluate the newly de- veloped 3DChemMol molecular editor tool and to investi- gate university students’ first encounter with a 3D struc- ture editing tool while performing three specific activities Table 1: Characteristics of selected freely available user interfaces for 3D model building Tool name Type Technology GUI elements Characteristics Avogadro S C++, Qt Menus, toolbars, dialogs Editing dialog, mode switching for rotation Jmol S Java Menus, toolbar, dialogs Editing menu on right click JSmol (original) W JavaScript, JQuery Menus (right click) Editing menu on right click CheMagic (JSmol) W JavaScript, JQuery Dashboard buttons All tool functionalities at once, editing buttons, undo, help MolCalc (JSmol) W JavaScript, JQuery Buttons Basic editing (adding, deleting), input for computational software 3DChemMol (JSmol) W JavaScript, JQuery Menus, toolbars Editing toolbar, undo, help Types: S = standalone, W = web-based 170 Acta Chim. Slov. 2022, 69, 167–186 Dolničar et al.: The Students’ Perceptions Using 3DChemMol ... for creating and editing molecular models. The research questions were as follows: – RQ1: How successful were students in performing sim- ple chemistry activities using the 3DChemMol molecular editor, and how was their success related to the time re- quired and the perceived difficulty of the activity? – RQ2: What types of difficulties did students encounter when performing activities with 3DChemMol molecular editor? What was the cause and a possible remedy? – RQ3: How did students use the different types of help available in 3DChemMol molecular editor and addition- al support when they encountered problems? 2. Methods 2. 1. Participants A total of 54 students of the University of Ljubljana participated in the study. They were enrolled in the second year of study (aged 20 to 21) at the Faculty of Education (17 students, 31.5%) or the Faculty of Health (37 students, 68.5%) in the study year 2020/21. They had already taken basic chemistry courses in general and inorganic chem- istry; therefore, basic knowledge and understanding of chemistry principles and basic ICT skills were assumed. Introduction to building 3D models of chemical com- pounds was designed as a foundation for organic chem- istry and other higher level chemistry courses that follow in their program of study. Apart from the field of study, there were no additional differences between the groups, important for the purpose of this research. 2. 2. Materials 2. 2. 1. Model Building Tool The editing module of the web-based tool 3DChem- Mol molecular editor (http://www2.arnes.si/~supddol- n/3dchemmol), previously created by the author of this study,58 was used to construct the molecular models. The tool is based on JSmol software for visualization and edit- ing of 3D molecular models. Model creation is performed in 3D using a graphical user interface consisting of the model window and toolbar (Figure 1). The tool contains basic model building functionalities, but also some ad- vanced features that allow the creation of different confor- mations and isomers. The available model interactions (e.g., clicking or drag- ging on atoms/bonds) depend on the current action mode. There are four atom action modes (add/edit, delete, move, invert-substitute switch) and three bond action modes (add/ edit, delete, rotate around bond). Switching between action modes is done by selecting a mode from the list. One of the additional elements implemented in the tool is the Undo/Redo function, which did not work in the original JSmol application. Four types of help are integrated and available at all times: a) status indicator of the currently available action mode, displayed at the bottom of the model window (op- tional), b) explanations of button actions when hovering the mouse over them, c) help menu with image and text explanations of the toolbar, d) video tutorial with examples of structure building, also available from the help menu. One of the standard functions of model building is geometry optimization. The tool also allows to quickly cre- ate an image from the model window. 2. 2. 2. Problem Set Three simple activities were designed to guide stu- dents in building and editing models using our tool. Each activity required students to create or edit a specific mole- cule with a limited number of actions. – Activity 1: Build a simple model of the molecule – buta- noic acid (new model, add/change atoms, change bond type). This activity did not require any change in action mode – all the functions needed to build a model were already present. – Activity 2: Convert from one to two models of the mol- ecules – hexane to ethene and butane (delete bonds, de- lete atoms, manually add hydrogen atoms, change bond type). – Activity 3: Convert from a non-cyclic to a cyclic model of the molecule – glucose (add bonds, rotate branches around a bond, change bond type). The full list of steps for each activity can be found in Table 2. All activities included common features such as changing the bond type (with some differences) and auto- matic geometry optimization. At the end of each activity, students had to create an image of the final model of the molecule. Time for each activity was not limited. Some steps required a simple click on a toolbar but- ton, while others required direct interaction with the mod- el or a combination of both (Table 2). The model interac- tions available depended on the current action modes.Figure 1: User interface of the 3DChemMol molecular editor 171Acta Chim. Slov. 2022, 69, 167–186 Dolničar et al.: The Students’ Perceptions Using 3DChemMol ... For each activity, students were provided with the editing tool interface, which contained the initial model on the left half of the screen and activity instructions in Google Forms on the right half (Figure 2). The activity in- structions consisted of a) general information about the availability of free model rotation, undo/redo functions, and various types of help; b) an image of the 3D output model (which was also displayed in the interface); c) a Table 2: Steps for each activity with the required interaction with the toolbar and the 3D model Interaction with the toolbar Interaction with the 3D model Step # Step content Button Type Mode Atom Atom Bond click change change click drag click Activity 1: Building a simple model of the molecule 1 New model x 2 Adding C atoms x 3 Adding heteroatoms x x 4 Changing the bond type x 5 Model centering x 6 Geometry optimization x 7 Creating an image x Activity 2: Converting one model into two models of the molecules 1 Deleting bonds x x 2 Changing the bond type x x 3 Deleting atoms x x 4 Adding hydrogen (manually) x x x 5 Geometry optimization x 6 Creating an image x Activity 3: Converting from a non-cyclic to a cyclic form of the molecule 1 Adding a bond x 2 Changing the bond type x 3 Geometry optimization x 4 Rotating a branch around the bond x x 5 Geometry optimization x 6 Creating an image x Figure 2: Activity display for the first activity (left: interface for model building, right: activity instructions) 172 Acta Chim. Slov. 2022, 69, 167–186 Dolničar et al.: The Students’ Perceptions Using 3DChemMol ... short, annotated video tutorial explaining relevant actions on another example model; d) step-by-step instructions on how to build the target model, which referred to indi- vidual actions rather than elements of the interface; e) an image of the 3D target model. Students could scroll up and down through the instructions. 2. 2. 3. Students’ Self-evaluation Questionnaires For each activity, a self-evaluation questionnaire was included at the end of the activity instructions (Google Forms) with the following items/questions: – degree of activity completion – completion level (Likert scale 1–5: 1 = started, 5 = fully completed); – time spent on the activity (in minutes, as reported by students); – perception of activity difficulty – perceived difficulty level (Likert scale 1–5: 1 = easy, 5 = difficult); – type(s) of help used (multiple choice: a) video tutorial (single view), b) video tutorial (multiple views), c) hov- er on toolbar, d) current action status, e) help menu); – other actions used (multiple choice: a) free view rota- tion, b) undo, c) redo, d) restart activity); – severity of difficulties encountered for each step of the activity – step difficulty level (Likert scale 1–5: 1 = no difficulties, 5 = severe difficulties); – difficulty description (text). Prior to the study, two researchers (the co-authors of the study) optimized the instrument by performing a face validity69 check. They completed the suggested activ- ities and reviewed the questionnaires and then suggested changes and adjustments. 2. 3. Data Collection The testing was conducted in May 2021 and was su- pervised by the authors in an online format. The Zoom videoconferencing tool and a web browser were used to display the tool and instructions with the questionnaires. Students consented to data analysis. Prior to testing, a standardized introductory pro- tocol was used that included clarification of purpose, in- structions, voluntary participation, and acknowledgement of participation. The research was approved by the com- petent authorities of University of Ljubljana. None of the students had any prior experience with the tool. The teach- er first gave a general introduction/demonstration of the entire 3DChemMol molecular editor. Students had access to the interface. Students were then given links to the ac- tivities. After completing each activity, they completed the questionnaire and moved on to the next activity. 2. 4. Data Analysis Data from the students’ self-evaluation questionnaires were collected in Google Spreadsheets and transferred to Excel and Statistical Package for the Social Sciences (SPSS), version 26 for analysis, which was performed for each of the three activities. – Mean scores were calculated for continuous and or- dinal questionnaire items, including completion lev- el, time spent, perceived activity difficulty level, and step difficulty levels. Step difficulty mean was also calculated for each activity. The two multiple-choice questions (type of help used, other items used) were transformed into multiple dichotomous variables, one for each response (1 if the response was selected and 0 if it was not). Means were calculated for each response. – The distributions of the variables were examined using the frequency of the results expressed as a percentage of students. This was done for ordinal items and multi- ple-choice responses, and also for time spent on activity, where scores were first divided into five groups. – Correlations between parameters were calculated using Spearman correlation coefficient (rs). – The open-ended questions from the student self-eval- uation questionnaires were also recorded in Google Spreadsheets and transferred to Excel. The students’ responses were coded using a coding table. The coding table was derived from a qualitative analysis of 20% of the questionnaires (n = 11 participants); the reliability of the coding was ensured by independent coding by two researchers (the authors of this article). Finally, both evaluations were contrasted at the points where differ- ences occurred and, after consideration, the more appro- priate one was selected. Altogether, 99% reliability was achieved. 3. Results and Discussion 3. 1. Completion Level of the Activities Completion level of the activities was measured by the self-evaluation questionnaire. For each activity, the time spent on the activity and the perceived level of diffi- culty were also reported. 3. 1. 1. Means and Distributions Students were relatively successful in completing the simple chemistry activities, as measured on a Likert scale of 1 to 5. The average score was above 4 for all 3 activities (Figure 3). For the first two activities, the completion level was very high with 91 and 96% of students reporting that they completed the activity, compared to only 53% for the third activity (Figure 4). The completion time, measured in minutes, showed that the majority of students took between 3 and 5 minutes for each of the first two activities, while most students took 6–10 minutes for the last activity (Figure 5), with a signifi- cantly higher mean (Figure 3). 173Acta Chim. Slov. 2022, 69, 167–186 Dolničar et al.: The Students’ Perceptions Using 3DChemMol ... Figure 3: Mean scores with SD (whiskers) by activity for activity completion level (1–5), time spent on activity (min), and perceived activity difficulty (1–5) Figure 4: Distribution of activity completion levels by activity (5 = fully completed) Figure 6: Distribution of perceived activity difficulty by activity (5 = difficult) Figure 5: Distribution of time spent on activity by activity Perceived difficulty, expressed on a Likert scale of 1–5 (5 being difficult), showed that the second activity was considered the easiest with a mean of 1.81, and the third activity was considered the most difficult, with a mean of 3.18 (Figure 3). The most common response for activity 1 was difficulty level 2, for activity 2 was difficulty level 1, and for activity 3 was difficulty level 3 (Figure 6). 3. 1. 2. Correlations No significant correlation was found between time spent and activity completion (Table 3). Some students took more time, but still completed the activity. An exam- ple is a comment on activity 1: “I had trouble adding atoms at first but figured it out after a few minutes.” As expected, time spent correlated positively with perceived difficulty (most strongly for the second – overall easiest activity). Students who spent more time on the activity perceived it to be more difficult. The negative correlation between completion and perceived difficulty was significant for the third – the hardest overall activity – suggesting that stu- dents who did not complete the activity perceived it to be more difficult. For example, a student’s comment was: “It is difficult to have spatial orientation.” The lower correlation between perceived difficulty and completion level for the first two activities was due to the high completion levels for these activities. Similar correlations between perceived difficulty as a determinant of Web search performance and time have been found in a study by Kim.70 Table 3: Spearman correlations between completion level (Compl.), time spent (Time) and perceived difficulty of activities (Perc. diff.) Param. Compl. Time Perc. diff. Activity 1: Building a simple model of the molecule Compl. 1.000 Time –0.069 1.000 Perc. diff. –0.239 0.404b 1.000 Activity 2: Converting one model into two models of the molecules Compl. 1.000 Time –0.081 1.000 Perc. diff. –0.266 0.584b 1.000 Activity 3: Converting from a non-cyclic to a cyclic form of the molecule Compl. 1.000 Time –0.158 1.000 Perc. diff. –0.469b 0.435b 1.000 bp < 0.01 174 Acta Chim. Slov. 2022, 69, 167–186 Dolničar et al.: The Students’ Perceptions Using 3DChemMol ... 3. 2. Difficulties During the Activities 3. 2. 1. Mean Scores by Activity The step difficulty mean for each activity reflects the average amount of difficulties students encountered during steps of an activity. The scores (Figure 7) show the same trend as the time spent and perceived difficulty of the activities (Figure 3). Students reported the greatest step difficulty mean on the third activity and the smallest on the second activity. Means ranged from 1.62 to 1.91, which is relatively low given the Likert scale of 1 and 5. For all activities, some students specifically stated: “No problems,” and several others made no comment. Mean scores are low due to the proportion of steps that are not problematic and those that are less problematic. Examples of repeated com- ments in all activities related to some technical difficulties were: “I can’t save the image.” Figure 7: Step difficulty mean with SD (whiskers) by activity (1 = no difficulties, 5 = severe difficulties) 3. 2. 2. Mean Scores by Interaction Type In the previous section the steps were grouped by activities. Here we grouped steps in multiple ways and calculated step difficulty mean for each group. The grouping in Table 4 by type of interaction shows that bond interactions caused more difficulties than atom in- teractions. Toolbar interactions with button click were the least problematic. Table 4: Step difficulty mean by interaction type Interaction type Step diff. mean Toolbar button click 1.48 Atom interaction 1.76 Bond interaction 1.86 Steps with atom and bond interactions were also classified into four groups (Table 5). Actions that re- quired selection of the atom or bond type on a tool- bar button prior to direct interaction with the model caused fewer difficulties than those that did not require a preceding action on the toolbar. On average, the most difficult actions were those that required a change of ac- tion mode (selection on the toolbar from a list of modes, e.g., add/change, delete). The action requiring a combi- nation of type and mode change was also deemed more difficult. Table 5: Step difficulty means for direct interaction with the model, depending on the preceding action Preceding action Step diff. mean Button type change 1.54 No action 1.80 Button mode change 1.95 Button type + mode change 1.98 Another classification of steps was applied to direct interactions with atoms and bonds: clicking, dragging and repeated actions (Table 6). Repeated mouse clicking caused the most difficulties, followed by mouse dragging. A single mouse click on a bond or on atom was the least problematic. Repeated clicking was related to geometry changes in our case. Table 6: Step difficulty means in direct interaction with the model, depending on the type of mouse interaction and repetition Direct interaction type Step diff. mean Mouse click 1.58 Mouse drag 2.07 Mouse click + repetition 2.90 The last grouping of atom and bond interactions concerned geometry change (Table 7). The fewest diffi- culties arose from automatic geometry optimization. No direct interaction with the model was required. Actions where no significant geometry change occurred (nothing added, no automatic hydrogen adjustment) were consid- ered less problematic. The most difficulties occurred when the geometry was changed, highlighting the importance of spatial abilities. Table 7: Step difficulty means when interacting directly with the model, depending on the type of geometry change Type of geometry change Step diff. mean Geometry optimization 1.35 Small geometry change 1.59 Significant geometry change 1.96 175Acta Chim. Slov. 2022, 69, 167–186 Dolničar et al.: The Students’ Perceptions Using 3DChemMol ... 3. 2. 3. Scores by Step Step difficulty levels for each step of the three activ- ities, presented in Table 8, were ranked from 1 (easiest) to 18 (most problematic). Scores for individual steps ranged between 1.21 and 2.90. Total step difficulty mean was 1.66. Activity 1: Building a model of butanoic acid. The eas- iest steps involved two actions available through a simple button click: creating a new structure (ranked 4 out of 18) and geometry optimization (ranked 5). Moderate difficul- ties were encountered in adding C atoms (rank 9) to build the main skeleton of the structure. This step is crucial. Some of the students reported difficulties, such as: “When click- ing with the mouse, an atom was deleted instead of added.” This was because the mouse was moved when clicking on a hydrogen atom. Instead, the “drag” event was registered, which in Jmol is associated with deleting an atom when ap- plied to a hydrogen atom. Comments also related to add- ing heteroatoms (rank 13): “I can’t position the chain as it is shown in the result.” and “Sometimes atoms are added in strange ways.” Another comment: “In the beginning, I had a lot of problems with adding atoms unevenly.” Students were paying attention to structure but not configuration. Adding and replacing atoms only required clicking on existing atoms. There was not much chance for error, so “strange ways” and “unevenly” likely refers to configura- tions that result in isomers of the target structure. In this first activity, students have not yet learned how to make configuration changes. Adding atoms correctly required good spatial orientation. There were some difficulties with centering the model (ranked 12). Comment: “I had trou- ble centering the model until I found the centering button. It would be beneficial if centering was automatic because centering has to be applied repeatedly when building larger structures.” This difficulty could have to do with fa- miliarity with the center button, but students also forgot that they could not only rotate the model during model construction but also zoom it out. The zoom button was not part of the editing toolbar, but was an available mouse shortcut (mouse wheel). Surprisingly, most of the difficul- ties with this activity occurred when it came to changing the bond type (ranked 15), which should be quite simple by just clicking on a bond to increase its order. Increasing the bond order was not included as a toolbar button but was part of the default add/delete action mode. There was no need to change the action mode. The comment “The number of hydrogens doesn’t automatically adjust.” sug- gests that students tried to use a different method where they selected the bond type and clicked on a bond. This process does not currently adjust the hydrogens. Students did not know the shortcut even though it was shown in the introductory video. The two methods should be made compatible. Creating an image (ranked 14) also caused dif- ficulties for some students, as expressed in a comment: “I can’t convert to an image. Numbers appear instead.” The reason here was that some system configurations automat- ically generated a text file with the structure in mol format Table 8: Steps for each activity with interaction types, step difficulty levels and ranks Step # Step content Button, Type, Mouse click, Geom. Inter. Step Step diff. atom, bond mode chg . drag, rep. chg. type* diff. level rank Activity 1: Building a simple model of the molecule 1 New model c c 1.44 4 2 Adding C atoms a – k y a 1.52 9 3 Adding heteroatoms a t k y t+a 1.67 13 4 Changing the bond type b – k y b 1.93 15 5 Model centering c c 1.63 12 6 Geometry optimization c g c 1.46 5 7 Creating an image c c 1.69 14 Activity 2: Converting one model into two models of the molecules 1 Deleting bonds b m k n m+b 1.49 8 2 Changing the bond type b t k n t+b 1.40 3 3 Deleting atoms a m k n m+a 1.47 6 4 Adding hydrogen a tm d n tm+ad 1.98 16 5 Geometry optimization c g c 1.21 1 6 Creating an image c c 1.47 7 Activity 3: Converting from a non-cyclic to a cyclic form of the molecule 1 Adding a bond a – d y ad 2.16 17 2 Changing the bond type b – k y b 1.57 11 4 Rotating a branch b m kr y m+br 2.90 18 3&5 Geometry optimization c g c 1.37 2 6 Creating an image c c 1.53 10 * Key to interaction types: c – toolbar button click, a – atom interaction, b – bond interaction, t – button type change, m – button mode change, k – mouse click, d – mouse drag, r – repetition, g – geometry optimization, n – small geometry change, y – significant geometry change 176 Acta Chim. Slov. 2022, 69, 167–186 Dolničar et al.: The Students’ Perceptions Using 3DChemMol ... instead of the image file. This technical issue needs to be addressed and fixed in the future. The issue mentioned in a comment: “I had no particular problems constructing the model, but the angles between atoms aren’t the same.” was either related to configuration or the student did not optimize the model geometry correctly. A comment from a student who reported no individual difficulties was: “The correct tool is not visible.” In this case, the comment could refer to shortcuts built into the editor that are not explicitly visible in the toolbar (e.g., changing the bond type in gen- eral mode). This activity did not require any action mode changes but some students had expected them. Activity 2: Splitting the model of hexane into models of butane and ethene (cracking). In this assignment, sev- eral students reported, “I had no problems.” Geometry optimization and bond change were considered the easi- est steps by students (ranked 1 and 3, respectively). Here, bond change was performed by first selecting the bond type from the toolbar (no shortcut used). This method did not automatically adjust the number of hydrogen atoms, but unlike the first activity, the subsequent steps were de- signed to solve this problem. Deleting atoms and bonds did not cause too many difficulties (rank 6 and 8), how- ever, a student commented: “Problems switching between adding and deleting atoms.” The reason is that the delete function is not immediately visible but is in a list of action modes in the toolbar. The most problematic part of the ac- tivity was the manual hydrogen addition (ranked 16). It consists of selecting the hydrogen atom type in the toolbar and then dragging out an existing atom with the mouse. A typical comment was: “Problems with adding the sin- gle H atom due to the fact that addition and modification appear together.” As with the first activity, more than one action is available in Add/Change mode, depending on the type of interaction (click, drag), the object of interaction (atom, bond), and sometimes the type of atom (hydrogen, non-hydrogen). There is no separate button or selection on the toolbar for this action. As with the first activity, students may have been looking for a separate mode and could not find the button. Adding the H atoms by drag- ging was otherwise covered in the tutorial video and also shown in the action mode text help at the bottom of the screen. Interestingly, some of the difficulties were relat- ed to a functionality not being available. A student com- mented: “The button to move one of the models did not work, so I could only rotate the left model.” The reason is that moving and rotating individual models is not possi- ble in edit mode. Only the entire view can be rotated. This functionality could be incorporated in the future, as it is already present in other toolbars of this software. Image creation difficulties were not rated as severe (rank 7) for this activity, although the same technical obstacles were encountered. Comment: “I could not save the image, so I took a screenshot instead.” Perhaps the severity changed or there were other novice difficulties saving the file in the first activity. Activity 3: Converting the noncyclic form to a cyclic form of glucose. The only unproblematic action in this activity was geometry optimization (rank 2). Changing the bond type from double to single bond was perceived moderately difficult (rank 11). Some students remem- bered the shortcut from the first activity, others did not. A typical comment was: “I had a problem changing the bond.” Creating an image was also still an issue (ranked Table 9: Summary of the most frequent difficulties with example student comments Act. # Step # Theme / Step Category* Possible issue Step diff. rank** Example student comment 1 2 Ading C atoms a Interface 9 “When clicking with the mouse, an atom was deleted instead of added.” 3 Adding heteroatoms a Spatial ability 13 “I can’t position the chain as it is shown in the result.” 4 Changing the bond b Interface 15 “The number of hydrogens doesn’t automatically adjust.” 5 Model centering c Interface 12 “I had trouble centering the model until I found the centering button.” 7 Creating an image c Technical 14 “I can’t convert to an image. Numbers appear instead.” 2 4 Adding hydrogen a Interface 16 „Problems with adding the single H atom due to the fact that addition and modifi- cation appear together.“ 3 1 Adding a bond b Interface 17 „I didn‘t know how to connect the O atom to the other side...“ 4 Rotating a branch b Spatial ability 18 “One of the groups was always oriented in the wrong direction.“ „It is difficult to have spatial orientation.“ * Key to categories – interaction types: a – atom interaction, b – bond interaction, c – toolbar interaction ** Key to step difficulty rank: 1 = easiest, 18 = most difficult among all steps 177Acta Chim. Slov. 2022, 69, 167–186 Dolničar et al.: The Students’ Perceptions Using 3DChemMol ... 10) to some, with a comment: “I couldn’t save the image.” Adding a bond between two existing atoms and especially rotating a branch around a bond were the two most prob- lematic steps overall (ranked 17 and 18). The latter step had a difficulty of 2.9, which is one grade above the former step at 2.16. Some students did not know how to connect two atoms, as evident in a comment: “I didn’t know how to connect the O atom to the other side and what the correct rotation was.” or a comment: “Having trouble connecting the structure properly.” Dragging was required in Add/ Modify mode, so no mode change was required in this step and no toolbar button was available. The appropriate ac- tion was demonstrated in the tutorial video and shown in the action status help at the bottom of the screen. Perhaps the model itself was part of the problem. It needed to be properly oriented so that the atoms could be reached with the mouse. Good spatial orientation could be related to this action. This was even more evident when the branch was rotated, as a student wrote in a comment: “I couldn’t get the model aligned the way it was in the picture. One of the groups was always oriented in the wrong direction.” or another student “I couldn’t place the atoms in the posi- tion shown in the resulting image.” The branch rotations around the bond were done in 60-degree increments. Stu- dents had to determine the correct degree of rotation by applying (repeating) the action the appropriate number of times. Another comment “It is difficult to have spatial ori- entation.” suggested that this activity required more spatial orientation than the first two activities. Comment, “It was difficult to begin the activity. Watching the tutorial video was crucial. Still, I had trouble rotating the bonds.” The first sentence (beginning of the activity) refers to the bond addition. Although this activity proved to be the most dif- ficult overall, four students indicated, “No problems.” This is consistent with the research of Harle and Towns who noted that rotational transformations were among the tasks that students had particular difficulty with.71 The most typical themes and categories of students’ difficulties that emerged from the above analysis are list- ed in Table 9. Of the eight themes, three each related to atom and bond manipulations and the remaining two to toolbar interaction. Two of the issues are probably relat- ed to the students’ lack of spatial orientation, which could be improved through training. Another requires solving a technical issue. The rest could be possibly avoided/fixed by redesigning parts of the user interface (e.g. even more visi- ble action status, separation of actions that are too similar, separate buttons instead of mode selection). 3. 2. 4. Correlations There are significant correlations between most steps within an activity in terms of difficulties (Tables 10–12). Mean of step difficulties is included as step mean. In the Table 10: Spearman correlations between step difficulty levels within Activity 1 Step Step Step Description 1 2 3 4 5 6 7 mean 1 New model 1.000 2 Adding C atoms 0.620b 1.000 3 Adding heteroatoms 0.479b 0.646b 1.000 4 Changing the bond type 0.441b 0.508b 0.482b 1.000 5 Model centering 0.408b 0.473b 0.414b 0.505b 1.000 6 Geometry optimization 0.426b 0.587b 0.392b 0.343a 0.718b 1.000 7 Creating an image 0.272a 0.408b 0.255 0.238 0.281a 0.506b 1.000 Step mean 0.629b 0.690b 0.625b 0.752b 0.690b 0.631b 0.568b 1.000 a p < 0.05, bp < 0.01 Table 11: Spearman correlations between step difficulty levels within Activity 2 Step Step Step Description 1 2 3 4 5 6 mean 1 Deleting bonds 1.000 2 Changing the bond type 0.562b 1.000 3 Deleting atoms 0.687b 0.740b 1.000 4 Adding hydrogen 0.401b 0.283a 0.332a 1.000 5 Geometry optimization 0.423b 0.672b 0.439b 0.351a 1.000 6 Creating an image 0.185 0.313a 0.254 0.119 0.357b 1.000 Step mean 0.675b 0.657b 0.698b 0.765b 0.537b 0.474b 1.000 a p < 0.05, bp < 0.01 178 Acta Chim. Slov. 2022, 69, 167–186 Dolničar et al.: The Students’ Perceptions Using 3DChemMol ... final step – saving the image of the result – the correlations are not as strong, as the difficulties with image creation were largely a technical issue. Difficulties with rotating a branch around a bound (third activity) also do not cor- relate with all other steps of the activity, as many students had difficulties in this step. 3. 2. 5. Correlations with Completion Level of the Activities The step difficulty mean for each activity correlated positively with time spent and perceived activity difficul- ty and negatively with activity completion (Table 13). The completion level for the second activity was very high, so the correlation with step difficulty mean was not signifi- cant. 3. 3. Help Tools Used During Activities The forms of help available included the tutorial vid- eo, the help menu, the description of the toolbar button when the user hovers over it, and the description of the ac- tions currently available on the structure (atom and bond actions). If students made mistakes, they could undo and Table 13: Spearman correlations of step difficulty mean with completion level (Comp.), time spent (Time) and perceived activity difficulty (Perc. diff.) Step Description Comp. Time Perc. diff. Activity 1: Building a simple model of the molecule 1 New model –0.419b 0.421b 0.522b 2 Adding C atoms –0.421b 0.422b 0.503b 3 Adding heteroatoms –0.235 0.512b 0.553b 4 Changing the bond type –0.135 0.503b 0.290a 5 Model centering –0.231 0.585b 0.353b 6 Geometry optimization –0.290a 0.458b 0.356b 7 Creating an image –0.121 0.224 0.292a Mean –0.334a 0.620b 0.539b Activity 2: Converting one model into two models of the molecules 1 Deleting bonds –0.305a 0.410b 0.307a 2 Changing the bond type –0.341a 0.488b 0.518b 3 Deleting atoms –0.296a 0.494b 0.492b 4 Adding hydrogen (manually) –0.041 0.467b 0.496b 5 Geometry optimization 0.083 0.506b 0.428b 6 Creating an image –0.152 0.205 0.276a Mean –0.215 0.558b 0.599b Activity 3: Converting from a non-cyclic to a cyclic form of the molecule 1 Adding a bond –0.104 0.305a 0.280a 2 Changing the bond type –0.333a 0.298a 0.344a 4 Rotating a branch –0.445b 0.409b 0.468b 3 and 5 Geometry optimization –0.115 0.203 0.236 6 Creating an image –0.211 0.063 0.179 Mean –0.424b 0.493b 0.508b a p < 0.05, b p < 0.01 Table 12: Spearman correlations between step difficulty levels within Activity 3 Step Step Step Description 1 2 4 3 and 5 6 mean 1 Adding a bond 1.000 2 Changing the bond type 0.350a 1.000 4 Rotating a branch 0.276a 0.270 1.000 3 and 5 Geometry optimization 0.285a 0.326a 0.086 1.000 6 Creating an image 0.365b 0.169 0.037 0.331a 1.000 Step mean 0.777b 0.579b 0.630b 0.501b 0.516b 1.000 ap < 0.05, bp < 0.01 179Acta Chim. Slov. 2022, 69, 167–186 Dolničar et al.: The Students’ Perceptions Using 3DChemMol ... redo previous actions. They were free to rotate the models during the construction process. If none of the previous actions helped, they could restart the activity. 3. 3. 1. Distributions Of the above actions with help tools, free rotation and the undo button were used by most students (70–94%) (Figure 8). The frequency of free rotation was lowest in the second activity because fewer configuration changes (de- leting atoms and bonds as opposed to adding them) were made than in the other two activities. Nevertheless, 14% of students reported not rotating the model in the third activity, which involved a larger configuration change when adding a bond to form a ring, as well as rotating a branch around a bond. The number of students who used the undo feature increased by 20% in the third activity, as only 6% of students did without it. This indicates the importance of the undo function, which did not work in the original JSmol interface. Redo function was not used as frequently, although its use increased with each activity and one in four students used it by the third activity. The most commonly used type of help was watching the tutorial video once, followed by the mouse-over button action. About 30% of students reported not watching the video in the first two activities, but in the third activity, the number of multiple video viewings increased significantly: One in three students watched the video more than once, compared to 4–9% in the previous activities. An example of a student comment on this activity is: “Watching the tu- torial video was crucial.” The use of the mouse-over action was comparable in all three activities and was used by less than half of the students. The last two help options (action status and help menu) were used less frequently, increasing from less than 10% in the first two activities to about 15% in the last activity. This could mean that students were not confused about the current action status (work mode) or that they missed the textual status display at the bottom of the screen. Interestingly, they also made little use of the help menu, which could indicate that they found the video tutorials largely sufficient. This is consistent with the con- clusion of a study by Van Der Meij,72 in which video tuto- rials that previewed the training activities were the most effective for learning software. The help menu provided similar information to hovering over the buttons. Finally, the level of activity restarting was low (9%) for the first activity, indicating that building a new structure by adding atoms and changing bonds was not a problem, especially because the undo function was available. This value in- creased slightly in the second activity and significantly in the third activity. Nearly two out of five students estimated that they were too far off course compared to the target model or did not get close enough, so they started over. They were not discouraged and there was no time limit on the activity. In this activity, the importance of good spatial ability was probably most pronounced. Starting over was the chosen strategy. 3. 3. 2. Correlations Interestingly, for all three activities, there was a significant negative correlation between using the video (once) and hovering buttons, suggesting that students who did not watch the video relied on hovering buttons in the toolbar (Table 14). No significant correlation with the four types of help was found for free rotation or the use of the undo button in any of the activities. This could mean that these two functionalities were used by all. In the first activ- ity, the negative correlation with button hovering was also observed for multiple video views. There, the use of redo was positively associated with the help menu and negative- ly associated with watching the video once. In the second activity, use of the help menu was negatively correlated with viewing the video once, indicating that students for whom viewing the video once was sufficient did not use it. With the fewest geometry changes in this activity, students who used free rotation were less likely to use the undo but- ton. In this way, the rotation helped. It is surprising that this was not the case in the third activity, where students could benefit from free rotation even more. There, use of the help menu correlated significantly with other types of help, aside from watching the video once. Students who Figure 8: Distribution of actions with help tools by activity 180 Acta Chim. Slov. 2022, 69, 167–186 Dolničar et al.: The Students’ Perceptions Using 3DChemMol ... needed help used all available types of help. Students who restarted the activity were also more likely to consult the help menu and watch the video multiple times. 3. 3. 3. Correlations with Completion Level of the Activities For the first two activities, activity completion cor- related negatively, and perceived difficulty correlated positively with help menu use (Table 15). Students who did not need to consult the help menu were more likely to complete the activity. Those who did consult the help menu perceived the activity to be more difficult. On the third activity, students who did not have to watch the in- structional video multiple times were more likely to com- plete the activity. Multiple video viewings also correlated positively with perceived activity difficulty. It seems that consulting the static help menu did not help solve the easi- er activities and that the tutorial videos were not sufficient to solve the more difficult activities. One of the possible remedies would be to create help tutorials/videos for in- dividual actions that students found particularly difficult, covering multiple examples. The use of undo correlated with time spent on the first two activities and redo did on the last two activities. Both also correlated positively with perceived difficulty – students who used them found the activities more difficult. With the third activity, the amount of time spent restarting was significantly higher, and these students were less likely to complete the activity they also perceived as more difficult. Starting over did not help enough. 3. 3. 4. Correlations with Difficulties by Activity The difficulty level referenced is the average step dif- ficulty for each activity (step difficulty mean). In the first activity, one video view seemed sufficient for students who reported fewer difficulties overall (Table 16). In the second Table 14: Spearman correlations between actions with help tools Video Video Button Action Help Free Undo Redo Restart once multi hover status menu rotat. button button activity Activity 1: Building a simple model of the molecule Video once 1.000 Video multi –0.416b 1.000 Button hover –0.270a –0.275a 1.000 Action status 0.071 –0.090 0.185 1.000 Help menu –0.152 –0.102 –0.017 –0.090 1.000 Free rotat. –0.168 0.152 0.218 0.135 –0.177 1.000 Undo button 0.006 0.067 0.067 0.184 0.067 –0.205 1.000 Redo button –0.369b –0.090 0.042 0.190 0.398b 0.135 0.029 1.000 Restart activity –0.020 0.118 0.242 0.154 0.118 0.152 0.207 –0.090 1.000 Activity 2: Transformation of one into two models of the molecules Video once 1.000 Video multi –0.058 1.000 Button hover –0.554b –0.154 1.000 Action status –0.084 –0.057 0.072 1.000 Help menu –0.297a –0.064 0.015 0.152 1.000 Free rotat. –0.149 0.130 0.258 0.188 –0.069 1.000 Undo button –0.002 –0.106 0.113 0.009 0.193 –0.301a 1.000 Redo button –0.057 –0.077 0.271a 0.100 0.255 0.135 0.234 1.000 Restart activity –0.057 –0.077 0.156 –0.111 0.065 0.014 0.107 0.012 1.000 Activity 3: Transformation from noncyclical to cyclical form of the molecule Video once 1.000 Video multi –0.610b 1.000 Button hover –0.337a –0.139 1.000 Action status –0.099 0.038 0.259 1.000 Help menu –0.163 0.322a 0.326a 0.298a 1.000 Free rotat. –0.179 0.161 0.132 0.015 –0.006 1.000 Undo button 0.078 0.177 –0.108 0.108 0.100 0.142 1.000 Redo button –0.174 0.098 0.333a 0.269 0.047 0.087 0.139 1.000 Restart activity –0.107 0.315a 0.198 0.002 0.282a –0.164 0.020 0.146 1.000 ap < 0.05, bp < 0.01 181Acta Chim. Slov. 2022, 69, 167–186 Dolničar et al.: The Students’ Perceptions Using 3DChemMol ... activity, more difficulties likely resulted in multiple video views. On the third activity, no correlation was found be- tween difficulty and video views. Difficulty level correlated positively with help menu use on the first two activities. This means that students who had difficulties were more likely to consult the help menu. In both activities where the action mode was changed (activities two and three), the difficulty level correlated with the use of the action sta- tus help. Students who had difficulties consulted this help. Hovering over buttons, free rotation, and restarting the ac- tivity did not significantly correlate with difficulty levels. For all activities, using the undo button, as well as the redo button, were positively correlated with problems. 3. 3. 5. Correlations with Difficulties by Step Activity 1. Consultation of the help menu correlated with step difficulty levels in almost all individual steps (Ta- ble 17). In general, students who had difficulties consulted the help menu. The exception was changing the bond type, where difficulties were inversely correlated with watching the video multiple times. Students who watched the vid- eo multiple times had fewer difficulties with this step. The shortcut for this step was not available in the toolbar but was visible in the action mode description. Those who had difficulties changing the bond type also used the undo and redo buttons. Difficulties with centering the model corre- lated with the use of button hover, indicating difficulty in visually identifying the correct button. Students who used free rotation were less likely to have difficulties with geom- etry optimization. Activity 2. The use of the help menu, as well as the use of the redo button, correlated with difficulty levels in this activity. The exception was manually adding hydrogen, the step that was perceived as the most difficult and, like the shortcut for changing the bond, was not explicitly shown in the toolbar. Undo was used most frequently with the manual hydrogen addition. In this activity, multiple video views correlated with difficulties changing bond type and deleting atoms. Students used multiple videos when they encountered these difficulties. Activity 3. In contrast to the previous two activities, correlations between difficulty and help menu use were absent or low (not significant). For the two most diffi- cult steps, bond addition and branch rotation, there was a low correlation with the use of action status and undo. Two problems were possibly associated with these steps: Table 15: Spearman correlations between actions with help tools and completion level (Comp.), time spent (Time) and perceived difficulty (Perc. diff.) Video Video Button Action Help Free Undo Redo Restart once multi hover status menu rotat. button button activity Activity 1: Building a simple model of the molecule Comp. 0.286a –0.114 0.022 0.090 –0.351b 0.024 –0.070 –0.149 –0.330a Time –0.114 0.079 0.062 –0.207 0.252 0.034 0.286a 0.078 0.033 Perc. diff. –0.108 0.028 0.088 –0.161 0.293a –0.008 0.205 0.033 0.101 Activity 2: Transformation of one into two models of the molecules Comp. 0.0584 0.039 –0.050 0.057 –0.275a 0.085 –0.119 –0.508b –0.215 Time –0.164 0.235 0.059 0.010 0.420b –0.068 0.371b 0.336a 0.185 Perc. diff. –0.106 0.266 0.122 0.129 0.351b 0.090 0.353b 0.502b 0.219 Activity 3: Transformation from noncyclical to cyclical form of the molecule Comp. 0.138 –0.419b 0.050 –0.084 –0.187 –0.187 –0.224 –0.228 –0.315a Time –0.185 0.165 0.308a –0.006 0.371b 0.130 0.147 0.435b 0.349a Perc. diff. –0.195 0.391b 0.106 0.086 0.177 0.181 0.330a 0.379b 0.297a ap < 0.05, bp < 0.01 Table 16: Spearman correlations between actions with help tools and step difficulty mean Activity no. Video Video Button Action Help Free Undo Redo Restart once multi hover status menu rotat. button button activity 1 –0.278a –0.012 0.238 0.117 0.313a –0.068 0.336a 0.243 0.187 2 –0.186 0.287a 0.110 0.305a 0.357b –0.191 0.331a 0.325a 0.151 3 –0.0724 0.169 0.026 0.299a 0.169 0.021 0.296a 0.313a 0.176 ap < 0.05, bp < 0.01 182 Acta Chim. Slov. 2022, 69, 167–186 Dolničar et al.: The Students’ Perceptions Using 3DChemMol ... recognizing the correct action and performing the action correctly. Action status could help with the first part. The tutorial video could help with the second part. Only for branch rotation difficulties was there a low correlation with multiple video views. 4. Conclusions and Implications for Teaching The experience of undergraduate students in con- struction and editing of molecular models of small organic compounds aimed to equip them with the knowledge and ability to create their own presentations and to proceed with further exploration and analysis of model properties in chemistry courses beyond introductory chemistry. The success of the course also depends on the design of the course and the teacher, which would be worth of further study. Manipulation of 3D molecular models has been as- sociated with the development of representational skills, particularly when used to support learning.12,13 Students of all ages encounter problems and misunderstandings when asked to explain chemical phenomena at the submi- croscopic level.73 Molecular modeling has long been used to support experimental work, and to teach fundamental concepts.39 Previous studies have also shown that software usability, expressed as perceived meaningfulness and ease of use, has an impact on learning.74 Spatial ability is an- other factor involved in learning science.75 Its active pro- motion in college-level chemistry and biochemistry has increased, but not to the same extent as other cognitive skills.76 The 3DChemMol molecular editor for building/edit- ing 3D molecular models was used in the study. Features implemented in the user interface allowed for ease of use: a toolbar; separation of the editing function from other functions; the ability to undo and redo changes for mul- tiple steps; various types of help, including video tutorials, button hovering, action status display, and help menu. The 3DChemMol molecular editor incorporating an editing toolbar was tested in a group of 54 university students using three model building/editing activities of varying difficulty: 1) building a simple model, 2) splitting a model into two, 3) creating a cyclic from a non-cyclic structure. Table 17: Spearman correlations between use of help tools and step difficulty level Step Video Video Button Action Help Free Undo Redo Restart once multi hover status menu rotat. button button activity Activity 1: Building a simple model of the molecule 1 New model –0.005 –0.178 0.125 0.015 0.416b 0.014 0.090 0.294a 0.290a 2 Adding C atoms –0.219 –0.047 0.040 –0.043 0.440b –0.149 0.215 0.308a 0.122 3 Adding heteroat. –0.298a 0.176 0.119 –0.065 0.303a 0.088 0.269 0.145 0.090 4 Chg. bond type –0.196 –0.272a 0.237 0.158 0.247 –0.014 0.314a 0.334a 0.050 5 Model centering –0.385b –0.023 0.300a –0.173 0.351b –0.017 0.129 0.227 0.135 6 Geometry optim. –0.192 –0.022 0.117 –0.158 0.520b –0.292a 0.280a 0.158 0.129 7 Creating image –0.048 –0.075 0.170 0.168 0.080 –0.310a 0.139 –0.018 0.115 Mean –0.278a –0.012 0.238 0.117 0.313a –0.068 0.336a 0.243 0.187 Activity 2: Converting one model into two models of the molecules 1 Deleting bonds –0.132 0.120 0.181 0.140 0.272a –0.203 0.158 0.433b 0.119 2 Chg. bond type –0.151 0.390b 0.124 0.204 0.442b –0.039 0.214 0.453b 0.127 3 Deleting atoms –0.078 0.408b 0.120 0.121 0.381b –0.203 0.192 0.329a 0.162 4 Adding hydrogen –0.238 0.137 0.096 0.218 0.2548 –0.097 0.450b 0.118 0.046 5 Geometry optim. –0.356b 0.218 0.098 0.098 0.427b 0.043 0.121 0.299a 0.129 6 Creating image –0.2206 0.185 –0.104 0.243 0.328a –0.122 –0.144 0.214 –0.187 Mean –0.186 0.287a 0.110 0.305a 0.357b –0.191 0.331a 0.325a 0.151 Activity 3: Converting from a non-cyclic to a cyclic form of the molecule 1 Adding a bond 0.032 0.015 0.079 0.203 0.241 –0.096 0.259 0.106 0.059 2 Chg. bond type –0.034 0.069 0.115 0.261 0.162 0.046 0.031 0.408b 0.130 4 Rotating a branch 0.015 0.198 –0.029 0.267 –0.022 0.152 0.228 0.293a 0.250 3,5 Geom. optim. –0.394b 0.061 0.180 0.198 –0.034 0.070 0.123 0.315a –0.185 6 Creating image 0.012 0.071 –0.350a 0.066 –0.196 –0.084 0.123 –0.050 –0.058 Mean –0.0724 0.169 0.026 0.299a 0.169 0.021 0.296a 0.313a 0.176 ap < 0.05, bp < 0.01 183Acta Chim. Slov. 2022, 69, 167–186 Dolničar et al.: The Students’ Perceptions Using 3DChemMol ... In relation with first research question (RQ1), it was found that students were successful overall in using the tool and graphical interface and in completing the ac- tivities. They were excellent on the first two activities and good on the third activity. As expected, the more time they spent on an activity, the more difficult it appeared to them. When they were unable to complete the activity, they per- ceived it to be more difficult. No relationship was found between time spent and success rate. As expected, the average step difficulty of the activity correlated inversely with activity completion and directly with perceived activity difficulty. The more difficulties stu- dents encountered, the more difficult the activity seemed to them; more difficulties also meant more time spent on the activity. When it comes to the second research question (RQ2), it was found that actions for direct model manip- ulation (atoms, bonds) caused more difficulties than using the toolbar buttons. There were more difficulties interact- ing with the model by dragging than by clicking. Steps that involved changing the model configuration or required changing the working mode of the interface were more problematic. It was also found that actions were perceived as easier if they were preceded by a clear mode change. This means that a lot of emphasis needs to be placed on displaying the state of the system so that the user is imme- diately aware of the actions available. The most difficult individual actions reported were 1) rotating a branch around a bond, 2) adding a bond be- tween two existing atoms, and 3) manually adding a hy- drogen atom, but also 4) changing a bond type, 5) creating an image, and 6) adding heteroatoms. Issue #5 was techni- cal in nature. Actions 2 and 3 involved dragging the mouse on or between model atoms. Issues 2–4 had a common denominator: the actions were not implicitly given in the toolbar but were available as part of the default add/change action mode, so students could not discover them without either watching the video tutorial or reading the available actions displayed at the bottom of the screen. Correct ad- dition of bonds and heteroatoms probably requires good spatial orientation, which could be especially true for branch rotation. Action 1 required repeated clicking on a bond until a satisfactory configuration was achieved. The latter was done in 60-degree increments. Difficulties related to the user interface will be ad- dressed in future improvements of the tool, such as high- lighting the action state or even separating actions. Diffi- culties related to spatial abilities could be mitigated by sim- ple video tutorials and exercises focusing on a single issue. Related to the third research question (RQ3), the study indicated that among the four types of help provid- ed, and regardless of reported difficulties, students most frequently watched video tutorials once or used hovering over buttons to indicate button meanings. Use of other forms of help increased only on the third activity, which was perceived as most difficult. Use of the multiple undo feature was high, indicating that it was absolutely neces- sary, and increased with activity difficulty. Similarly, free rotation compensated for the use of the undo function on the second activity. The most difficult and complex activity was found to have a relatively high rate of restarting the activity and re-watching the learning video. When difficulties occurred, students most often used the help menu and the undo/redo actions. Use of the undo function increased for the most difficult steps. For activ- ities/steps that required a mode change, more students consulted the action state that contained the correct an- swer. Individual activities were associated with multiple video views, with video views generally increasing on the most difficult activity. Mouse hovering over the toolbar was used more often when students could not visually identify the correct button. Sometimes the wrong type of help was consulted, such as button hovering (looking for an appropriate action) when no toolbar interaction was re- quired. Reading the action status would have helped there. In other cases, consulting the action status did not con- tain the answer and the tutorial video should be watched. Negative correlations between difficulties and single video views may indicate that the video was a sufficient aid in activity completion for many students. Despite using all the help available (multiple tutori- al video views and restarting the activity), some students were not able to complete the most difficult activity. This could be related to the difficulty of the activity and the need for good spatial orientation and/or mean that the help menus and system status were not fully utilized. Some of the lessons learned in this work, particular- ly the shortcomings of the user interface for editing, have already been implemented and further improvements are planned. Video tutorials became an important part of the help menu. Bond change methods will be unified so that they always include hydrogen adjustment. The toolbar will be upgraded with additional buttons, e.g., for actions that were part of the working modes but were not explicitly present. The action status display will be improved, and video tutorials for individual actions that proved most dif- ficult will be added and immediately available. Alternative help display could be considered, e.g., when you hover over the model parts. The implications for teaching of this study are mul- tifaceted. Using the new tool, students successfully creat- ed 3D models with the help of video tutorials and various types of help. In general, the availability of tools is not yet sufficient for students to use them for learning. Their use must to be encouraged through pedagogical approaches. We suggest that the tool is suitable for direct instruction or self-study. Students can easily use this tool to visualize the structure of chemical compounds during their studies and create images of 3D models to include in their own prod- ucts, such as seminar works, reports, and theses. 3DChemMol could also be used to improve students’ development of chemistry knowledge and representation- 184 Acta Chim. Slov. 2022, 69, 167–186 Dolničar et al.: The Students’ Perceptions Using 3DChemMol ... al skills. Some students may be afraid of special chemistry visualization software because they think it requires spe- cial skills. Because of its simplicity, even students who were not previously familiar with molecular modeling tools and may not have had experience drawing 3D representations or molecules can use it after studying short tutorials. Using 3DChemMol allows students to construct molecular mod- els to visualize the structure of compounds and under- stand their properties, rather than memorizing facts and writing about them. The accessibility of the 3DChemMol tool makes it easy to incorporate into various educational settings. The models created form the basis for further investigation and study of chemistry concepts through display of chem- ical properties. Teachers can use the tool directly in the classroom during lectures or prepare study materials for students in electronic or printed form. For example, vis- ualizations created in 3DchemMol can be part of lectures on various topics. Moreover, it can be used in students’ in- dividual work when they can check their understanding on new examples. Different levels of task difficulty can be accommodated in the tool by the teacher. We are aware that our observational study has some limitations. One of them is the self-reporting nature of the questionnaires. Further insight into students’ behavior and efficiency in building molecular models could be gained by using additional recording and analysis methods, such as eye-tracking, video recording during activity perfor- mance, and structured interviews afterwards. Anoth- er limitation was that the study was focused only on the editing feature of the tool. Future research could include experimental studies such as comparing the usability and effectiveness of other features of the tool (e.g. molecular property display and exploration), comparing it with other 2D and 3D model editing tools, and with building physical models, investigating correlations with other internal or external factors such as students’ spatial skills, representa- tional competence, chemistry knowledge and teaching methods. However, this is already beyond the scope of this study. In further development of 3DChemMol more inter- active online tutorials and exercises tailored to specific chemistry courses could be prepared. Acknowledgements This study was co-financed by University of Ljublja- na (grant no. 704-8/2016-229). The authors would like to thank all of the students who participated in the survey. 5. References 1. K. L. Vavra, V. Janjic-Watrich, K. Loerke, L. M. Phillips, S. P. Norris, J. Macnab, Alta. Sci. Educ. J. 2011, 41, 22–30. 2. I. Vekiri, Educ. Psychol. Rev. 2002, 14, 261–312. DOI:10.1023/A:1016064429161 3. S. R. Hinze, D. N. Rapp, V. M. Williamson, M. J. Shultz, G. De- slongchamps, K. C. Williamson, Learn. Instr. 2013, 26, 12–21. DOI:10.1016/j.learninstruc.2012.12.002 4. J. K. Gilbert, D. F. Treagust, in: J. K. Gilbert and D. Trea- gust (Eds.): Multiple Representations in Chemical Education, Springer Netherlands, 2009, pp. 1–8. DOI:10.1007/978-1-4020-8872-8_1 5. A. H. Johnstone, Sch. Sci. Rev. 1982, 64, 377–379. DOI:10.1017/S0035336100178700 6. A. H. Johnstone, J. Comput. Assist. Learn. 1991, 7, 75–83. DOI:10.1111/j.1365-2729.1991.tb00230.x 7. V. Ferk Savec, I. Sajovic, K. S. W. Grm, in: Multiple representa- tions in chemical education, Springer Netherlands, 2009, pp. 309–331. DOI:10.1007/978-1-4020-8872-8_14 8. R. Kozma, J. Russell, in: J. K. Gilbert (Ed.): Visualization in Science Education, Springer Netherlands, 2005, pp. 121–145. DOI:10.1007/1-4020-3613-2_8 9. M. Stieff, A. Origenes, D. DeSutter, M. Lira, L. Banevicius, D. Tabang, G. Cabel, J. Educ. Psychol. 2018, 110, 1160–1174. DOI:10.1037/edu0000258 10. N. A. Kiernan, A. Manches, M. K. Seery, Chem. Educ. Res. Pract. 2021, 22, 626–639. DOI:10.1039/D0RP00354A 11. M. Stieff, S. Scopelitis, M. E. Lira, D. Desutter, Sci. Educ. 2016, 100, 344–363. DOI:10.1002/sce.21203 12. A. T. Stull, M. Gainer, S. Padalkar, M. Hegarty, J. Chem. Educ. 2016, 93, 994–1001. DOI:10.1021/acs.jchemed.6b00194 13. A. T. Stull, M. Hegarty, J. Educ. Psychol. US, 2016, 108, 509– 527. DOI:10.1037/edu0000077 14. Y. J. Dori, M. Barak, J. Educ. Technol. Soc. 2001, 4, 61–74. 15. S. R. Hinze, V. M. Williamson, M. J. Shultz, K. C. William- son, G. Deslongchamps, D. N. Rapp, Cogn. Process. 2013, 14, 129–142. DOI:10.1007/s10339-013-0539-3 16. E. Francoeur, Soc. Stud. Sci. 1997, 27, 7–40. DOI:10.1177/030631297027001002 17. D. Gabel, D. Briner, D. Haines, Sci. Teach. 1992, 59, 58–63. DOI:10.1038/s41578-019-0146-8 18. A. M. Ingham, J. K. Gilbert, Int. J. Sci. Educ. 1991, 13, 193– 202. DOI:10.1080/0950069910130206 19. N. Barnea, in: J. K. Gilbert and C. J. Boulter (Eds.): Devel- oping Models in Science Education, Springer Netherlands, Dordrecht, 2000, pp. 307–323. DOI:10.1007/978-94-010-0876-1_16 20. H. Hugel, Chem. Aust. 2005, 72, 31. 21. H.-K. Wu, J. S. Krajcik, E. Soloway, J. Res. Sci. Teach. 2001, 38, 821–842. DOI:10.1002/tea.1033 22. W. L. DeLano, CCP4 Newsl. Protein Crystallogr. 2002, 40, 82–92. 23. Free Molecular Modeling Software, 1999, https://www.edin- formatics.com/mathmol/free-molecular-modeling-software. html, (accessed: December 20, 2021) 24. A. Moll, A. Hildebrandt, H.-P. Lenhof, O. Kohlbacher, Bioin- formatics 2006, 22, 365–366. DOI:10.1093/bioinformatics/bti818 25. E. F. Pettersen, T. D. Goddard, C. C. Huang, G. S. Couch, D. M. Greenblatt, E. C. Meng, T. E. Ferrin, J. Comput. Chem. 185Acta Chim. Slov. 2022, 69, 167–186 Dolničar et al.: The Students’ Perceptions Using 3DChemMol ... 2004, 25, 1605–1612. DOI:10.1002/jcc.20084 26. R. A. Sayle, E. J. Milner-White, Trends Biochem. Sci. 1995, 20, 374–376. DOI:10.1016/S0968-0004(00)89080-5 27. M. D. Casselman, J. F. Eichler, K. Atit, Sci. Educ. 2021, 1–30. DOI:10.1002/sce.21675 28. Erlina, C. Cane, D. P. Williams, J. Chem. Educ. 2018, 95, 991– 995. DOI:10.1021/acs.jchemed.7b00687 29. B. Mohamed-Salah, D. Alain, Chem. Educ. Res. Pract. 2016, 17, 862–877. DOI:10.1039/C5RP00180C 30. A. T. Stull, M. J. Gainer, M. Hegarty, Learn. Instr. 2018, 55, 80–92. DOI:10.1016/j.learninstruc.2017.09.008 31. A. Taly, F. Nitti, M. Baaden, S. Pasquali, Interface Focus 2019, 9, 1–11. DOI:10.1098/rsfs.2018.0065 32. C. A. Terrell, M. A. Franzen, T. Herman, S. Malapati, D. L. Newman, L. K. Wright, in: T. J. Bussey, K. Linenberger Cortes and R. C. Austin (Eds.): Biochemistry education : from theo- ry to practice, American Chemical Society, Washington, DC, 2019, pp. 43–62. (ACS Symposium Series; vol. 1337). DOI:10.1021/bk-2019-1337.ch003 33. T. Thayban, H. Habiddin, Y. Utomo, M. Muarifin, Acta Chim. Slov. 2021, 68, 736–743. DOI:10.17344/acsi.2021.6836 34. V. Ferk Savec, M. Vrtacnik, J. K. Gilbert, in: J. K. Gilbert (Ed.): Visualization in Science Education, Springer Netherlands, 2005, pp. 269–297. DOI:10.1007/1-4020-3613-2_14 35. M. Barak, in: J. P. Suits and M. J. Sanger (Eds.): Pedagogic Roles of Animations and Simulations in Chemistry Courses, Amer Chemical Soc, Washington, 2013, pp. 273–291. DOI:10.1021/bk-2013-1142.ch011 36. M. T. Springer, J. Chem. Educ. 2014, 91, 1162–1168. DOI:10.1021/ed400054a 37. R. Kelly, S. Akaygun, Chem. Educ. Res. Pract. 2019, 20, 657– 658. DOI:10.1039/C9RP90009H 38. T. J. José, V. M. Williamson, J. Chem. Educ. 2005, 82, 937–943. DOI:10.1021/ed082p937 39. W. J. Hehre, A. J. Shusterman, J. E. Nelson, The molecular modeling workbook for organic chemistry, Wavefunction, Irvine, CA, 2004. 40. A. D. Clauss, S. F. Nelsen, J. Chem. Educ. 2009, 86, 955–958. DOI:10.1021/ed086p955 41. K. J. Linenberger, R. S. Cole, S. Sarkar, J. Chem. Educ. 2011, 88, 962–965. DOI:10.1021/ed100727r 42. J. Raiyn, A. Rayan, World J. Chem. Educ. 2015, 3, 1–4. DOI:10.12691/wjce-3-1-1 43. E. Rothe, W. Zygmunt, Chem. Eng. Educ. 2016, 50, 125–129. DOI:10.1007/s10749-016-0672-0 44. Y. J. Dori, M. Barak, N. Adir, J. Chem. Educ. 2003, 80, 1084– 1092. DOI:10.1021/ed080p1084 45. J. B. Ealy, J. Sci. Educ. Technol. 2004, 13, 461–471. DOI:10.1007/s10956-004-1467-x 46. R. Kozma, Learn. Instr. 2003, 13, 205–226. DOI:10.1016/S0959-4752(02)00021-X 47. J. Lundell, Educ. Quím. 2017, 33–36. DOI:10.2436/20.2003.02.157 48. K. Kolar, R. Dolezal, N. Karaskova, N. Maltsevskaya, S. Kriz- kova, Chem.-Didact.-Ecol.-Metrol. Warsaw, 2019, 24, 45–51. DOI:10.2478/cdem-2019-0003 49. L. L. Winfield, K. McCormack, T. Shaw, J. Chem. Educ. Wash- ington, 2019, 96, 89–92. DOI:10.1021/acs.jchemed.8b00145 50. S. S. Johnson, L. Winfield, S. H. Sung, J. Chem. Educ. 2021, 98, 982–985. DOI:10.1021/acs.jchemed.0c01134 51. L. Soulère, J. Chem. Educ. 2021, 98, 2709–2713. DOI:10.1021/acs.jchemed.1c00017 52. A. Grushow, M. S. Reeves, in: Using Computational Methods To Teach Chemical Principles, American Chemical Society, 2019, pp. 1–10. (ACS Symposium Series; vol. 1312). DOI:10.1021/bk-2019-1312.ch00110.1021/bk-2019-1312.ch001 53. J. Rodríguez-Becerra, L. Cáceres-Jensen, T. Díaz, S. Druker, V. B. Padilla, J. Pernaa, M. Aksela, Chem. Educ. Res. Pract. 2020, 21, 638–654. DOI:10.1039/C9RP00273A 54. M. Aksela, J. Lundell, Chem. Educ. Res. Pract. 2008, 9, 301– 308. DOI:10.1039/B818464J 55. M. Aksela, J. Pernaa, P. Blomgren, I. Rautiainen, Collabora- tive science education at the University of Helsinki: Chemis- tryLab Gadolin: a science lab as an inspiring environment for learning, development and research since 2008, Unigrafia Oy, Helsinki, 2018. 56. J. Pernaa, M. Aksela, S. P. Ghulam, Introduction to Molecular Modeling in Chemistry Education, e-Oppi Ltd. & Edumendo Publishing, Jokioinen, 2017. 57. J. Pernaa, LUMAT Int. J. Math Sci. Technol. Educ. 2015, 3, 960–975. DOI:10.31129/lumat.v3i7.977 58. D. Dolničar, B. Boh Podgornik, in: I. Devetak (Ed.): Eurova- riety 2021: 9th European Variety in University Chemistry Education Conference: 7-9 July 2021, Ljubljana, Slovenia, online conference: book of abstracts, University of Ljubljana, Faculty of Education, 2021, p. 74. http://pefprints.pef.uni-lj. si/id/eprint/6794 59. L. A. Abriata, Inform.-Basel 2017, 4, 1–18. DOI:10.3390/informatics4030028 60. S. Pirhadi, J. Sunseri, D. R. Koes, J. Mol. Graph. Model. 2016, 69, 127–143. DOI:10.1016/j.jmgm.2016.07.008 61. M. C. Burger, J. Cheminformatics 2015, 7, 1–7. DOI:10.1186/s13321-015-0085-3 62. B. Rayan, A. Rayan, World J. Chem. Educ. 2017, 5, 136–141. DOI:10.12691/wjce-5-4-4 63. R. M. Hanson, J. Appl. Crystallogr. 2010, 43, 1250–1260. DOI:10.1107/S0021889810030256 64. N. Rego, D. Koes, Bioinformatics 2015, 31, 1322–1324. DOI:10.1093/bioinformatics/btu829 65. H. Bergwerf, ACS CHED CCCE Newsl. 2015, https://conf- chem.ccce.divched.org/2015FallCCCENLP9, (accessed: De- cember 20, 2021) 66. R. M. Hanson, J. Prilusky, Z. Renjian, T. Nakane, J. L. Suss- man, Isr. J. Chem. 2013, 53, 207–216. DOI:10.1002/ijch.201300024 67. O. Rothenberger, Virtual Molecular Model Kit, https://che- magic.org/molecules/amini.html, (accessed: December 20, 2021) 68. J. H. Jensen, J. C. Kromann, J. Chem. Educ. 2013, 90, 1093– 1095. DOI:10.1021/ed400164n 69. B. Nevo, J. Educ. Meas. 1985, 22, 287–293. DOI:10.1111/j.1745-3984.1985.tb01065.x 186 Acta Chim. Slov. 2022, 69, 167–186 Dolničar et al.: The Students’ Perceptions Using 3DChemMol ... 70. J. Kim, Inf. Res. 2008, 13, 13–14. DOI:10.1111/j.1469-8749.1971.tb03071.x 71. M. Harle, M. Towns, J. Chem. Educ. 2011, 88, 351–360. DOI:10.1021/ed900003n 72. H. Van der Meij, J. Van Der Meij, Comput. Educ. 2014, 78, 150–159. DOI:10.1016/j.compedu.2014.06.003 73. M. Slapničar, V. Tompa, S. A. Glažar, I. Devetak, J. Pavlin, Acta Chim. Slov. 2020, 67, 904–915. DOI:10.17344/acsi.2020.5908 74. Z. Merchant, E. T. Goetz, W. Keeney-Kennicutt, O. Kwok, L. Cifuentes, T. J. Davis, Comput. Educ. 2012, 59, 551–568. DOI:10.1016/j.compedu.2012.02.004 75. M. Tanweer, Eur. J. Phys. Educ. Sport Sci. 2019, 4, 145–151. DOI:10.5281/zenodo.3407001 76. M. Oliver-Hoyo, M. A. Babilonia-Rosa, J. Chem. Educ. 2017, 94, 996–1006. DOI:10.1021/acs.jchemed.7b00094 Povzetek V članku je predstavljena študija, v kateri je 54 univerzitetnih študentov preizkusilo in ovrednotilo 3DChemMol - no- vo razviti, brezplačni spletni urejevalnik modelov molekul z orodno vrstico. Namen orodja je povečanje reprezentaci- jske kompetence v povezavi s submikroskopskimi predstavitvami. Študenti so programsko opremo uporabili prvič. Z orodjem za gradnjo/urejanje modelov so izdelali modele molekul v naslednjih treh aktivnostih z različnimi stopnjami težavnosti: 1) gradnja preprostega modela (butanojska kislina), 2) pretvorba enega modela (heksan) v dva modela, 3) pretvorba iz neciklične v ciklično obliko (glukoza). Študenti so za izvedbo vsake od aktivnosti potrebovali od dveh do 15 minut. V orodni vrstici urejevalnika 3DChemMol je bilo na voljo več vrst pomoči, ki so študentom olajšale izvajanje aktivnosti, vključno z video vodnikom, prikazom pomoči ob preletu gumbov orodne vrstice z miško, prikazom statusa/ načina dela in menijem pomoči. Na voljo so bile tudi možnosti razveljavitve in ponovne uveljavitve posameznih korakov ter ponovnega začetka celotne aktivnosti. Stopnjo dokončanja aktivnosti, težave in uporabo pomoči smo preučevali s pomočjo vprašalnikov za samoocenjevanje študentov. Urejevalnik molekul 3DChemMol se je izkazal kot koristna pod- pora študentom pri preprostih kemijskih aktivnostih. Študenti so bili pri gradnji modelov uspešni, čeprav so naleteli na nekatere specifične težave, zlasti pri korakih, ki so vključevali prostorske operacije, kot je vrtenje izbranega dela mod- ela molekule okoli vezi. Po mnenju študentov so bila video navodila najprimernejša in najpogosteje uporabljena vrsta pomoči, funkcija razveljavitve pa je bila pri delu bistvenega pomena. Rezultati kažejo, da lahko urejevalnik modelov molekul 3DChemMol učinkovito uporabljamo pri osnovnih predmetih kemije na terciarni ravni izobraževanja, bodisi za poučevanje, samostojno učenje študentov ali druge oblike podpore v pedagoškem procesu. Rezultati in ugotovitve študije bodo uporabljeni tudi za nadaljnjo optimizacijo uporabniškega vmesnika v prihodnjih različicah ovrednotenega orodja. Except when otherwise noted, articles in this journal are published under the terms and conditions of the  Creative Commons Attribution 4.0 International License